Determinants of CSR Disclosure of Tunisian Listed Banks: A Multi Support Analysis

Abstract This aim of this study is to investigate the extent and trends of the voluntary corporate social responsibility disclosure (CSRD), as well as to analyze the determinants of (CSRD) in annual reports and websites of listed banks in an emergent capital market namely Tunisia. We examine the level of CSRD via a manual content analysis, and sentence is used as the unit of the analysis. We use the index of Branco and Rodrigues (2006 & 2008) including 23 items. We focus on the annual reports of 11 Tunisian listed banks (TLB) in during the (2007- 2012) period and on their websites in December 2013. We also use regression analysis to identify the determinants of CSRD in TLB. The results obtained show that Tunisian listed banks disclose CSR information primarily in a narrative form. Human resources are the main focus in the annual reports, whereas in the websites, the most widespread theme is the community involvement. Regarding the determinants, it appears that bank age, financial performance and State shareholding are the factors which influence positively CSR disclosure in the banks ‘annual reports. Furthermore, it was found a positive (negative) relationship between leverage (financial performance) and CSRD in the banks websites. In this regard, the results showed that there were different determinants of CSRD for the two supports. Moreover, bank size, foreign shareholding and the type of auditor are not related to banks CSRD neither in the annual reports neither in the websites. This study enables us to provide the Tunisian banks’ stakeholders (like regulators, investors and managers) with a diagnosis of the determinants of CSRD. We contribute to the scare literature on CSRD by financial institutions. It is the first study to investigate banks’ CSRD in Tunisia. It represents also a first attempt that demonstrates how banks’ characteristics and banks’ ownership structure influence their CSRD in both annual reports and websites. Keywords: corporate social responsibility disclosure - annual reports - websites - Tunisian listed banks

[1]

Introduction Because corporate governance is essentially about decision making, it is inevitable that corporate social responsibility (CSR) norms and structures play a role. CSR is a concept that emerged as soon as the fifties; however, during the recent decades, it is becoming more and more significant. It has become a topical issue and the focus of several countries and international organizations and a prominent topic in the both business and academic press. Mostovicz et al. (2011) show that the four pillars of corporate responsibility are: ethics, leadership, personal responsibility and trust. In this regard, Huang and Watson (2015) summarize the work on CSR as follow: (1) determinants of CSR, (2) the relation between CSR and financial performance, (3) consequences of CSR, and (4) the roles of CSRD and assurance. In addition, the European Union considers 2005 the year of social responsibility. Moreover, in the UK, social responsibility has been devoted an entire department dedicated (Douglas et al., 2004). Given the worsening of the social and environmental problems that the world has dad in the recent decades, the concept has become more prevalent (e.g., Exon Valdez, Erika, the explosion of the AZF factory in Toulouse…). These damages should emphasize the need to establish managerial models based on ethical practices, such as the respect for workers' rights, the environmental protection, the charitable activities... Therefore, besides making profits and paying debts, dividends and taxes, companies should be concerned about their ethical and philanthropic responsibilities. Consequently, they have to manage their social responsibilities on the same level as they do with their economic activities. This may apply to all the business sectors, although there are scientists who believe that not all the sectors have the same priorities regarding social responsibility (e.g., Scholtens, 2009; Crawford and Williams, 2010). Banks play a crucial role in social and economic development of a country. Bank transparency is currently of crucial importance. Actually, the implementation of regulatory tools such as Basel II needs to improve the transparency of banks, but the quality of financial reporting remains very irregular between Tunisian banks (Dhouibi and Mamoghli, 2013). More specifically, the banking sector, which is considered as the backbone of the socioeconomic development (Joseph, 2008), like all the other sectors, is called upon to adopt a socially responsible behavior. In addition, CSR is an ethical dilemma that financial institutions face (Masud and Hossain, 2012). In this regard, Paulet (2011) discuss the interest to solve the antagonism between profit and ethics in the banking sector. She provides an insight for banks to satisfy social pressure on more ethical behavior. In order to defend their [2]

reputation, as citizens banks with ethical behavior, the banks had better share with the stakeholders, their stocks in favor of the society and the environment as well as the impact of their activities, hence the use of CSR reporting (Yeung, 2011; Zappi, 2007), which has been widely dealt with by many studies in the literature but excluded financial institutions from their sample. These studies focused exclusively on companies outside the banking sector due to its specific disclosure requirements and financial characteristics (Khlif et al., 2015; Ben Rhouma and Cormier, 2007; Zéghal and Ahmed, 1990...). In fact, few studies tackled CSR reporting in banks in spite of the fact that the level of awareness of ethical standards is expected to be higher in the case of financial institutions. This paper seeks to fill this gap by providing an insight into CSR practices and determinants of the banking sector. Importantly, this study also focuses on Tunisia and contributes to the growing number of empirical disclosure research in the banking sector in the North African emergent markets (e.g Kribat et al., 2013). There is a great focus on these markets given their growing influence on the global economy (Peters et al., 2011). At the time of writing, three published papers have examined the determinants of CSRD of Tunisian companies (e.i., Rabah Gana and Dakhlaoui, 2011; Khemir and Baccouche, 2010 and Driss and Jarboui, 2014). Only the work of Rabah Gana and Dakhlaoui (2011) does not exclude the companies from financial sector, and show that the membership of the company to this sector affects its societal disclosure. No attempt has been made to examine the determinants of CSRD for banks in Tunisian setting although the calls for an extensive focus on CSRD in African countries of many international organizations (e.g Basel Committee, the International Monetary Fund; the World Bank) and of previous work (e.g Mohadeo et al., 2011). Furthermore, on the one hand, most of these studies are descriptive and/or crosssectional (Sharma, 2011; Tsang, 1998; Tarna, 1999; Clarke and Gibson-Sweet, 1999; Douglas et al., 2004; Upadhyay-Dhungel and Dhungel, 2012; Yeshmin, 2012; Masud and Hossain, 2012; Lipunga, 2013). On the other hand, there are only a few studies that dealt with the determinants of banks’ CSR reporting in financial institutions were conducted in European, African Sub-Saharan and Asian countries, namely, Ireland (Douglas et al., 2004), Portugal (Branco and Rodrigues, 2006 & 2008), Bangladesh (Khan, 2010; Yeshmin, 2012), Malaysia (Abdul Hamid, 2004; Htay et al., 2012 & 2013), India (Hossain, 2008; Sethi, 2013; Hossain and Reaz, 2007), Malawi (Lipunga, 2013), Kenya (Barako and Brown, 2008) and Nigeria (Akano et al., 2013; Joseph, 2008). These studies used the annual reports as the only communication medium of CSR information. Moreover, their findings are mixed. [3]

