Jurnal Voluntary Disclosure

8/19/2019 Jurnal Voluntary Disclosure

1/28

1

BOARD COMPOSITION, OWNERSHIP STRUCTURE AND VOLUNTARY

DISCLOSURE IN ANNUAL REPORTS: EVIDENCE FROM TUNISIA

Hamadi Matoussi

Professor of Finance-ISCAE, TunisiaLaboratoire Interdisciplinaire de Gestion Université-Entreprise (LIGUE)

E-mail: [email protected]

Raida Chakroun

Doctorate student-ISCAE, Tunisia

Laboratoire Interdisciplinaire de Gestion Université-Entreprise (LIGUE)

Telephone: (+216) 98 953 605

E-mail: [email protected]

AbstractIn the setting of our research, we study the interactions between the composition of the board

of directors, ownership concentration and voluntary disclosure in annual reports. This

research uses panel data of Tunisian listed firms that don’t belong to the financial sector and

this for the years 2003-2005. Our results show that the extent of voluntary disclosure tend to

increase across the time. The results reveal also that the independence and the structure of the

leadership of the board of directors and the familial control of the firm doesn’t lead to more

voluntary disclosure. The results highlight that the size of the board of directors has no effect

on the extent of voluntary disclosure. However, we have observed that managerial ownership

and a good quality of corporate governance are represented as factors that incite firms to

disclose voluntarily.

Keywords: Voluntary disclosure. Annual reports. Ownership structure. Board of directors’

composition. Corporate governance.

1. Introduction

The quality of voluntary disclosure contained in the annual reports, the principle

source of information, is nowadays in the heart of financial modern problems. Firms are

confronted to serious crisis of trust and they can’t think about the efficiency of their financial

communication. Thus, transparency and a better disclosure make the stakeholders of the firm

better informed. This will lead to a better capital allocation in the securities market.Financial reporting is not focusing on the accounts anymore and on the numerical and

quantitative information. It includes qualitative information accompanied with comments in

addition to extra financial information, and then develops global economic information. It

happens that the financial statements are insufficient to give a true image for the firm. To

complete the accounting “portrait” of the firm, managers will be conducted to voluntarily

disclose information. Voluntary disclosure can be defined as a facultative publication that is

not part of the public rights for information.

Besides, the importance of information accompanying the financial statements in the

annual reporting is recognized in the preface of the IFRS standards. The IASB make

references to the management report in the “IAS 1” relative to the presentation of the financial

statements that stipulates that: “the firms are encouraged to present, apart from the financialstatements, a management report that describes and explains the main characteristics of the

financial performance and of the financial situation of the firm as well as the main


2/28

2

uncertainties to which the firm is facing.” In the same sense, the IASB have published in

October 2005 a project of accounting standard about the management report called

“management commentary” considered as been an integral part of the financial reporting of

the firm. This standard project has as objectives to guide the firm in the preparation of a

narrative report that includes non mandatory information. In addition, several directives areappeared in the last years, such as the recommendations about « Management Discussion and

Analysis » (MD&A) recommended in Canada and the United States and recommended by the

commission of the IOSCO in its report about the general principles of the reporting of MD&A

in February 2003

Besides talking about accounting choices, we can talk about disclosure choices. Thus,

the accounting policy is not only in line with logic of optimization of the accounting choices,

but also of the economic communication. That means that managers have a large margin of

discretion allowing him to publish or no an information. Thus, voluntary disclosure is not a

fortuity act or related to the ignorance of the legal frame of disclosure, but an accounting

policy instrument that aims to reach certain objectives.

Based on the idea that a good governance guarantee an economic communication witha better quality, the firm should put in place an efficient internal system of governance that

assure the protection of the shareholders against the managers’ opportunism. In this context

the studies of Cheung, Connelly, Limpaphayom and Zhou (2006), of Ho and Wong (2001), of

Eng and Mak (2003), of Chen and Jaggi (2000) and of Cheng and Courtenay (2006) have

supported that putting in place a guide of good practices of corporate governance helps

enhance voluntary disclosure in annual reports.

In this sense, the role of the board of directors consists in controlling and disciplining

the corporate management and then to assure that the managers behaves in the sense of

increasing the shareholders’ interests. This board prepares the annual reports under his

responsibility, thus his composition affect the extent of voluntary disclosure in annual reports.Firm’s ownership structure is also likely to influence voluntary disclosure.

Tunisia is a developing country, with an emerging capital market, and whose

economic environment has been subject to major changes during the last years. In this

country, economic communication constitute an under regulated field and several Tunisian

firms don’t disclose sufficient information in their annual reports, that helps users evaluate the

management performance, lighten their perception of the firm and evaluate the future

profitability of their investment.

The main question of this research is: how far the structure of ownership and the

composition of the board of directors influence the offer of voluntary information in the

annual reports of listed companies in Tunisia?

The principal objective of this research is then to analyze the interaction between thecomposition of the board of directors (internal control mechanism) and the offer of voluntary

information in the annual reports (external control mechanism) in a context characterized by a

high ownership concentration.

This paper is organized as follows: the following section will be designed to present

the theoretical framework of transparency and voluntary disclosure. The third section will

present a revue of the previous research and the development of our hypothesis. The

verification strategy will be exposed in the fourth section while the fifth section presents a

discussion of our empirical results. And the last section includes our conclusions.

2. Theoretical framework of the transparency and voluntary disclosure in annual

reportsIn this section, we exhibit the theoretical framework on which our study is based.


3/28

3

2.1. Informational environment in the Tunisian stock exchange

2.1.1. How is voluntary disclosure regulated by Tunisian Law?

Yaich (2004) stipulates that management reports’ recommendations consist in

presenting a detailed management report. Thus, managers have wide margin to exercisediscretion in fixing the report content.

2.1.1.1. Legal requirements and regulatory framework for management report in

Tunisia

A- Commercial code (CSC)

According to the article 201 of CSC: “At the end of every fiscal year, the board of

directors establishes, under its responsibility, the financial statements of the firm according to

the law relative to the firm’s accounting

- the board of directors has to attach to the balance sheet a state of guarantees given by

the firm, and a state of securities she has agreed for.- he should, in conjunction with accounting documents, present to the annual meeting

a detailed annual report about the firm management”.

Among the obligatory documents communicated to the shareholders by the board of

directors of the firm, we find thus the management report. Nevertheless, regulators don’t give

any precision about the form nor the content of this report, except that it should be “detailed”.

B- Firms accounting system

The Tunisian accounting system (1997) has been created by adopting standards in

harmony with the IASB’s standards. Tunisian regulators didn’t give precisions about the

information to include in the management report. In this sense, the Tunisian accounting

framework stipulates in the section N° 83 that: “ other financial and non financialinformation, that the publication makes the information more useful, can be communicated

under the form of reports or separated states that completes the financial statements.” This

orientation shows that Tunisian standards setters are more aware about the information’s

importance that goes beyond the financial dimension to reach other dimensions such as social

dimension, ecological dimension, as well as technological dimension.

C- The board of financial market’s regulation relative to the initial public offerings

Contrary to the financial statements that have legal framework material enough and in

constant evolution with the firms accounting system, there are no accountant standards

governing the additional information to disclose in the management report.

Only article 44 of board of financial market’s regulation relative to the initial publicofferings, states some obligatory information to provide in the management report. This

article indicate that apart from the documents stipulated in article 3 of the law N° 94-117 of

14 November 1994, the firm should present to the board of financial market, a report on its

march foreseen by the article 201 of the commercial code.

In order to comply with the requirements of the article 3 of the law N° 94-117 of 14

November 1994 related to the reorganization of the financial market as modified by the law

N° 2005-96 of 18 October 2005, a proposal to amend the article 44 of the regulation of the

financial market board have been established in order to specify the content of the elements of

the internal control that should be provided in the annual management report. The new

version of the proposed article refers to an annex that establishes a standard model for the

management report gathering a set of recommendations provided par the legislation and theregulation within certain headings that might confer a better legibility to the report.


4/28

4

In addition, it is indicated in the new version of the article that the firm should insert in

its annual report other headings specific to its industry.

2.1.1.2. The means to control voluntary disclosure in Tunisia

The elastic perimeter that characterizes voluntary disclosure, the malleability of nonaudited and non standardized numbers, the suspicion that weighs upon the managers’ attitude,

shed light on the non credibility of the reporting information.

Accordingly to the article 266 of CSC, the verification of the concordance of the

information given in the management report with the annual accounts, is the responsibility of

the external auditors who gives their opinion toward financial statements.