Examining the relationship between CSRD and their determinants is crucial for all the stakeholders. A long-standing theoretical literature based on the legitimacy and the stakeholders’ theories (Brown and Deegan, 1998; Clarke and Gibson-Sweet, 1999; Deegan, 2002; Guthrie and Parker, 1989; Patten, 1991 & 1992; Roberts, 1992; Wilmshurst and Frost, 2000) and the political-contractual theory (Hill and Jones, 1992; Ness and Mirza, 1991) has been developed in order to understand this relationship. Branco and Rodrigues (2008) have called for more examination of the association between CSRD and its determinants in an environment when lower level of disclosure exists. Following this recommendation, our study seeks to investigate the determinants of the voluntary disclosure of CSR information in the annual reports and websites of the Tunisian listed banks1. In fact, most of the empirical studies analysing CSRD have focused on the annual report, which is considered as the most important media of corporate communication. However, internet has become an important medium through which companies can disclose information of different natures and especially corporate social information via their websites (Douglas et al., 2004; Williams and Ho Wern Pei, 1999; Branco and Rodrigues, 2006). Therefore this study constitutes the first attempt to examine the relationship between CSRD and a number of bank characteristics (size, age and financial performance) and two form of ownership structure including State ownership and foreign ownership. To achieve this target and in order to explain the variations between banks CSR reporting, we selected a sample of 11 Tunisian commercial banks listed on the stock exchange of Tunis for the period spanning from 2007 to 2012. We conducted a content analysis to estimate CSRD and we used Branco and Rodrigues (2006 & 2008) index. Our choice is taken due the fact that no previous research has tackled CSR reporting for banks in the Tunisian context, which makes these practices and their determinants unknown. Tunisia is an Islamic North African, civil law country that is characterized by high ownership concentration, State ownership, lack of firms’ accountability, low protection of minority shareholders and weak legal enforcement concerning the corporate disclosure (Chakroun, 2012). These characteristics give this research some originality, since the findings may be applicable in other emergent countries sharing the same legislative and economic characteristics with Tunisia. In addition, even in the Arab and Maghreb countries, the studies interested in CSRD for the financial institutions, in general, and banks, in particular, are rare or almost absent. For this reason, is worth answering the following questions:

1

TLB thereafter

[4]

-

To what extent do listed Tunisian banks use annual reports and websites in their voluntary disclosure of CSR information?

-

What are the main determinants of the CSRD in both medium (the annual reports and the websites) of listed Tunisian banks?

Tests conducted show that TLB disclose CSR information primarily in a narrative form. Human resources are the main focus in the annual reports, whereas in the websites, the most widespread theme is the community involvement. Also, our findings provide evidence that CSRD in annual reports is positively and significantly associated with bank age, financial performance and State ownership. Leverage affects positively CSRD websites. In contrast, financial performance affects negatively CSRD in websites. This study’s contributions can be summarized as follows. First, our study adds to the disclosure literature by focusing in an emergent market characterized by poor information environment, high ownership concentration (especially governmental) and high need of foreign funds. It also explores the associations between banks CSRD in two supports, namely annual reports and websites and banks characteristics and their ownership structure. Second, our results provide evidence that bank age, financial performance and State ownership plays an important role in improving the level of voluntary CSRD in banks annual reports. Consequently, investors cans use such information in excess of mandatory information to distinguish between banks when they make their investment decision. Finally, we focus on the bank sector that is almost usually excluded from the previous studies. This paper is structured as follows: the first section contains the theoretical framework and research hypotheses. The second section is devoted to the adopted methodology so as to empirically test the suggested hypotheses. The results of statistical tests and their interpretations are the subject of the third section. 1. Banks CSR reporting in Tunisia: an overview 1.1 Background to the study Transparency and disclosure are essential for sound and effective corporate governance. Kribat et al. (2013) list several reasons for the importance of accounting disclosures in the banking industry. First, accounting reports are often the sole source of information for bank stakeholders. Second, earnings numbers alone are not adequate for assessing the valuation of banks. Similarly, profitability alone does not give investors a full picture regarding bank performance. Third, the financial statements are less informative for banks than they are for [5]

industrial firms, because the useful information lies in the details and breakdowns of the financial items. The quality of banks’ financial reporting is central to the efficacy of market discipline and to the stability of the financial system. In Tunisia, we are seeing an increase in the number of banks’ annual reports that take into account environmental and social issues. Crawford and Williams (2010) assert that country contexts have different pressures on CSRD in banking sector. Tunisia is one of the emergent countries in which the banking sector is the core of the financial system. In fact, Tunisian business funding is mainly through bank loans. In this context, banks play a critical supporting role as economic operators. In fact, more than 95% of the lending to the economy passes through these financial institutions. Funding includes business development, renewal of the production facilities or innovation, supporting businesses internationally as well as the operating cycle. According to Dhouibi and Mamoghli (2013) Tunisian banks face several problems: a high level of non-performing loans and insufficient capitalization and profitability. The Tunisian banking sector consists of: -

11 commercial banks

-

11 listed banks

-

21 domestic banks

-

2 investment banks

-

2 mixed- development banks

-

8 offshore banks

-

3 Islamic banks

Currently, the Tunisian banking industry is concentrated; in fact 4 banks monopolize 65% of the market. It is dominated by 3 large public banks (namely BNA, STB and BH), 6 private and foreign banks2 and 3 major domestic private banks; which holding respectively 37%, 28% and 28% of banking sector assets. Also, it is controlled by the Government (the central bank) and dominated by commercial banks (11 banks). Besides, it is depending on the real estate sector and it represents a significant funding of the tourism industry. Banking firms are typically faced with more large disclosure requirements than other firms. CSR is a very important issue for the banks. It is a different way of being a bank. It should not be seen as a new or a specific area of activity for the bank. In other words, it should not be considered as an ad hoc activity added to the traditional functions of the financial institutions (Zappi, 2007). We mention that CSRD refers to the disclosure of information about companies’ interactions with society. This type of information seeks to reflect several social 2

4 of these institutions are major foreign banks (from France, Jordan and Morocco).

[6]

and environmental aspects upon which companies’ activities have an impact: human resources issues, community involvement, environmental concerns and other ethical issues (Branco and Rodrigues, 2008). Companies in the sectors of banking and finance do disclose relatively little environmental information but the case is not the same with other aspects of social responsibility, such as employee related information and other information of ethical nature (e.g., community involvement and consumer relations information) (Branco and Rodrigues, 2006). On the one hand, unlike the other sectors (such as chemical, paper…), the banking sector is seen to be environmentally friendly because it has a weak impact on the environment (Branco and Rodrigues, 2008). However, according to Jeucken and Bouma (1999), banks have both a direct and indirect impact on the environment. Concerning the direct impact, the banking sector is regarded as a strong consumer of resources, such as paper, energy ... besides, it produces a lot of waste. As a result, banks need to preserve natural resources and energy by resorting to recycling because these activities are part of their social responsibility (Coulson and Monks, 1999). Besides, regarding the indirect impact, banks indirectly affect the environment through their funding and investment policies. They must therefore take into account the social and environmental risk in their financial analysis, in other words, they must not accept to fund companies engaged in environment damaging activities, such as, highlyindustrial waste creating companies. Consequently, they should orient their lending policies to environmentally favourable projects and green or ethical investment products.... On the other hand, banks should tend to reflect several social aspects in their reporting upon which banks’ activities and products have an impact, such as: employee-related issues, community involvement and other ethical issues….In fact, it seems that social disclosure is popular in highly visible industries such as banks. For instance Tsang (1998) focused on CSRD practices of banking, food and beverages and hotel industries in Singapore. He found that the banking industry had the highest proportion of companies disclosing social information. But they disclosed, in terms of quantity, less information than companies in the other industries. In addition, Zéghal and Ahmed (1990) studied three supports of CSRD by banks: annual reports, advertisements and brochures. They found out that human resources was the most important disclosure category for the annual reports, while for advertisements it was products, and for the brochures it was community involvement. In line with the (GRI, 2006) guidelines, banks are asked to ensure that steps have been undertaken to control environmental pollution before financing enterprises. However, enforcements of these have [7]

been very weak in Tunisia. Consequentially, environmental protection is not in the priorities of the banks. 1.2. Tunisian CSRD environment in the banking sector Since the focus of this study is on CSRD of the TLB, it is appropriate to provide a brief overview of the regulatory framework in the Tunisian context. In this regard, corporate disclosure is poorly regulated in Tunisia in spite of the existence of laws that regulate corporate disclosure. To date, in the Tunisian context, IFRS are not yet applied and CSR reporting is voluntary. In this sense, Tunisian standards setters, aware about the information’s importance that goes beyond the financial dimension, encourage firms in the Tunisian accounting conceptual framework to disclose social and environmental information in their annual reports. But, they don’t give any precision about the form and the content of CSRD. Besides, the general framework for corporate financial reporting in Tunisia is composed by: -

Articles 18 and 83 of the Tunisian accounting conceptual framework.

-

Article 201 of the Tunisian Companies Legal Code.

-

(New) Article 44 of the financial market regulation on public offering (Minister of Finance Decree of 17 September 2008).

-

(New) Article 3 of Act No. 94-117 of November 14, 1994.

In addition, we noted a change in the Tunisian corporate information environment. In fact, the Tunisian government has made great effort to motivate the firms to improve their disclosures in recent years and the Tunisian financial market moves toward a strengthening of transparency and disclosure. This was reflected by the Law No. 2005-96, dated 18/10/2005, concerning the strengthening of financial security which put a greater obligation on publicly traded companies to enhance their accountability, transparency and disclosure quality. Furthermore, corporate governance rules encourage the improvement of transparency and disclosure. In fact, the Arab Institute of Business Leaders that is the main driving force with respect to corporate governance reforms in Tunisia. In 2009, this institute publishes a Guide of the Annual Report of the Tunisian Companies. This Guide included guidelines on corporate financial reporting and disclosure. For companies, adoption of the guide is voluntary and its effects are yet to be seen clearly. The guide guidelines encourage Tunisian companies to improve their corporate social and environmental reporting, but inadequate information is disclosed in the annual reports of most Tunisian companies. In a recent article by Chakroun (2012) in the Tunisian context, it was evidenced that voluntary disclosure in Tunisia is problematic, as she assert he overall degree of voluntary disclosure amounts to [8]

43.14%, indicating a mismatch between supply and demand for voluntary disclosure and especially for social and environmental information. A probable explanation for the low level of voluntary disclosure may be related to the poor legal protection of investors and the poor enforcement process applied by the Financial Market Council as argued by Chakroun (2012). Recently, banking sector was subjected to significant regulatory changes. This sector is often excluded from sample companies in the previous studies on corporate disclosure due to the sector’s more stringent regulatory regime such as Basel agreements. To overcome the problems assigned to the banking sector, Tunisian banks implement the guidelines and directives of the first and second Basel Accord (Dhouibi and Mamoghli, 2013). In this regard, Basel II agreement is based on three pillars. The first defines the minimum required capital for bank to cover the three major risks they are exposed to. The second establishes the principle of an individual prudential supervision. Finally, the third focuses on the concepts of transparency and market discipline because disclosure and bank transparency are currently of crucial importance. More specifically, pillar three aims to establish rules on financial transparency by improving the way of disclosing information to the public about the assets, the risks and how to manage them. The underlying objective is to standardize the banking practices in terms of financial communication and therefore facilitate the dissemination of the banks’ accounting and financial information from one country to another. This disclosure rule tends to raise the level of transparency for the banking industry. In contrast, the status of CSRD by banks is not satisfactory in emergent countries like Tunisia, largely due to poor legal enforcement and inadequate pressure from civil society. In light of the recent changes in the disclosure rules for banking industry in Tunisia, it is expected that TLB will undertake necessary steps in light of the implementation of certain decisions with regard to CSRD. Consequently, they will have more incentives to increase their CSRD level, although the needs in terms of formulation of new policies and implementations of existing regulations. CSRD could be mandatory or voluntary, thus Paulet (2011) shows that regulation is a necessary but not sufficient condition to ensure the efficiency of banking institutions. In this regard, we examine in this research the effect of several banks characteristics and ownership structure on the level of CSRD. 2. Theoretical framework and hypothesis development A voluminous body of theoretical research has been developed to examine the determinants of voluntary disclosure in emergent markets, but few studies examined the determinants of CSRD for banks. [9]