2.1.2. Voluntary disclosure: a recent fact in Tunisia

El Aoun (2008) outlined that the strategy of reporting in Tunisian firm is in line with

the strategy of its former French colonizer.

2.1.2.1.

The development and the internationalization of capital markets

Trabelsi and al. (2005) outlined that in the last decades, we are witnessing a huge

mutations in the Tunisian economic environment, marked especially by the stimulation of the

financial market through motivating Tunisian firms by a financial direct access and the

improvement of the reporting of the listed firms.

The internationalization requires transparent economic information. Thus, the

presentation of annual reports with a high quality of disclosure guarantees the attraction of

foreign investors as well as local investors. According to Omran and al. (2008), the proportion

of the foreign blockholders in the Tunisian firms is 53% and that it is the highest percentage

comparing to Egypt, Oman and Jordan. Beyond their financial function, foreign investors

have an essential role in the reinforcement and the professionalization of the Tunisiancorporate disclosure. These investors required a very high level of transparency and

disclosure.

2.1.2.2. Some characteristics of the economy in Tunisia

A- The financing of firms by banks

The Tunisian Financial system is fragmented, dominated by banks. Moreover,

financial institutions, including insurance, investment and securities companies own

significant proportions of the shares in listed companies and are often among the five largest

block holders.B- An Emerging capital market

The legal framework of securities markets are relatively new, however in spite of the

financial liberalization and development programs, carried out in the last decade, the market is

characterized by a few number of listed enterprises.

C- A Low protection of minority shareholders

In the Tunisian setting where ownership is highly concentrated, legal system does not

protect minority investors. These latter are dominated by the controlling shareholders. This is

explained by the opaque nature of the firms’ disclosure policies and their lack of transparency.

Indeed, Tunisia has inherited the cultural values of the euro-continental model (Kamla, 2007),characterized by the uniformity, conservatism and discretion.

D- Corporate governance development


5/28

5

The corporate governance is composed by a set of mechanisms that has as function to

harmonize the managers’ attitude with the shareholders interest. The Tunisian financial

market moves toward a reinforcement of the right of information. In addition, corporate

governance rules encourage the improvement of the quality of transparency and disclosure.

In this sense, the guide of good corporate practices of the Tunisian enterprises (2008)insists in the right of the shareholders to a better transparency and disclosure. It affirms that

all firms that decide to adopt this guide should guarantee an equal treatment of all

shareholders and have to make sure that all the shareholders have all the required information

and all the ways that enable them exercise their rights.

2.2. Theoretical setting of voluntary disclosure

Voluntary disclosure is a game of contradictory powers (motivating forces, and

dissuasive forces). The process of voluntary disclosure results thus from an arbitration

between the economy of costs (agency costs, political, capital) that this publication can

procure to the company, and the generation of costs (direct and indirect) as a result of this

publication.

2.2.1. Incentives for voluntary disclosure

Voluntary disclosure is a discretionary act. Indeed, managers manage different levels of

disclosure for two types of factors:

- To reach economic objectives.

- To respect rules and societal standards.

2.3.1.1. Economic perspective

The economic perspective is based on the information asymmetry, and agency

conflicts between the managers and the financial information users. Healy and Palepu (2001)outlined three hypotheses about the theoretical justification of the voluntary disclosure:

financial market transactions, agency conflicts, and the signalization.

A- Financial market transactions

Matoussi, Karaa and Maghraoui (2004) confirm the existence of a positive relation

between voluntary disclosure and the liquidity of listed shares in Tunisia. Sengupta (1998)

outlined that capital cost decreases with specific corporate information. More precisely, he

has proven that firms with a good communication strategy (evaluated by financial analysts)

are more likely to have better liabilities conditions (lower effective interest rate). In addition,

the results of Botosan (1997) research suggest the existence of a negative relation between

financial information quality and capital cost.

B- Agency theory versus stewardship theory

Voluntary disclosure can reduce information asymmetry between the firm and its

shareholders. According to the agency theory, investors ask for more information when the

information asymmetry is very high between managers and shareholders, and that because of

the separation of the propriety and the control.

Information asymmetry between investors and managers results in two types of

problems. On one hand, an adverse selection problem can occur because managers can take

advantage from information that only them have access to. In the other hand, this situation of

information asymmetry engenders a moral hazard problem if manager’s behavior is in

contradiction with investors recommendations.Voluntary disclosure can mitigate these problems. Thus, in order to reduce bond costs

he supports, the manager has to show a good attitude of management. Shareholders want to


6/28

6

encourage a voluntary disclosure strategy insomuch as it allows them to have information

they need to evaluate the management, and thus reduce monitoring costs they support.

Basly (2006) and Klein, Pieper and Jaskiewicz (2005) affirm that in family firms same

actors have both the role of shareholder and manager. Thus, agency costs are absent in this

case. According to Morck and Yeung (2004) and Davis et al. (1997)1

, stewardship theory have been introduced in order to study relations between the actors, basing on behavioral

hypotheses that are different from the dominant paradigm, that is based on the agency theory.

According to the stewardship theory, managers don’t always have intention to maximize their

personal power and wealth. As stewards they can have the same preoccupations than

shareholders [Ngobo and Capiez (2004)].

C- Signals theory

Our study is based on signals theory: voluntary disclosure practices are destined to

inform shareholders and the capital market. In the annual reports firms have to justify, argue

and defend the position they adopt in different points. In this frame, the voluntary disclosed

items are signals addressed to the investors in the purpose to reduce information asymmetrybetween the “insiders” and the “outsiders”. Thus, voluntary disclosure decisions aim to

influence investor’s anticipations. Indeed, according to this theory, the existence of

information asymmetry between managers and potential investors will conduct these investors

to prudently evaluate the corporate shares. In the literature, this mechanism is known as anti-

selection.

2.3.1.2. Social perspective

Neo institutional sociologic theories provide a more appropriated frame to analyse the

communication of societal information (social and environmental).

A- Stakeholders information’s needs

Non financial information is useful for stakeholders to take the appropriate decisions.Damak-Ayadi (2005) affirm that according to stakeholders theory, the organization is

presented as been in the center of a set of relations between parties with diverse natures and

that are likely to be influenced by its activity or to influence it.

B- Public pressure

Legitimacy theory consider the society as a whole, while stakeholders theory

recognises that only some group in the the society are more powerful than others. In this

sense, Khor (2005) have presented a model that translates the link between the legitimacy, the

social contract and the corporate disclosure policy.

C- The processes and the institutional constraints

Ben Rhouma (2006) considers that the societal corporate disclosure quality is

regulated by its institutional setting that is with:

- What other firms in the same country and in the same industry disclose (mimetic

Isomorphism): In general, imitating another firm’s disclosure practices perceived as

the leader or the model is a good way for the firm to justify its actions (increase its

legitimacy).

- What the firm has disclosed in the past: The ritual that is the actions that result from a

highly habitual and standardized process, are well known as the cornerstone of the

institutions.

1 Mentioned by Trébucq (2003)


7/28

7

2.2.2. Argues in favor to the retention of the voluntary disclosure

The costs that are likely to be incurred by the firms for voluntary disclosed

information limit their communication.

2.2.2.1.

Direct costs (material)According to Depoers (2000), they are extra direct costs induced by the publication of

voluntary information (collect, treatment, production, diffusion, printing…). According to

Leuz and Wysocki (2006), it is difficult to quantify direct costs associated to disclosure

activity especially if they consist on opportunity costs such as managerial time.

2.2.2.2. Indirect (or strategic) costs

A- Propriety costs hypothesis

According to Trabelsi (2005): « The impact of the propriety costs on the firm value

and its competitive position can lead to retention of the information ». It can be noted that

propriety costs appear when a third party, whose interests are not aligned with the firm’sinterests, use the disclosed information against this firm welfare.

The studies of Clinch and Verrecchia (1997) and of Darrough and Soughton (1990)

show that the argument brought by the firms willing to restrain their disclosure is an argument

of a competitive disadvantage. According to the propriety costs theory, the firm can limit its

disclosure to avoid its strategic exploitation by competitors. This threat can thus limit the

voluntary disclosure in order to be protected from the risk of adverse action.

B- Litigation costs

The communication of certain information can, in some cases, significantly causes

harm to the reputation of the firm and risk to lead to legal pursuits. For example, inexact

forward looking financial statements’ publication can incite investors to sue the firm.

3. Literature review and hypotheses development

In this section, we present some empirical evidence then we develop our study

hypotheses linking voluntary disclosure level to factors in relation with corporate governance.

3.1. Board characteristics impacts on voluntary disclosure

3.1.1. Impact of the board independence on voluntary disclosure

The first objective to the corporate board is to assure its function of control. External

directors are perceived as being a way to control the manager’s behavior. In Tunisia, the

notion of non executive independents directors has been introduced in the article 196 of thecommercial code. Furthermore, the quality of shareholder is not required to be a member of

the board of a public company. The governance guide of good practices of Tunisian

companies (2008) call firms to appeal to independent directors that should be chosen for their

qualifications and expertise. To that purpose, this guide recommends that at least the third of

the board members should be independents.

The studies of Leung, Morris and Gray (2005), of Nasir and Abdullah (2005) and of

Cheung, Connelly, Limpaphayom and Zhou (2006) demonstrate the existence of a positive

relation between the independence of the board of directors and the volume of the voluntary

disclosed information. Indeed, external directors, can be at the origin of a better information

disclosure, by the exercise of their mission of control and their domination on the firm

decisions.


8/28

8

We suppose then, when the board is independent, it encourages a better management

control and a better voluntary disclosure. Hence, we propose the following hypothesis.

H 1.1: The extent of voluntary disclosure in annual report is positively related to the

independence of the board of directors.

3.1.2. The impact of the size of the board of directors on voluntary disclosure

The optimal size of the board of directors is a problem for the firm. A big board size is

difficult to coordinate. On the other hand, a small board size is a favorable field to

coordination, but, it can suffer from a lack of experience and competence of its members.

According to the governance guide of good practices of Tunisian companies (2008):

“Every firm is free to choose, according to its needs, the number of the members that

compose the board of directors in the limits of the law. The board of directors have to be

restrained enough to help a rapid decision making and as large as possible in order to take

advantage from the diversity of the competences and the experience of the members”.

According the article 189 of CSC: “The public company is directed by a board of directors

composed from at least three members and twelve at the most”.

Lakhal (2003) demonstrates, in the French context, that the size of board of directors is

not related to the decision of the results voluntary disclosure. Arcay and Vázquez (2005)

demonstrate that the impact of size of the board of directors on voluntary disclosure is not

significant. We suppose the existence of a non significant relation between the size of the

board of directors and voluntary disclosure. Thus, we suppose the following hypothesis.

H 1.2: There is no relation between the size of the board of directors and the extent of

voluntary disclosure in annual report.

3.1.3. Impact of the existence of a dominant personality on voluntary disclosure

The governance guide of good practices of Tunisian companies (2008) recommends aseparation between the functions of the chairman of the board and of the chief executive

officer (CEO), and this in the interest of efficiency. This guide announces that when the boarddecides to cumulate both functions, he is called to justify to shareholders the reasons for this

choice. Tunisian firm can choose a strict separation of functions as indicated by the article

215 of CSC.

The duality of the CEO role and the chairman of the board’s role constitute a risk for

shareholders. Indeed, the dominant personality can adopt a strategy in way that maximize his

personal interest because he plays at the same time the role of the supervisor and the

supervised.

The results of Gul and Leung (2004) indicate that there is a negative and significant

relation between results voluntary disclosure in annual report and the existence of a dominantpersonality. In this instance, the firms don’t disclose enough information, which is prejudicial

towards the shareholders. In the same sense, according to Bouri and Khlifi (2007), the duality

of functions results in decreasing the level of voluntary disclosure by 21%.

We wait for a negative relation between the duality of functions of CEO and of

chairman of the board and the extent of voluntary disclosure in annual reports. This

expectation leads us to develop the following hypothesis.

H 1.3: The extent of voluntary disclosure in annual report is lower for firms where the chief

executive officer is at the same time the chairman of the board.

3.1.4.

Impact of the directors’ ownership on voluntary disclosureAccording to Ben Ali (2005), when directors hold a small proportion of capital (week

managerial ownership), the risk to replace them is high and their performance is directly


9/28

9

associated to the firm’s performance. In this case, they have interest to maximize the firm

value, and then improve its communication strategy and respond better to investors’

expectations from disclosure especially when the firm’s performance is bad.

In the same sense, according to Mohd Ghazali and Weetman (2006), Siala Ghorbel

(2005) and Eng and Mak (2003), there is a negative relation between the managerialownership and the extent of voluntary disclosure. In addition, Trabelsi and al. (2005) outlined,

in the Tunisian setting, that the managerial ownership has a negative effect on voluntary

disclosure.

When the directors hold an important proportion of the company capital, the

ownership and the management are combined. Thus, the company has no interest to disclose

further information while agency costs are low. We propose then the following hypothesis.

H 1.4: The extent of voluntary disclosure in annual report is negatively related to the

directors’ ownership

3.1.5. Impact of combined Board’ characteristics on voluntary disclosure

Patelli and Prencipe (2007) insist on the interest of the researches that treat the

different mechanism of control and that constitute an interactive system. These mechanisms

reduce agency costs that results from the separation between the propriety and the

management.

The board of directors constitutes an important internal mechanism of control and it is

in the heart of the managers’ control mechanisms. Thus, we can’t talk about good corporate

governance without mentioning the efficiency of the board of directors. This board prepares

annual reports under its responsibility. Therefore, its composition structure affects the

strategic choices of disclosure in annual report and more specifically the extent of voluntary

disclosure.

Voluntary disclosure is considered as an external mechanism of control. It protectsshareholders against the opportunism of the managers. It contributes to diminish agency costs

resulting from information asymmetry. In other terms, corporate governance attributes are

introduced in order to control agency problems and assure that managers behave in the

interest of the shareholders. According to Li and Qi (2008), corporate governance guarantees

good quality of the information disclosed. Good corporate governance can reduce managers’

opportunism through reducing information asymmetry.

The first objective of the creation of a board of directors is to assure its functions of

control of agencies’ conflicts in order to reduce information asymmetry between shareholders

and financial backers in one hand, and between the managers and shareholders in the other

hand by providing more voluntary information.

Since the voluntary disclosure extent comes under the discretion of the board ofdirectors, we expect that the greater the degree of compliance with the guide of good

corporate practices of the Tunisian enterprises (2008), the greater the improvement in

corporate voluntary disclosure. Hence, we try to verify the following general hypothesis.

H 1: The quality of corporate governance positively affects the extent of voluntary disclosure

in the annual reports.

3.2. Impact of ownership structure on voluntary disclosure

3.2.1. Impact of ownership concentration on voluntary disclosure

According to Ho and Wong (2001), when a high proportion of the capital is held by a

low number of shareholders, conflicts of interest are not between managers and shareholders,

but between large and small shareholders. In such situation, managers have the incentive to


10/28

10

behave against the interests of smaller shareholders by reducing the quality of financial

disclosure. Indeed, a better financial disclosure helps reducing conflicts of interests between

large and small shareholders and allows protecting these latter.

Summa and Ben Ali (2006) outlined a negative and significant relation between the

quality of disclosure and ownership concentration. Therefore, voluntary disclosure is morelikely to be intensive in diluted ownership environment. In the same sense, Siala Ghorbel

(2005) stipulates that a diffused ownership increases agency costs and information

asymmetry, and as a result the needs of signaling. More the ownership concentration is low,

more the need to signalling increases and more the extent of disclosure is high.

However, Raffournier (1995) and Depoers (2000) didn’t detect a significant influence

of the ownership concentration on the extent of disclosure.

These controversial results bring the research around to a non linear relation between

voluntary disclosure and ownership concentration. According to Labelle and Schatt (2005),

there is a curvilinear relation between diluted ownership and the quality of financial

communication. In the light of the results of previous studies, we verify the following

hypothesis.

H 2.1: The extent of voluntary disclosure in annual report is negatively related to the level of

ownership concentration.

3.2.2. Impact of institutional ownership on voluntary disclosure

According to Summa and Ben Ali (2006), because of their professional experience and

their power on managers, Institutional investors have the means to make respect and apply

corporate governance principals in order to protect shareholders’ rights and wealth. They

disclaim a transparent communication that shows all the risk that the firm faces and the main

success key factors in order to better evaluate and estimate the distribution of future cash

flows. Trabelsi and al. (2005) affirm that institutional ownership improves voluntarydisclosure strategy. In addition, Ajinkya and al. (2005) affirm that firms that have an

important number of institutional investors disclose more forward looking information than

other firms. In addition, these anticipations are more frequent, more precise and less biased.

Indeed, these investors represent a guarantee for the market that the firm is protecting small

shareholders’ rights.

However Bushee and al. (2003) demonstrate the existence of a negative relation

between institutional ownership and the extent of voluntary disclosure. Bouri and Khlifi

(2007) also demonstrate that an increase of institutional ownership by 10% is associated with

a decrease of the extent of voluntary disclosure by 31.9%. We attempt thus to verify the

following hypothesis.