2.1. Theoretical framework Alhtaybat et al. (2012) assert that we cannot base our explanation of corporate disclosure on one theory. In fact these authors propose and discuss a map of numerous corporate disclosure theories. To explain the level of CSRD in the annual reports and websites of the TLB, it is worth applying the most frequently used approaches in this context, such as, the economic and the socio-organizational approaches (Cormier and Magnan, 1999). The legitimacy and shareholder theories are a part of the socio-organizational approach whereas the economic approach includes the politico-contractual theory, which can bring together both the agency and the regulation theories. 2.1.1. The legitimacy theory The legitimacy theory is often associated with the concept of social contract, according to which the society and the firms are bound by a contract. That is why firms disclose their operations associated with CSR to inform the society about the degree of fulfilment of the social contract (Branco and Rodrigues; 2008). In this vein, banks can legitimize its existence and earn social approval by disclosing their CSR activities (Cormier and Gordon, 2001 and Branco and Rodrigues, 2006). Hogner (1982) was the first researcher to use the legitimacy theory in the study of social reporting. According to this researcher, societal information disclosed by companies is a response to what the society expects from business behavior. According to the legitimacy theory, CSR reporting is a means used by the company to ensure its sustainability and justify its existence when there is a discrepancy between its values and the ones usually accepted by the society (Wilmshurst and Frost, 2000; Patten and Crampton, 2004; Brown and Deegan, 1998; Clarke and Gibson-Sweet, 1999). In this regard, Branco and Rodrigues (2008, p 164) assert that: “CSR disclosure is seen as one of the strategies used by companies to seek acceptance and approval of their activities from society. It is seen as an important tool in corporate legitimation strategies, as it may be used to establish or maintain the legitimacy of the company (that is, influence public opinion and public policy)”. Therefore, the company makes disclosure of CSR information for the purpose of making its actions legitimate and showing that it is a corporate citizen that adopts a socially responsible behavior. For Suchman (1995), a company shall disclose CSR information to give the impression that its actions are desirable and appropriate compared to the social system built essentially on social standards and values. According to Deegan (2002), CSR reporting is a means used by the leaders to affect their companies’ external perceptions. Other researchers suggest that when social [10]

information is disclosed, the company seeks to reduce public pressure (Brown and Deegan, 1998; Abdul Hamid, 2004; Akano et al., 2013; Khemir and Baccouche, 2010; Guthrie and Parker, 1989; Patten, 1991 & 1992; Deegan and Gordon, 1996…). Furthermore, Mohadeo et al. (2011) confirm that the legitimacy, as a strategic and managerially driven approach favouring symbolic actions, is the prevailing motivation underlying the progression of CSRD in Mauritius. Consequently, we assert that legitimacy theory may be an explanation of CSRD by Tunisian banks. 2.1.2. The stakeholder theory The implicit assumption of the stakeholder theory is that a firm can continue its long-term operations only if it receives support of all related parties. Stakeholders express specific expectations with regard to CSR actions (Barako and Brown 2008; Ullmann, 1985; Roberts, 1992). The stakeholder theory, which is part of the socio-organizational approach, was developed by Freeman (1984). According to him, stakeholders are considered as: “any group of individuals that can affect or is affected by the achievement of business objectives". In accordance with the stakeholder theory, CSRD is considered as a basis for dialogue with the company's various partners (Ullmann, 1985). For Freeman (1984), the dissemination of CSR information is regarded as a means which helps the company manage its relationships with its partners. Therefore, the company shall disclose CSR information so that it can cope with the pressure from its partners and affect their perceptions and actions. In addition, CSRD provides a way of communicating with stakeholders, and of convincing them that the bank is fulfilling their expectations (Branco and Rodrigues, 2008). This theory was frequently used in previous research on the determinants of CSR reporting (Roberts, 1992; Gray et al., 1995; Khemir and Baccouche, 2010 ...). 2.1.3. The politico-contractual theory The politico-contractual theory was founded by Watts and Zimmerman (1978). It has an explanatory power of CSR reporting since it has the advantages of bringing together the regulation theory and the generalized agency one. This theory also enables to make a synthesis of the stakeholder theory and that of legitimacy through the combination of regulation theory and the generalized agency one. The generalized agency theory is also called the extended agency theory. It is a theory on which the politico-contractual theory is based. It helps explain the role of CSR information in the management of contracts between the stakeholders of the company. According to Hill and [11]

Jones (1992), any entity that has a legitimate claim on the firm is considered as a business stakeholder. The politico-contractual theory has not been widely used in previous studies dealing with the determinants of CSR reporting (Belkaoui and Kaprick, 1989; Ness and Mirza, 1991; Hackston and Milne, 1996; Cowen et al., 1987; Khemir and Baccouche, 2010). Actually, as far as we know, no previous study that tackled banks’ CSR reporting referred to this theory. 2.2. Hypotheses development Several empirical studies have been undertaken to investigate the determinants of CSRD in the developed countries. But this topic has recently attracted a great deal of attention in some African emergent markets characterized by legal environments with weak governance rules and disclosure levels (Lipunga, 2013 in Malawi; Barako and Brown, 2008 in Kenya; Akano et al., 2013 and Joseph, 2008 in Nigeria). However, there is still a lack of substantial empirical evidence about CSRD in African emergent markets for the banking sector. In our study, we will test six hypotheses to explain the level of CSR reporting in the annual reports and websites of the TLB. There are hypotheses related to the bank’s characteristics and others to ownership structure. 2.2.1. Hypotheses related to the bank's characteristics We test the effect of size, age, financial performance and indebtedness on CSRD for Tunisian banks. Adams (2002) argues that corporate characteristics are influential factors in determining the extent of corporate social and ethical reporting. Most of the previous studies showed that the size is a key factor in explaining the level of CSRD. In this regard, several arguments have been advanced. For instance, Branco and Rodrigues (2008) focus on the standpoint of the public visibility as a major determinant of the social reporting of Portuguese banks. Actually, there are researchers who believe that largesized companies should disclose more information than small-sized ones as they are more visible and therefore more exposed to pressure and political costs (Watts and Zimmerman, 1978). Moreover, they have more players who are concerned with its social and environmental actions (Damak Ayadi, 2004). In addition, larger companies have more resources than smaller ones which enable them to bear the expenses of additional disclosure (Chavent et al., 2006). Therefore, larger companies have a big number of skilled and knowledgeable staff, the thing which helps them set up a reporting system as developed as the small businesses’ one. Another thing which encourages large companies to disclose more CSR information is the fact that they are always looking to improve their reputation with their [12]

various partners (Branco and Rodrigues, 2008). In fact, these authors argue that banks with a higher public visibility exhibit greater concern to improve the corporate image through social responsibility disclosure. In addition, Henault and Lemoine, 2008 state that companies tend to disclose a high level of CSR information in order to become attractive to the top workers level. On reviewing the literature, it appears that previous studies did not reach the same results. In fact, there are researchers who found a positive relationship between the level of disclosure of CSR information and the company’s size (Akano et al., 2013; Belkaoui and Kaprick, 1989; Cowen et al., 1987; Hackston and Milne, 1996; Abdul Hamid, 2004; Haniffa and Cooke, 2005; Hossain et al., 1995; Neu et al., 1998: Patten, 1991..). On the other hand, some other researchers found a negative relationship (Stanwick and Stanwick, 1998), and some others found no relationship between both variables (Cowen et al., 1987...). Therefore, we expect that larger banks should disclose more information about their socially responsible operations and by this they can be more recognizable in the society. A positive relationship between the level of CSRD and the bank size is expected. Hence, our first hypothesis is as follows: H1: Size has a positive impact on the level of the CSRD in the annual reports and websites of the TLB. Age is one of the banks characteristics that have a significant influence on its CSR reporting level. It is a proxy of the bank stage of development and growth. However, arguments about the meaning of the relationship between both variables in previous research are contradictory. In fact, Baccouche et al. (2010) pointed out that older firms disclose more CSR information since they have a longer experience which helps them identify the resources needed for their survival and safeguarding of their reputation through their social actions. On the other hand, Cabagnols and Le Bas (2006) state that younger firms are those that need to be more engaged in societal activities because CSR is a contemporary phenomenon. Despite its importance, the "age" variable has not been frequently studied. Several studies found that this variable has a positive impact on the level of CSRD (Abdul Hamid, 2004). In this regard, a one-way relationship between indeterminate age and CSR reporting is expected. Given the mixed results of the earlier studies, the second hypothesis is as follows: H2: Age has an impact on the level of the CSRD in the annual reports and websites of the TLB. Financial performance was often studied in the literature on the determinants of CSRD. However, previous research that dealt with the relationship between performance and CSRD [13]

was contrasting. For some researchers, better performing firms have greater incentives to disclose more CSR voluntary information. Actually, they argue the existence of a positive relationship between both variables (Hossain, 2008; Abbott and Monsen, 1979; Roberts, 1992; Stanwick and Stanwick, 1998...). Whereas other studies found a negative relationship (Damak Ayadi, 2004; Fry and Hock, 1976). Furthermore, there are others who concluded that there is no relationship between both variables (Roberts, 1992; Hackston and Milne, 1996; Cowen et al., 1987; Belkaoui and Kaprick, 1989…). Therefore, an undefined direction relationship between financial performance and the level of CSRD is expected. Hence, our third hypothesis is follows: H3: Financial performance has an impact on the level of the CSRD in the annual reports and websites of the TLB. The indebtedness level is one of the banks characteristics that have an impact on the level of CSRD. Some researchers showed that, the more the company is in debt, the less it discloses CSR information (Belkaoui and Kaprick, 1989; Cormier and Magnan, 1999; Oxibar, 2005). For Oxibar (2005), indebted companies prefer accounting policies which favor the increase of the results, or that spending on social activities will lead to a decline of earnings and therefore over-indebted companies will not be involved in this kind of activity and, even if they do, they will not disclose such information. However, there are others who state the opposite. In fact, Roberts (1992) considers that the company in debt is more involved in CSR activities to meet the expectations of its creditors concerning the social aspect. Jensen and Meckling (1976) consider that the most indebted companies disclose more voluntary information in order to reduce the agency costs and therefore their capital costs. Due to this contrast in the justification of the relationship direction between debt and CSR reporting, an unknown direction association is expected between the two variables. Hence, our fourth hypothesis is as follows: H4: The indebtedness level has an impact on the degree of CSRD in the annual reports and websites of the TLB. 2.2.2. Hypotheses related to the ownership structure Ownership structure has been one of the most frequently used variables in the context of CSRD. Given the fact that Tunisian setting is characterized by two main features (State ownership and foreign ownership), we test the effect of such variables on CSRD. The impact

[14]