H 2.2: The extent of voluntary disclosure in annual report is positively related to the level of

institutional ownership.

3.2.3. Impact of family control on voluntary disclosure

According to the guide of good practices of Tunisian companies (2008): “in addition

to traditional mechanisms aimed at producing, sharing information, and assuring the

transparency of activities, family corporate governance should allow to :

- Introduce instruments, procedures, techniques and information system that enable to

produce in quality, quantity and in a suitable time, a reliable information about the real

functioning of the firm,

- Assure a quality and equality of information for all shareholders and all members of

the family…”


11/28

11

Chau and Gray (2002) and Trabelsi, Omri and Turki (2005) affirm the existence of a

negative relation between family control and voluntary disclosure. This can be explained by

the fact that family owned companies are less motivated to disclose beyond the obligatory

threshold. Ho and Wong (2001) demonstrate that firms that have a large proportion of family

members in the board of directors are more likely to have a less extent of disclosure.In a family owned company, ownership is concentrate in the hand of the same family.

That represents a particular case of ownership concentration. This family participates actively

in the management of the company and in most of time part of the direction or of board of

supervision or of directors. This allows the family to access to all information that enables it

to be reassured about the future of its investment. Family controlled firms tend to provide less

external information since asking for information is low [Summa and Ben Ali (2006)]. We

propose to verify the following hypothesis.

H 2.3: The extent of voluntary disclosure in annual reports is lower for firms controlled by a

family.

3.3.

Mediator effect of the composition of the board of directors on the influence of

the ownership concentration on voluntary disclosure

Having almost all the information, large shareholders dominate the boards of directors

[Jilani and Ben Hammouda (2006)]. Mak and Li (2001) affirm that there is a relation between

ownership structure and the board of directors’ composition.

In order to investigate the indirect effect of many firms’ characteristics (especially

ownership concentration) on voluntary disclosure, Arcay and Vázquez (2005) use internal

mechanism of governance as mediator variable. According to these authors, ownership

structure has a negative and significant impact on adopting rules of good practices and

governance that also affects corporate disclosure. Hence, there is an indirect relation between

ownership concentration and voluntary disclosure mediated by internal governancemechanisms.

We suppose that the mediator (board composition) represent a mechanism by which

ownership concentration influences voluntary disclosure extent. Ownership concentration

influences the mediator and this latter influence voluntary disclosure extent because it’s the

board of directors who prepares the annual reports (that contains voluntary disclosure) under

his responsibility. In other words, ownership concentration has an indirect effect on the extent

of voluntary disclosure. Thus, we attempt to verify the following hypothesis.

H 3: The composition of the board of directors is a mediator of the effect of the ownership

concentration on the extent of voluntary disclosure.

4.

Methodology

4.1. Sample and data collection

Our sample is composed by Tunisian companies listed in the BVMT. Our sample is

limited to enterprises that operate in non financial industry and this like: Hasnan (2005),

Bertrand (2000..., because of the characteristics of disclosure for the financial sectors (banc,

leasing, assurances…).

Collecting annual reports for our sample was not an easy task for us since they are not

directly downloadable via internet. These reports are copied from CMF and from the stock

exchange intermediate. Our data are extracted from the annuals reports collected and from the

link of the BVMT (http://www.bvmt.com.tn). They will be organized in the form of panel

data in order to increase the number of observations (because the number of listed company in


12/28

12

Tunisia is low) [Barako, Hancock and Izan (2006), Omri and al. (2005), Trabelsi and al.

(2005) and Hassan and al. (2006)].

Table 1: Distribution of observations

Years 2003 2004 2005

Number of firms 22 23 25

4.2.

Variables measures and research design

4.2.1. Dependant variable : voluntary disclosure extent

4.2.1.1. Elaboration of the index

We will develop an index of voluntary disclosure based on the index of Botoson

(1997) that is structured around five categories by completing it with three information

categories: information about intangible assets, social and environmental information and

information about governance. These categories of information are added in order to fill theinsufficiencies of the index of Botoson (1997). Thus a list of information likely to be

voluntarily disclosed is preset.

A- Information about intangible assets

This category includes information about intangible assets that aren’t taken into

consideration in the financial statements. This category is added to the grid of Botoson

(1997), because of the importance and relevance of the intangible assets that are increasing for

the firms as well as for investors. These items have been extracted from Trabelsi (2005)

index: description of the main clients, description of the main suppliers and description of the

axes of R&D.

B- Social and environmental information

Social and environmental information include the wording of social objectives of the

company and the description of its engagement toward the community. This category has

been added to the grid of Botoson (1997) for its eventual usefulness for making an investment

decision. The following items have been extracted from the index of Trabelsi (2005):

description of charity gifts, subventions, financial helps, description of the engagement

toward the community through social specific project (cultural, educative, recreational and

sportive) and the description of the activities that reduces the pollution linked to the firm

activities.

C- Information about corporate governance

Information about corporate governance has a major importance for the appreciation

of the company.

4.1.2.2. Measurement of the index of voluntary disclosure in annual reports

The methodology consists in reading carefully the annual reports of our sample of

companies. Then, we compare the information presented by each company with those

established in the grid. After this content analysis, an index is calculated for each company.

The methodology of points’ attribution can be weighted or unweighted.

A- Unweighted approach

The list of voluntary items will be applied to the annual reports of the firms. For each

one, a score of disclosure will be calculated. The first procedure used is dichotomous: an item

takes the value “1” if it is disclosed and “0” otherwise. This approach doesn’t reflect the

relative importance of each item and suppose that all the items have the same informationalinterest for the users. The amount of disclosure will be measured by dividing the score


13/28

13

obtained by the maximum possible score for each firm. Thus, the firms will not be penalized

by the non disclosure of some items if they are not relevant for their activities.

B- Weighted approach

We also use a weighted score, because we have chosen financial analysts as aparticular group of annual reports users. We have distributed 62 questionnaires to the

population of analysts. We obtained a level of response of 64.51%. The sample size of our

study is composed by 40 financial analysts and portfolio managers that work for 21

intermediate of the Tunisian stock exchange. Respondents have been called to express the

degree of importance they give to the items by giving a note in the Likhert 5 points scale

(with 1= not important to 5= very important).This approach consists in assigning to each item

a weighting that translates the relative importance of this item depending on the group of

users [Baker and Haslem (1973), Firth (1984), Bertrand (2000), Chow and Wong-Boren

(1987) and Michailesco (1999)]. Every diffused item is weighted by the mean of the points

that are attributed to him [Buzby (1975)]. The second used procedure is the following: an item

takes the “weight” that has been attributed to him by the financial analysts if it is disclosedand “0” otherwise.

4.2.2. Definition and measurement of the independent variables

4.2.2.1. The variables related to the ownership structure

A- Ownership concentration

Ownership concentration is measured by the percentage of shares held by the most

important shareholder of the company [Nekhili and Fakhfakh (2005) and Bouri and Khlifi

(2007)].

CONC = percentage of shares held by the dominant shareholder

B- Institutional ownership

This variable PINST is measured by the percentage of shares held by institutional

investors [Summa and Ben Ali (2006) and Trabelsi and al. (2005)].

PINST= percentage of shares held by institutional investors

C- Family control

The definition proposed by Rosenblatt de Milk, Anderson and Johnson (1985)2

will beadopted in our study. These authors consider a company as family controlled one when he

majority of the ownership or the control is in the hand of one family that at least two of these

family are directly involved in the management (members of the board). This variable FAM

is a dummy, and it takes “1” when the company is controlled by a family and “0” otherwise.

This measure has been adopted by Summa and Ben Ali (2006).

4.2.2.2. Variables relative to the characteristics of the board of directors

A- Independence of the board

2 Mentioned by Poulain-Rehm (2006)


14/28

14

We measure the independence of the board ICA by the proportion of the external

directors since they are supposed to be competent and more concerned by the shareholders’

interest. This measure has been adopted by Nasir and Abdullah (2005). In our study, we

consider as externals: the directors that has no familiar link with the managers and that don’t

have a direct and significant part in the capital of the firm.

ICA = number of external directors / number of directors

B-Size of the board

The size of the board of directors LNTCA is measured by the logarithm of the total

number of directors that compose the board. The use of this measure is justified by the

objective of mitigating heteroscedasticity problems. This measure has been used by Bouri and

Khlifi (2007).

LNTCA = Log (number of directors)

C-The existence of the structure of leadership in the boardThis variable DUAL is a dummy variable that takes “1” if one person holds at the

same time the functions of chief executive officer (CEO) and chairman of the board and “0”

otherwise.

D- Directors’ ownership

Like Bouri and Khlifi (2007), Eng and Mak (2003) and Trabelsi and al. (2005) we

measure managerial ownership (PADM) by the percentage of shares held by the members of

the board.

PADM = percentage of shares held by directors

E- Governance quality index

Arcay and Vázquez (2005) and Hajri and Omri (2005) insist on the interest to

synthesize all the governance’s mechanism in one variable because of the effect of

complementarities and substitutability that exist between them. We construct an index of the

governance quality IGOUV that is a variable that synthesize the characteristic of the boardexcept the dummy variable DUAL. We didn’t take into consideration this variable because

we should use only continuous variables for factorial analysis. IGOUV is determined by a

factorial analysis that integrates the variable (ICA, TCA and PADM). It will constitute a

common factor that will be then used as a proxy for corporate governance quality.

We have chosen this index because the inclusion of all the variables in the same model

can be in the origin of multicolinearity problems. The choice of this methodology results

essentially from a need to gather in an objective way the variables that characterizes the

composition of the corporate board of directors.

Before starting factorial analysis, we have verified that the three variables are

significantly correlated (according to the correlation matrix). Indeed, the partial correlation of

the variables (taken two per two) presented by the index KMO is strong (0,624 > 0,5) which

means that the variables are highly correlated and thus we can wait for a weak number of

factors. In addition, Bartlett test indicates that the matrix of correlation is different from the

matrix of identity Test of Bartlett, chi-2 = 18,577; Sig = 0,000) which confirm that the

variables are highly correlated and thus factorable.

4.2.3. Definition and measure of control variables


15/28

15

4.2.3.1. Competitiveness in the commodity market

Competitiveness in the market of goods and services is weak when the barrier of entry

of the sector is important, that is when the risk of new entries is low. We will use the

capitalistic intensity as proxy for the barriers of entry of the sector [Mami (2003), Aerts,

Cormier and Magnan (2004), Raffournier (1995), Chow and Wong-Boren (1987), Clarksonand al. (1994) and Depoers (2000)]. The variable INTCAP is measured as follows:

INTCAP = gross fixed assets / total assets

We expect a positive relation between the capitalistic intensity and the extent of

voluntary disclosure in annual reports.

4.2.3.2. Size of the firm

Ben Ali (2005), Lakhal (2003), Depoers (2000), Raffournier (1995), Buzby (1975),

Hassan, and al. (2006), Alsaeed (2005), Aksu and Kosedag (2006) and Lang and Lundholm

(1993) have confirmed the existence of a positive and significant relation between the firm

size and the quality of financial reporting.Firm size is measured by the logarithm of total assets LNTA. The use of the logarithm

is justified by the objective of mitigating heteroscedasticity problems.

LNTA = Log (total assets)

We expect a positive relation between the size of the firm and the extent of voluntary

disclosure in annual reports.

4.2.3.3. Indebtedness of the firm

Leung and al. (2005) have found a positive relation between indebtedness and

voluntary disclosure. Naser and al. (2006) affirm that the firms with high level of debts are

considered more risky. These companies meet difficulties during capital increase. And it is forthis reason that they are called to disclose more information explaining their financial

situation.Indebtedness (END) is measured by the weight of creditors compared with

shareholders. Thus END is measured by dividing financial debts by equity.

END = total debts / equity

We wait for a positive relation between the indebtedness of the firm and the extent of

voluntary disclosure in annual report.

4.2.3.4. Firm age

Gibbins and al. (1990) outlined that the history of the company influences its strategy

of reporting through:

- The traditions of the enterprise (the traditions, the way of doing things)

- The organizational training of individuals.

Haniffa and Cooke (2002) have used the variable listing duration among the variables

that characterize the firm. Alsaeed (2005) has used the variable age among the factors that can

influence the level of voluntary disclosure because mature firms have the possibility to

improve their disclosure across the time.

The age of the company LNCOT is measured by the logarithm of the duration of the

firm listing in the stock exchange in years. The use of the logarithm is justified by theobjective of mitigating heteroscedasticity problems.


16/28

16

LNCOT = Log (duration of the listing of the firm in the stock exchange in years)

We expect a positive relation between the period of listing and the extent of voluntary

disclosure.

4.2.3.5.

Auditors’ quality

In the meta-analysis made by Ahmed and Courtis (1999), it seems that the results of

the researches that investigate the effect of the quality of the auditor on the extent of

disclosure are mitigated. Some have confirmed a positive and significant relation; others have

found a positive but not significant relation, and others found no relation between the two

concepts, while others confirmed the existence of a negative relation. Archambault and

Archambault (2003), Leung and al. (2005) and Raffournier (1995) demonstrate that the firms

that are audited by an auditor among the big 4 audit companies disclose more information

than the others

This variable QAU is a dummy variable and takes “1” if the firm is audited by a big 4

and “0” otherwise. This measure has been adopted by Archambault and Archambault (2003)and by Raffournier (1995).

We wait for a positive relation between the auditor quality and voluntary disclosure.

Indeed, big 4 audit companies motivate the audited companies to disclose information beyond

the obligatory threshold.

4.2.4. Empirical design

IDIV it = 0 + 1 INTCAPit + 2 CONC it + 3 FAM it + 4 PINST it + 5 DUAL it + 6

NTCA it + 7 ICA it + 8 QAU it + 9 LNCOTit + 10 END it + 11 LNTA it + 12 PADM it+ it (Model 1) IDIV it = a0 + a1 INTCAPit + a2 CONC it + a3 FAM it + a4 PINST it + a5 DUAL it + a6

IGOUV it + a7 QAU it + a8 LNCOTit + a9 END it + a10 LNTA it + it(Model 2)IDIV= index of voluntary disclosure. IGOUV= factor of variables (ICA. TCA et PADM). CONC = percentage of

shares held by the dominant shareholder. PINST= percentage of shares held by institutional investors. FAM=1

when the firm is controlled by a family and =0 otherwise. ICA = number of external directors / number of

directors. LNTCA = Log (number of directors). DUAL =1 if one person holds at the same time the functions of

CEO and chairman of the board and =0 otherwise. PADM = percentage of shares held by directors. INTCAP =

gross fixed assets / total assets. LNTA = Log (total assets). END = total debts / equity. LNCOT = Log (duration

of the listing of the firm in the stock exchange in years). QAU=1 if the firm is audited by a big 4 and = 0

otherwise.

Our study is based on the panel data, that’s why it is convenient to verify the

homogenous specification of the generator process of the data. We should then distinguish

between the specific effect and the common effect through Fisher statistic (test of Chow).

5. Empirical results

In the following section we will analyze and comment our different empirical results.

5.1. Descriptive statistics

We are particularly interested in examining the evolution of the weighted and

unweighted index of voluntary disclosure IDIV between 2003 and 2005.

Table 2: Evolution of the weighted and unweighted indexes of voluntary disclosure2003 2004 2005

IDIVunweighted

IDIVweighted

IDIVunweighted

IDIVweighted

IDIVunweighted

IDIVweighted

Mean 35,36 35,73 40,99 41,34 40,00 40,55


17/28

17

Median 38,36 38,02 38,00 38,53 41,94 42,65

Standard Deviation 11,45 11,27 12,58 12,24 11,46 11,92

Variation coefficient 0,323 0,315 0,306 0,296 0,286 0,294

N 22 21 24

IDIV= index of voluntary disclosure

Through this table, we notice an increase in the extent of voluntary disclosure between2003 and 2004. The studies of Bughin and al. (2007) and of Naser and Nuseibeh (2003) also

demonstrate that the level of voluntary disclosure in annual reports tend to increase across the

time. We also notice that the standard deviation is important comparing to the mean, which

make us conclude the existence of variability in the behavior of voluntary disclosure among

the firms of our sample. In addition we have done an additional analysis like Singleton and

Globerman (2002) to verify if the voluntary disclosure practices tend to diverge or converge

among the enterprises of our sample. The comparison of the coefficient of variation of the

indexes IDIV of 2003 and 2004, and of 2004 and 2005 allow us to detect a decrease of this

coefficient which means that voluntary disclosure behavior in our firm sample tend to

converge across the time.

We have applied the non parametric test of means comparison of Mann-Whitiney on

the of voluntary disclosure global index variable IDIV weighted and unweighted across the

years: 2003-2004, 2003-2005 and 2004-2005. The procedure of tests for two independent

samples compares two groups of observations according to a continuous variable.