of internal corporate governance mechanisms may be complementary or substitutive (Ho and Wong, 2001). The State continues to play an important role as principal owner in some of the most important banks in Tunisia. When the State is a shareholder in a company, its target is beyond the profit but seeks to play a social role which is its priority. Therefore, the firm in which the State is a shareholder will disclose more CSR information (Baccouche et al., 2010). According to the politico-contractual theory through the regulation theory, companies where the State is a shareholder will further disclose social information because these companies’ leaders will voluntary disclose this type of information to avoid the regulator’s intervention (Baccouche et al., 2010). The empirical results regarding the causality of the relationship between CSR reporting and the State’s ownership do not always coincide. In fact, some studies led to a positive association between both variables (Eng and Mak, 2003), while others found a negative association (Al Janadi et al., 2013), and a third group found no relationship (Naser et al., 2006). We expect a convergence between the pure profit goals of banks characterized by a strong State ownership and goals related to public interests .Thus, a positive relationship is expected between the State’s shareholding and the level of CSRD. Hence, our fifth hypothesis: H5: The State’s shareholding has a positive impact on the level of the CSRD in the annual reports and websites of the TLB. In general, foreign shareholders face a higher level of information asymmetry. In order to compete effectively in the capital market, banks with a strong foreign ownership would disclose more CSR information. In fact, CSRD is a mean to monitor the actions of management by foreign shareholders. Indeed, in the basis of the previous studies, it appears that the level of CSRD rises with foreign shareholding for several reasons. Actually, foreigners attach great importance to the CSR concept; therefore, they can assess a company according to level of its participation in social activities. As a consequence, to attract foreign shareholders, the company will disclose more societal information. Several researchers showed that foreign shareholders prefer to invest more in companies that disclose more CSR information (Haniffa and Cooke, 2002 & 2005). Furthermore, Adams and Mehran (2005), state that "foreign shareholding induces the leaders to disclose more societal information". In fact, in doing so, the leaders can avoid the shareholders’ control mainly when ownership and management are separated. Therefore, the empirical results of the previous research are varied. Moreover, in some studies, researchers found a positive relationship between the level [15]

of the societal information disclosure and foreign ownership (Haniffa and Cooke, 2002 & 2005; Barako et al., 2006), while in others, this relationship does not exist (Baccouche et al., 2010). In this regard, a positive association is expected between the level of the societal information disclosure and foreign shareholding. Hence, our six hypothesis is as follows: H6: Foreign shareholding has a positive impact on the level of the CSRD in the annual reports and websites of the TLB. 3. The methodology 3.1. The sample selection and data Our sample consists of all listed banks on the Tunis Stock Exchange (see Appendix 1). Our sample consists of eleven banks, three state-owned banks and eight private banks four which are foreign. All these banks have websites. The choice of studying banks is justified, first, by the fact that there is, to our knowledge, no previous study in Tunisia that dealt with banks’ CSRD and, secondly, because the studies which dealt with CSR reporting have always excluded the financial sector to their sample due to its specificity (Ben Rhouma and Cormier, 2007). The annual reports were chosen as the first CSRD support for several reasons. The first is that it has a degree of credibility that does not exist in other media (Neu et al., 1998; Baccouche et al., 2010 ...). Second, it is characterized by a wide disclosure (Ernst and Ernst, 1978). Third, it is distinguished by the regularity of its production (Oxibar, 2005). Finally, it is seen as the most reliable and available document in Tunisia (Baccouche et al., 2010). Parker (1982), states that despite all its advantages, the annual report is characterized by some rigidity of its format and rules governing the preparation and dissemination, which causes some social inaccessibility as there are some people who find a difficulty in decoding its messages. On the other hand, Zéghal and Ahmed (1990) point out to the deficiencies that may result after studying CSRD based only on the annual reports. For this reason, beside the annual reports, we chose to analyze the banks websites which differ from them by their periodicity, quality and scope of their audience (Zéghal and Ahmed, 1990). The choice of making a multi-support analysis is justified by the objective to have a broader and comprehensive view of CSRD (Oxibar, 2005). Moreover, Gray et al. (1995) assert that source diversification is required because: "the ideal thing is to try to take into account all the company's communication to study social disclosure practices". We should mention that all the websites are in French language, while the annual reports are written in Arabic, English or French. [16]

The annual reports were analyzed over a six-year period spanning from 2007 to 2012. To collect the annual reports, we download them from the websites of banks, the website of Financial Market Council (CMF)3, the website of Tunis Stock Exchange (BVMT)4 and Tustex’ website5. For annuals reports that are not available on the Internet, we collect them directly from the headquarters of banks. As part of this research, we preferred to analyze the CSRD through websites over one month (December 2013). The data regarding banks characteristics and ownership structure is hand‐collected and obtained through annual reports and the website of BVMT. 3.2. Variable measurement 3.2.1. Measurement of the dependent variable The variables to be explained in our study are: the level of voluntary CSRD in the annual reports and websites such as (Branco and Rodrigues, 2008). To measure these variables, we proceed with the manual content analysis technique of the annual reports and websites of the TLB. The content analysis method is defined by Abbott and Monsen (1979) as: "a technique of collecting information which consists in codifying literal information into categories and identify quantitative scales". This method was chosen because, firstly, it can make our research consistent with the previous studies and facilitates the comparison and, secondly, because it is the most often used in the previous studies dealing with CSRD (Khan, 2010; Lipunga, 2013; Gray et al., 1995; Williams and Ho Wern Pei, 1999; Branco and Rodrigues, 2008; Hossain and Reaz, 2007; Damak Ayadi, 2006; Khemir and Baccouche, 2010; Al Janadi et al., 2013; Driss and Jarboui, 2014...). There are several classes of information categories that have been used in the previous literature on CSRD. For instance, Zéghal and Ahmed (1990) used the following categories: the environment, the energy, the human resources, the products, the involvement in the community, the ethics and others. For our study, we adopt the grid computing CSRD used by (Branco and Rodrigues, 2006 & 2008). This grid is composed by 23 items and divided into four information categories representing (see Appendix 2): the environment, the human resources, the products and customers and, the community involvement. This choice is justified by the fact that these information categories are the most used in previous studies dealing with the CSRD (Williams and Ho Wern Pei, 1999; Lipunga, 2013; Akano et al., 2013;

3

www.cmf.org.tn www.bvmt.com.tn 5 www.tustex.com 4

[17]