Table 3: Results of Mann-Whitiney test2003-2004 2004-2005 2003-2005

IDIV

unweighted

IDIV

weighted

IDIV

unweighted

IDIV

weighted

IDIV

unweighted

IDIV weighted

Year 2003 2004 2003 2004 2004 2005 2004 2005 2003 2005 2003 2005

Mean rank 19,81 24,28 19,77 24,33 23,28 22,75 23,04 22,95 20,5 26,25 20,54 26,20

U of Mann-

Whitney

183 182 246 251 198 199

Sig. 0,24 0,23 0,89 0,98 0,14 0,15

IDIV= index of voluntary disclosure

According to the mean rank, we notice that the extent of voluntary disclosure increases

from 2003 to 2004 (24,28 >19,81 and 24,33 >19,77) and becomes stable from 2004 to 2005

(23,28 22,75 and 23,04 22,95). These tests are significants, which make us accept the null

hypothesis H0 of equality of means between the three groups of the firms (taken two by two).

This means that the extent of voluntary disclosure doesn’t significantly vary across the three

observed years. But it should be noted that the extent of voluntary disclosure is more

important for the year 2005 compared to the year of 2003 (26,25 > 20,5 and 26,20 > 20,54 ;

sig=14% and 15%).

5.2. Descriptive statistics and univariate and bivariate tests

5.2.1. Comparison between weighted IDIV and unweighted IDIV

We have used the non parametric test of Wilcoxon to demonstrate the existence of a

difference between two variables in the same sample.

Table 4: Results of Wilcoxon testN Mean rank

Negative ranks 14

31,57 IDIV weighted - IDIV unweighted

Positive ranks

53

34,64

Z (based on negative ranks) -4,354

sig 0,000


18/28

18

The table 3 shows that there are few differences between weighted IDIV and

unweighted IDIV. In spite of that, we found a significant difference (at 1%) between the two

measures of the voluntary disclosure index, which means that financial analysts give a

different interest to voluntary disclosed items in annual reports. This result corroborates with

the results of the study of Naser and Nuseibeh (2003), but it is contradictory with the resultsof the study of Chow and Wong-Boren (1987). It should be noted that the coefficient of

correlation between the weighted IDIV and unweighted IDIV is 0.99 and is significant at 1%.

It is suitable then to use the weighted IDIV variable in our multivariate analysis.

5.2.2. Descriptive statistics of independent variables

Table 5: Descriptive statistics of independent variablesN Mean Median Std Dev

Independent and continuos variables

CONC 67 37,59 35 18,01

PINST 67 19,33 9,6 22,87

LNTCA 66 2,09 2,19 0,26

ICA 66 20,60 14,28 22,65

PADM 66 59,17 61,8 20,03

IGOUV 66 1,31 0,19 1

Control variables

INTCAP 68 84,57 72,79 119,28

LNCOT 69 1,86 1,94 0,73

END 68 199,50 123,45 394,20

LNTA 68 17,90 17,77 0,98

Binary variables N Variable = 1 Variable = 0 Median

FAM 67 19

(28,35%)

48

(71,64%)

0

DUAL 66 48

(72,72%)

18

(27,27%)

1

QAU 69 28

(40,57%)

41

(59,42%)

0

IGOUV= factor of variables (ICA. TCA et PADM). CONC = percentage of shares held by the dominant

shareholder. PINST= percentage of shares held by institutional investors. FAM=1 when the firm is controlled by

a family and =0 otherwise. ICA = number of external directors / number of directors. LNTCA = Log (number ofdirectors). DUAL =1 if one person holds at the same time the functions of CEO and chairman of the board and

=0 otherwise. PADM = percentage of shares held by directors. INTCAP = gross fixed assets / total assets. LNTA

= Log (total assets). END = total debts / equity. LNCOT = Log (duration of the listing of the firm in the stock

exchange in years). QAU=1 if the firm is audited by a big 4 and = 0 otherwise.

This table shows that the firms of our sample are characterizes by a high ownership

concentration (the minimum of this variable is equal to 11%) and especially by a very high

managerial ownership. The firms investigated are highly indebted and have a capitalisticintensity. The standard deviation of the variables PINST and ICA are very important

comparing to the means, which demonstrates the existence of a high disparity between the

firms of our sample. This result justifies the use of panel data to control the heterogeneity of

the observation in their individual dimensions.

5.2.3. Correlation analysis

The matrix of correlation of Spearman doesn’t show any correlation higher than “0.6”

between the independent variables. We have also calculated the VIF. It is under the threshold

of “3” for all the independent variables, but it exceeds the value of “2” for the following

variables: CONC, PADM, DUAL and ICA. From where there is a very little problem of

mutlicolinearity in our Model 1. It is then suitable to replace the variables (ICA, TCA andPADM) by the variable IGOUV (already defined) in our Model 2 to avoid any source of

multicolinearity.


19/28

19

5.3. Panel data tests

We have used STATA to make the different statistics tests and the regressions of this

sub-section.

5.3.1. Test of presence of individual effect

Table 6: Test of Chow of the presence of individual effectModel 1 Model 2

Fisher test 4,66***

(0,0000)

6,09***

(0,0000)

*** significant at 1%

The results of this test allow to reject the null hypothesis H0 and to accept the

alternative hypothesis: the presence of individual effects.

5.3.2. Test of Hausman

Table 7 : Test of HausmanModel 1 Model 2Chi-2 test 12,72

(0,2398)

7,80

(0,5543)

We have, also, used the test of Hausman in order to specify the model whether by

taking into consideration the individual fixed or random effect. The Hausman test is not

significant for the Model 1 and for the Model 2, thus, we apply the specification in random

effects.

5.3.3. Heteroscedasticity test

Table 8: Test of Breusch-PaganModel 1 Model 2

Chi-2 test 5,67**

(0,0173)

5,80**

(0,0160)

** significant at 5%

Breush-Pagan test have confirmed the existence of heteroscedasticity problem for both

Model 1 and Model 2. Thus, we use the method of GLM (generalized least squares) that takes

into consideration the presence of heteroscedasticity.

5.4. Interpretation of the results of the multivariate analysis

5.4.1. Results of Model 1

Table 9: Results of Model 1IDIV weightedVariables Pred sign

Coefficients z-statistic

Intercept ? 60,70278*** 3,03

ICA + -0,1899977*** -3,03

LNTCA - / + -5,982983 -1,60

DUAL - -10,13566*** -3,29

PADM - 0,1491854** 2,00

CONC - 0,1306006 1,33

PINST + 0,0197181 0,31

FAM - -5,229894** -2,26

INTCAP + -0,0250128* -1,74

LNTA + -0,2370524 -0,20QAU + 1,574098 0,50

LNCOT + -2,237565 -1,16


20/28

20

END + 0,0023512 0,86

N

Log likelihood

Wald Chi-2

Prob > Chi-2

64

-222,730761,53***

0,0000

*, ** et *** significant at 10%, 5% et 1% respectivelyIDIV= index of voluntary disclosure. CONC = percentage of shares held by the dominant shareholder. PINST=

percentage of shares held by institutional investors. FAM=1 when the firm is controlled by a family and =0

otherwise. ICA = number of external directors / number of directors. LNTCA = Log (number of directors).

DUAL =1 if one person holds at the same time the functions of CEO and chairman of the board and =0

otherwise. PADM = percentage of shares held by directors. INTCAP = gross fixed assets / total assets. LNTA =

Log (total assets). END = total debts / equity. LNCOT = Log (duration of the listing of the firm in the stock

exchange in years). QAU=1 if the firm is audited by a big 4 and = 0 otherwise.

The negative and statistically significant coefficient of the variable ICA highlights the

fact that when the percentage of external directors in the board increases by 10%, the extent of

voluntary disclosure decreases by 18.99%. Thus, our hypothesis H1.1 is confirmed. This

result corroborates with the study of Bouri and Khlifi (2007) who found that when the

percentage of independent members in the board of directors increases by 10%, the level ofvoluntary disclosure decreases by 47.4%. The studies of Eng and Mak (2003) and of Barako

and al. (2006) also highlight the existence of a negative relation between the percentage of

external directors and the extent of voluntary disclosure. Nevertheless, this result, that is not

logical, can be explained by the choice of the measure used for the variable ICA. It’s probable

that the used criterion is not sufficient to qualify an external director. But the use of these

chosen criteria is justified by the lack of data about the identity of the directors of the Tunisian

firms. This result can also be explained by the effect of substitution between voluntary

disclosure and the proportion of external directors. A good voluntary disclosure can replace

the absence or the lack of external directors. The inverse reasoning is also valid. The presence

of external directors can take the place of voluntary disclosure.

The link between voluntary disclosure and the size of the board is not significant. This

makes us confirm our hypothesis H1.2. This result is coherent with the results of Lakhal

(2003) and of Arcay and Vázquez (2005).

Our result for the variable DUAL corroborates with the results of Gul and Leung

(2004), but it’s in contradiction with the results of Ho and Wong (2001), of Arcay and

Vázquez (2005), of Raffournier (1995) and of Barako and al. (2006) that reveal a negative but

not significant relation between the extent of voluntary disclosure and the existence of a

dominant personality (that holds the functions of CEO and chairman of the board). Our results

lead us to confirm the hypothesis H1.3 and thus to highlight the fact that the structure of the

leadership of the board of directors negatively influence the extent of voluntary disclosure.

The coefficient of the variable PADM is significant and its sign is contrary to theexpected sign. This result is not coherent with the results of Eng and Mak (2003), of Mohd

Ghazali and Weetman (2006), of Siala Ghorbel (2005) and of Trabelsi and al. (2005) and

makes us infirm our hypothesis H1.4 and to affirm that more the managerial ownership

increases, more the voluntary disclosure extent increases.

For the variable CONC, its coefficient is positive (contrary to the expected sign) and

not significant which make us infirm the hypothesis H2.1. This result joins the results of the

studies of Loukil and Triki (2008) and of Bouri and Khlifi (2007), conducted in the Tunisian

context using cross sectional data. This positive relation we have found can be explained by

the fact that when the capital is highly concentrated, the problems of interest’s conflicts

between small shareholders and large shareholders don’t push the latter to minimize voluntary

disclosure. The measure we used may not be suitable for our context of research.


21/28

21

As far as the variable PINST is concerned, its coefficient is positive and not

significant, which infirm the hypothesis H2.2. This result doesn’t corroborate with the results

of Trabelsi and al. (2005) and of Ajinkya and al. (2005).

The coefficient of the variable FAM is negative as expected and is significant, which

confirm the hypothesis H2.3 and corroborate the results of Chau and Gray (2002) and ofTrabelsi and al. (2005). This means that the family controlled firms disclose less than the

other firms.

The control variables have not significant coefficients except for “capitalistic

intensity” variable. Its negative sign makes us affirm that when the entry barriers of the

sectors are important, the extent of voluntary disclosure is weak. But, what attracts our

attention, is the fact that our results show that the size of the firm has a negative but non

significant effect on the extent of voluntary disclosure. This result is matching to neither the

sign nor the results of previous empirical evidences of Buzby (1975), of Chow and Wong

(1987), of Ho and Wong (2001) and of Raffournier (1995).

5.4.2.

Results of Model 2Table 10: Results of Model 2

IDIV weightedVariables Pred sign

Coefficients z-statistic

Intercept ? 79,85696*** 3,74

IGOUV + 4,250724*** 4,88

DUAL - -7,418199** -2,38

CONC - 0,1190179 1,49

PINST + 0,0166937 0,28

FAM - -5,235595** -2,33

INTCAP + -0,0165384 -1,20

LNTA + -1,580097 -1,30

QAU + 3,064574 1,19LNCOT + -4,84571*** -2,72

END + 0,001728 0,60

N

Log likelihood

Wald Chi-2

Prob > Chi-2

64

-225,0055

41,78***

0,0000

*, ** et *** significant at 10%, 5% et 1% respectively

IDIV= index of voluntary disclosure. IGOUV= factor of variables (ICA. TCA et PADM). CONC = percentage

of shares held by the dominant shareholder. PINST= percentage of shares held by institutional investors.

FAM=1 when the enterprise is controlled by a family and =0 otherwise. DUAL =1 if one person holds at thesame time the functions of CEO and chairman of the board and =0 otherwise. INTCAP = gross fixed assets /

total assets. LNTA = Log (total assets). END = total debts / equity. LNCOT = Log (duration of the listing of the

firm in the stock exchange in years). QAU=1 if the firm is audited by a big 4 and = 0 otherwise.

The coefficient of the variable IGOUV is significant and its sign is positive. This

means that the quality of corporate governance influence significantly the extent of voluntary

disclosure. This result allow us confirm our hypothesis H 1. This result is logical because

disclosure is an integral part of the corporate governance system. A good quality of corporate

governance allow then to improve the extent of voluntary disclosure that is considered as a

mean to control the managers, to protect the shareholders and to decrease agency costs

resulting from information asymmetry.

We found that the coefficient of the variable LNCOT is significant but doesn’t have

the expected sign. This makes us confirm the fact that the firms recently listed in the Tunisian

stock exchange voluntary disclose more than old firms. The new firms in the market disclosemore than the others in objective of signaling.


22/28

22

5.5. Tests of the mediator effect of the composition of the board of directors on the

influence of the ownership concentration o voluntary disclosure

In what follows, we apply the regression approach (simple and multivariate) and we

test the four conditions enunciated by Caceres and Vanhamme (2003) in addition to the

specific tests of the panel data.Table11: First step of the test of the mediator effect

Estimated Equation: IDIV weighted it = 0 + 1 CONC it + ait

Variables Pred Sign Coefficients z-statistic

Intercept ? 37,23247*** 15,06

CONC - 0,0613336 0,96

N

Test of Chow

Prob > Fisher

Test of Hausman

Prob > Chi-2

Test of Breusch-Pagan

Prob > Chi-2

Log likelihood

Wald Chi-2

Prob > Chi-2

657,08***

0,0000

3,80*

0,0514

22,45***

0,0000-240,8149

0,92

0,3375

*, ** et *** significant at 10%, 5% et 1% respectively

IDIV= index of voluntary disclosure. CONC = percentage of shares held by the dominant shareholder.

We find that the relation between the ownership concentration and the voluntary

disclosure is positive and statistically significant. Thus, our first condition is not verified since

the coefficient of the variable “CONC” is not significant.

Since our first condition is not verified, it is useless to verify the conditions 2, 3 and 4

of the existence of a mediator effect. We can’t thus confirm the existence of the mediator

effect of the composition of the board of directors on the relation between the concentration of

ownership and voluntary disclosure. This result doesn’t match with the study of Arcay andVázquez (2005) and then our hypothesis H 3 is infirmed.

6.

Conclusion

The major interest of this study is that it contributes to the analysis of the behaviors of

the Tunisian firms in matter of voluntary disclosure in annual reports and the examination of

the effects of the board composition and the ownership structure on voluntary disclosure. The

institutional setting of Tunisia is of interest because it is characterized by its high ownership

concentration, low level of investor protection and poorly developed capital market.

The results of our study highlight that the extent of voluntary disclosure tends to

improve through the time. According to the used statistic tests, we have proved that the

independence of the structure of the leadership of the board of directors and the familyownership don’t improve the extent of voluntary disclosure. We have demonstrated, in the

other hand, that the size of the board of directors has no effect on the extent of disclosure.

Nevertheless, we find that managerial ownership and the quality of corporate governance are

represented as incentive factors to corporate voluntary disclosure.

This study has both theoretical and practical implications. From a theoretical

standpoint, our analysis reveals a positive relationship between the two control mechanisms

(i.e. board composition and voluntary disclosure). From a practical standpoint, the study

offers insights to policy makers and regulators in order to evaluate the effectiveness of

corporate governance rules and the interaction between control systems.

However, our study has certain limits. The most important are its small sample and the

manual analysis of the content of the annual reports. It is convenient, also to conceive an

index of voluntary disclosure prepared according to the needs of the Tunisian annual reports


23/28

23

users instead of using the modified index of Botoson (1997). The use of the classic factorial

analysis for the construction of a governance index is also questionable: the non linear

factorial analysis allows us include qualitative variables such as the structure of the leadership

of the board. Also, the lack of publicly available information on Tunisian listed companies

limited to some extent our empirical research. This lack of data stopped us from refiningmeasures of variables.

Core (2001), Lim and al. (2007), Gul and Leung (2004), Cheng and Courtenay (2006)

and Nikolaev and Van Lent (2005) affirm that the endogeneity problem embarrasses most of

the empirical studies that treat voluntary disclosure. Therefore, it is important to examine the

problem related to the endogenous nature of the quality of corporate governance through the

verification of the existence of a bi-directional relation between the quality of corporate

governance and the extent of voluntary disclosure.