Tsang, 1998; Abdul Hamid, 2004; Douglas et al., 2004...). Moreover, we chose these information categories to facilitate the comparison between the practices of the Tunisian banks’ CSR reporting with those of their counterparts in other countries. Furthermore, the other reason behind choosing this list is that, after a first reading of the annual reports and websites, it appeared that this list is the most suitable for our study as it does not even need to be adapted to the Tunisian context. Previous studies on the CSRD used several measurement units. For example, (Zéghal and Ahmed, 1990; Douglas et al., 2004; Deegan and Gordon, 1996) retained the word as a unit of measurement; whereas Gray et al. (1995) used the pages; while others like (Williams and Ho Wern Pei, 1999; Hackston and Milne, 1996; Khemir and Baccouche, 2010...) chose the sentences. As part of our study, we choose the sentence as a unit of measure as it provides coherent ideas that make sense (Hackston and Milne, 1996). Moreover, the sentence helps count more accurately than words. In addition, the sentence can overcome the problem of the margin size and the graphic inclusion. Finally, the choice of this measurement unit is made because most researchers consider it as the most appropriate instrument to measure the level of CSRD (Unerman, 2000). However, according to Ernst and Ernst (1978): “monetary or non-monetary quantification of societal information improve its quality, however, information quality would be enhanced by an explanation in a narrative form”. For this reason, in our study, we opted for the classification of the CSR information disclosed in the annual reports and websites as: narrative, monetary quantitative and non-monetary quantitative forms. Unerman (2000) argues that annuals reports are not the only medium through which companies can report their CSR behavior and other media like the websites enable more timely communication to the stakeholders. Thus we add the content analysis of banks websites to our study. Besides, the researcher faces several methodological difficulties in analyzing the website content. The first difficulty is spatial, which means that the websites contain links to pages and external sites. Therefore, the ideal solution is to fix the space frontiers of the site. The second difficulty is a time characteristic since the sites are always updated. To deal with these challenges, we will adopt the solutions announced by Oxibar (2005). Regarding the space delineation, we will try to follow the plan of the site by taking into account the information available from the site map. Then, to cope with the temporal frontiers, we should print the whole site while taking into account the space boundaries, which makes the comparison with the annual reports very easy. [18]

3.2.2. Measurement of the independent variables The table below summarizes of the independent variables, their measurement instruments and the expected signs. TABLE 1 ABOUT HERE

3.2.3. The empirical models To identify the determinants of the level of CSRD in the annual reports and in the websites of the TLB, both of the following models should be used. A counting model analysis is performed (Poisson regression or the negative binomial regression) for a sample of 66 observations spanning from 2007 to 2012 (Model 1) and a sample of 11 observations in December 2013 (Model 2). In our analysis we include six independent variables and one control variable that have been found to the literature to be correlated with CSRD. Model 1: The first model (panel-data model) CSRDIS_AR

it

= β0 + β1 BKSIZE it + β2 BKAGE it + β3 ROE it + β4 LEV it + β5 STATE it

+ β6 FORGN it + β7 AUDIT it + ɛ Model 2: The second model (cross-section model) CSRDIS_WS i = β0 + β1 BKSIZE i + β2 BKAGE i + β3 ROE i + β4 LEV i + β5 STATE

i

+ β6 FORGN i + β7 AUDIT i + ɛ Where; CSRDIS_AR: is the dependent variable of the first model, which represents the level of the CSRD in the annual reports of the TLB measured by the number of sentences. CSRDIS_WS: is the dependent variable of the second model, which represents the level of the CSRD in the websites of the TLB measured by the number of sentences. BKSIZE: is the size variable measured by the number of bank branches. BKAGE: is the age variable measured by the number of the operating years since the setting up of the business. ROE: is the financial performance variable. LEV: is the debt variable measured by the proxy of overall debt ratio to total assets. STATE: is the State shareholding variable measured by the percentage of shares held by the State. FORGN: is the foreign ownership variable measured by the percentage of shares held by foreigners. [19]

AUDIT: is the control variable, which represents the type of auditor. 4. Empirical results and discussion As the present study is novel in investigating CSRD practices and determinants specifically in the banking sector of an emergent economy, direct comparison with equivalent earlier studies is difficult. 4.1. Descriptive statistics 4.1.1. Results of the content analysis The content analysis helped us to identify the overall level of the CSRD of the TLB in their annual reports during the (2007-2012) period, as well as in their websites in December 2013. Moreover, through this content analysis, we were able to know the most frequently disclosed items either through the websites or through the annual reports. The level of the CSRD in the annual reports in 2012 was 400 sentences whereas in their websites, it was 214 sentences allover December 2013. These findings suggest that Tunisian banks disclose more CSR information on their annual reports than in their websites. On the contrary, Williams and Ho Wern Pei (1999) assert that organizations in Australia, Singapore, Malaysia and Hong Kong appear to provide more narrative information on websites than annual reports. The following table summarizes the overall level of the CSRD in the annual reports of the TLB during the (2007-2012) period. The comparison of the level of the CSR information disclosed in the banks’ annual reports helps us note that there are differences across the six-year study. Actually, a drop by 10 sentences was registered in 2008, which brought down this level from 202 to 192 sentences. In 2009, there was a rise by 135 sentences, which brought up the level of CSRD in the annual reports from 192 to 327 sentences. This is the most important increase recorded during the six-year study. In 2012, the level of CSRD peaked as it reached 400 sentences. A possible explanation for some changes in CSRD practices by some banks is the legitimization of their activities. This result is in accordance with the finding of the longitudinal study of Mohadeo et al. (2011) and provides evidence of the increased importance of CSRD in the African developing economy. TABLE 2 ABOUT HERE

Table 3 displays the level of CSRD in each category of CSR information disclosed in the annual reports and the websites of the TLB. Based on the comparison between the disclosure levels in each category, it can be concluded that the category the most disclosed in the annual reports during the 6-year study is that of the human resources which accounts for 71.4% of the [20]

global disclosed level (that is 1297 sentences). This result seems consistent with that of Driss and Jarboui (2014) for the non-financial firms in the Tunisian context, Mohadeo et al. (2011) for the Mauritius context, Douglas et al. (2004) for the Irish context, Akano et al. (2013) for the Nigerian context and Zéghal and Ahmed (1990) for the Canadian context; but inconsistent with that of Barako and Brown (2008) for the Kenyan context, where there is a complete lack of disclosure on the categories of human resources. The following category is that of products and services of which the disclosure level reached 269 sentences over the 6-year period, which represents 14.75% of the overall CSRD in the annual reports. Whereas Abdul Hamid (2004) for the Portuguese context found out that product/service related disclosure seems to be more frequent than environmental and energy, human resources or community related disclosures. Information category regarding the community involvement represents 12.4% of the total CSR information (that is 221 sentences). As a conclusion, it turned out that the environment is the least disclosed category in the annual reports as it represents only 1.97% (that is 36 sentences). Regarding the websites, it appears that the community involvement was the most widespread category in December 2013. In fact, it represents 58.43% of the total CSRD (that is 125 sentences). This result complies with those of Branco and Rodrigues (2006 & 2008) for Portuguese. The second category is that of the human resources with 25.23% of the overall CSRD (that is 54 sentences). The third category, is about of the products and services which represents 15.88% of the overall CSR disclosure, and in the last place, we find the environment, which is the least disclosed category, with only 0.46% of the total CSR information in the websites of the TLB. This result seems consistent with those of the previous studies (Clarke and Gibson-Sweet, 1999; Lipunga, 2013). TABLE 3 ABOUT HERE