REFERENCES

Ahmed K et Courtis J K, (1999), « Associations between corporate characteristics anddisclosure levels in annual reports: a Meta Analysis », British Accounting Review, vol 31, pp

35-61

Aerts W, Cormier D et Magnan M, (2004), « Performance disclosure transparency by

continental European and north American firms: Contrasting practices and implications »,

working paper , Ecole des sciences de gestion, UQAM, Canada

Ajinkya B, Bhajraj S et Senguta P, (2005), « The association between outside directors,

institutional investors and the properties of management earnings forecasts », Journal of

Accounting Research, vol 43, n°3, pp 343-376

Aksu M et Kosedag A, (2006), « Transparency and disclosure scores and their determinants in

the Istanbul stock exchange », Journal compilation, vol 14, pp 277-296

Alsaeed K, (2005), « The association between firm-specific characteristics and disclosure:

The case of Saudi Arabia », Journal of American academy of business, Cambridge, vol 7, n°1,

pp 310-321

Arcay R B et Vázquez F M, (2005), « Corporate characteristics, governance rules and the

extent of voluntary disclosure in Spain », Advances in Accounting, vol 21, pp 299-331

Archambault J J et Archambault M E, (2003), « A multinational test of determinants of

corporate disclosure», The international Journal of Accounting, vol 38, pp 173-194

Baker A K et Haslem J A, (1973), « Information needs of individual investors », The journal

of accountancy, pp 64-69

Barako D G, Hancock P et Izan H Y, (2006), « Relationship between corporate governance

attributes and voluntary disclosure in annual reports: The Kenyan experience », working

paper, University of Western Australia

Basly S, (2006), « Propriété, décision et stratégie de l’entreprise familiale : Une analyse

théorique », working paper , Université Montesquieu - Bordeaux 4

Ben Rhouma A, (2006), « Mesure de la qualité de la divulgation sociétale des entreprises

françaises », working paper , Congrès de l’AFC 2006 à Tunis, CNRS-UNSA, France

Ben Ali C, (2005), « La communication financière et la structure de propriété: le cas

français», working paper , Congrès de l’IFC 2005 à Hammamet, Université Paris Dauphine IX


24/28

24

Bobet F et Flageul C, (2004), « Le rapport de gestion », Revue française de comptabilité

n°363

Bouri A et Khlifi F, (2007), « L'impact des mécanismes de gouvernement d’entreprise sur le

niveau de divulgation volontaire : Etude empirique dans le contexte tunisien », working

paper , Congrès de l’IFC 2007 à Hammamet, Université de Sfax

Botosan C,A, (1997), « Disclosure level and the cost of equity capital », The Accounting

Review, vol 72, n°3, pp 323-350

Bushee B et Noe C, (2000), « Corporate disclosure practices, institutional investors, and stock

return volatility», Journal of Accounting Research, vol 38 (Supplement), pp171-202

Bushee, B J, Matsumoto D A et Miller G S, (2003), « Open versus closed conference calls:

the determinants and effects of broadening access to disclosure », Journal of Accounting and

Economics, vol 34, pp 149-180

Bughin C, Colot O, Croquet M et Niyuhiré P, (2007), « L'approche stakeholders de la

performance : étude de rapports annuels de sociétés cotées belges », working paper , Congrèsde l’IFC 2007 à Hammamet, Université de Mons-Hainaut, Belgique

Buzby S L, (1975), « Company size, listed versus unlisted stocks, and the extent of financial

disclosure », Journal of Accounting Research, vol 13, n°1, pp16-37

Caceres B C et Vanhamme J, (2003), « Les processus modérateurs et médiateurs : distinction

conceptuelle, aspects analytiques et illustrations », Recherche et applications en marketing,

vol 18, n°2, pp 67-100

Cheung S Y L, Connelly J T, Limpaphayom P et Zhou L, (2006), « Determinants of corporate

disclosure and transparency: Evidence from Hong Kong and Thailand », working paper ,

Chulalongkorn UniversityChow C W et Wong-Boren A, (1987), « Voluntary financial disclosure by Mexican

corporations », The accounting review, vol 62, n°3, pp 533-541

Chen S, Chen X et Cheng Q, (2006), « Do family firms provide more or loss voluntarydisclosure? », working paper , University of Washington Business School

Cheng E C M et Courtenay S M, (2006), « Board composition, regulatory regime and

voluntary disclosure », The international journal of accounting, vol 41, p 262-290

Chen J P et Jaggi B L, (2000), « The Association between independent non executive

directors, family control and financial disclosures », Journal of Accounting and Public Policy,

vol 19, pp 285-310Chau G,K et Gray S J, (2002), « Ownership structure and corporate voluntary disclosure in

Hong Kong and Singapore », The international Journal of Accounting, vol 37, n°2, pp 247-

265

Clinch G et Verrecchia R E, (1997), « Competitive disadvantage and discretionary disclosure

in industries », Australian journal of management, vol 22, n°2, pp 125-138

Clarkson P M, Kao J L et Richardson G D, (1994), « La présentation volontaire d’information

à caractère prévisionnel dans le rapport de gestion intégré aux rapports annuels », Recherche

comptable contemporaine, vol 11, n°1, pp 451-488

Core J E, (2001), « A review of the empirical disclosure literature: discussion », Journal of Accounting and Economics, vol 93, n°1, pp 441-456


25/28

25

Code des sociétés commerciales (CSC)

Darrough M et Soughton N, (1990), « Financial disclosure policy in an entry game », Journal

of Accounting and Economics, vol 12, pp 219-243

Damak-Ayadi S, (2005), « Analyse des déterminants de la diffusion d’informations sociétalesdes rapports annuels par les entreprises françaises », working paper , Congrès de l’ATSG 2005

à Hammamet, Université de Paris Dauphine

Depoers F, (2000), « A cost-benefit study of voluntary disclosure: some empirical evidence

from French listed companies », The European Accounting Review, vol 9, n° 2, pp 245-263

El Oun M, (2008), « Les déterminants de la transparence et de la divulgation des entreprises

cotées sur les marchés émergents de la zone du moyen orient et de l’Afrique du nord »,

Mémoire pour l’obtention de Mastère en comptabilité, FSEGS

Eng L L et Mak Y T, (2003), « Corporate governance and voluntary disclosures », Journal of

Accounting and Public Policy, vol 22, n°4, pp 325-345

Firth M, (1984), « The extent of voluntary disclosure in corporate annual reports and its

association with security risk measures », Applied Economics, vol 16, pp 269-277

Forker J, (1992), « Corporate governance and disclosure quality », Accounting and Business

Research, vol 22, n° 86, pp 111-124

Gibbins M, Richardson A et Waterhouse J, (1990), « The management of corporate financial

disclosure: opportunism, ritualism and processes », Journal of Accounting Research, vol 28,

n°1, Spring, pp 121-143

Gelb D S, (2000), « Managerial ownership and accounting disclosure: An empirical study »,

Review of Quantitative Finance and Accounting, vol 15, n° 2, pp,169-185

Guide de bonnes pratiques de gouvernance des entreprises tunisiennes, (2008), Institut arabe

des chefs d’entreprise

Gul F A et Leung S, (2004), « Board leadership, outside directors’ expertise and voluntary

corporate disclosures », Journal of Accounting and Public Policy, vol 23, pp 351-379

Hajri J et Omri A, (2005), « Qualité de gouvernance et performance boursière des entreprises

tunisiennes », working paper , Congrès de l’IFC en Mars 2005 à Hammamet, Institut supérieur

de gestion de Tunis

Hasnan A, (2005), « Corporate voluntary reporting practices in India », The cost and

management, vol 33, n°5, pp 73-79

Hassan O A G, Giorgioni G et Romilly P, (2006), « The extent of financial disclosure and its

determinants in an emerging capital market: the case of Egypt », Accounting, Auditing and

Performance Evaluation, vol 3, n°1, pp 41-67

Haniffa R et Cooke T, (2002), « Culture, corporate governance and disclosure in Malaysian

corporations », Abacus, vol 38, n°3, pp 317-347

Ho S S M et Wong K S, (2001), « A study of the relationship between corporate governance

structures and the extent of voluntary disclosure », Journal of International Accounting

Auditing and Taxation, vol 10, pp 139-158

Healy P M et Palepu K G, (2001), « Information asymmetry, corporate disclosure, and the

capital markets: A review of the empirical disclosure literature », Journal of Accounting and Economics, vol 31, pp 405–440


26/28

26

Hossfeld C et Klee L, (2003), « Informations financières sur la performance des entreprises

allemandes et françaises », working paper , ESCP-EAP, Paris

IASB Discussion paper , (2005), « Management Commentary »

Jilani F et Ben Hammouda R, (2006), « Le gouvernement d’entreprise dans les paysémergents : Approche comparative », working paper , Congrès de l’ATSG 2006 à Hammamet,

Institut Supérieur de Commerce et de Comptabilité de Bizerte

Kamla R, (2007), « Critically appreciating social accounting and reporting in the Arab middle

east: A postcolonial perspective », Advances In International Accounting, vol 20, pp 105-177

Jurnal Voluntary Disclosure

Documents