Table 4 reports the level of CSRD depending on its form in the annual reports and websites of the TLB. The analysis of CSRD based on the form shows that it is essentially communicated in a narrative form. Actually, narrative CSR information represents 57.27% of the overall volume disclosed in the annual reports of TLB during the (2007-2012) period (that is 1044 sentences). Whereas, the level of CSR information’ dissemination under a non-monetary quantitative form is relatively large. In fact, it represents 34% of the overall CSRD in the annual reports (that is 620 sentences). However, the TLB rarely use the monetary quantitative form in their CSR reporting as it accounts only for 8.73% of the overall voluntary offer of CSR information in their annual reports during the study period. [21]

Regarding the websites, it appears that the form preferred by the TLB in communicating CSR information is the narrative one, with 200 sentences, which represents 93.45% of the total CSRD in their websites. These banks rarely use the monetary and non-monetary quantitative forms, which represent respectively 5.14% and 1.41% of the overall volume of CSR information. Thus, these banks prefer to disclose CSR information mainly under a narrative form whether in their annual reports or their websites. This result is consistent with those of the previous studies (Yeshmin, 2012; Khemir and Baccouche, 2010; Zéghal and Ahmed, 1990). TABLE 4 ABOUT HERE

4.1.2. Results of the descriptive analysis It can be seen that the TLB disclosed, on average, 27.62 sentences in their annual reports during the (2007-2012) period. The minimum of their disclosure was four sentences recorded by the BH in 2008, whereas the maximum was 86 sentences achieved by ATTIJARI in 2009. Besides, the TLB disclosed an average of 19.45 sentences in their websites in December 2013. The level of CSRD in the websites of the TLB varies between a minimum score of zero and a maximum score of 25 sentences, which was recorded by UBCI in December 2013 (Panel B of table 5). The comparison between the two supports reveals that the TLB use annual reports more than the websites to disclose CSR information. This result is in concordance with that of Branco and Rodrigues (2006 & 2008) for the Portuguese context. The Panel A of table 5 displays that bank size, which is measured by the number of bank branches, varies between a maximum value of 185 branches recorded by ATTIJARI in 2012, and a minimum value of 5 branches recorded by the BTE in 2007. The number of operating years of the TLB, between 2007 and 2012, ranges from a minimum of 19 years, recorded by the BH in 2007, and a maximum of 129 years, recorded by the BT in 2012. The age of the TLB is on average equal to 47.95 years, which means that they are relatively old. On average, the Tunisian banks had 108.59 branches between 2007 and 2012, which means that most of them are large-sized. The average financial performance, which is measured by the ROE, is equal to 9.62%. It actually varies between a minimum value of -11.87% recorded by the UIB bank in 2008 and a maximum value of 29.76% recorded by that of ATTIJARI in 2008. This means that the TLB were performing during the (2007-2012) period. The leverage of the TLB between 2007 and 2012 was on average 89.65%. This level peaked at the UIB in 2007 with 101.09% and reached a trough of 64.55% in 2007 at the BTE. These values show that the banks of our sample are carrying too much debt. On the other hand, the State shareholding [22]

variable varies between a maximum percentage of 66.67% recorded by the BNA in 2008 and a zero minimum percentage. The percentage of shares held by the State is on average about 20.54%. The banks in which the State shareholding exceeded 50% during the (2007-2012) period are: BH, BNA and STB. The variable foreign shareholding is on average equal to 30.11% during the (2007-2012) period. Moreover, the percentage of shares held by foreigners varies between a maximum rate of 64.24%, recorded by the ATB between 2007 and 2012, and a zero minimum rate, recorded by the BNA during the (2007-2012) period. This shows that the participation of foreign shareholders in Tunisia is not too high. From the table 5, we can see that 16 observations out of 66 in which the TLB were audited by small audit firms during the (2007-2012) period. However, there are 50 observations in which banks were audited by at least a large audit firm (big 4). TABLE 5 ABOUT HERE

4.2. Results of the multivariate analyses The examination of the variables correlation matrix (Panel A of table 6) and the VIF results (Panel B of table 6) shows us the absence of a significant correlation between the independent variables. TABLE 6 ABOUT HERE

The dispersion tests in both models showed the inadequacy of the Poisson regression. Therefore, the negative binomial regression is the most appropriate one. The results of the regressions of both models are summarized in the tables 7 and 8 below. In the first model, with the presence of panel data, we could have a problem of heterogeneity. In response, a fixed-effects or random-effects model should be used. The result of Hausman test shows that the fixed-effect model is a better option than the random-effect one. In terms of global significance, our first model is significant since it has a Chi-2 value of 39.07 and significant at a 1% level. As for the adjustment quality, the high Log-Likelihood absolute value is high (185.02) indicates a good adjustment quality. For our second model, it is not too significant since it has a Chi-2 value of (13.16), significant at a 10% level only. In terms of adjustment quality, the low Log-Likelihood absolute value (35 236) indicates an intermediary adjustment quality. This regression quality of the second model may be explained by the limited number of observations (11). The size-related hypothesis has not been validated in both models (Panel C table 7 & Panel B table 8). Actually, a negative but non-significant association was found between the bank size [23]

and the level of CSRD in the banks’ annual reports (-0.0067, z-statistic = -1.28). While, a positive but non-significant association was found between the size and the level of CSRD in the banks websites (0.0049, z-statistic = 0.36). Therefore, it can be concluded that banks’ CSR reporting, either in the annual reports or in the websites, does not depend on the banks’ size. This interpretation has already been provided by Cowen et al. (1987) and Kribat et al. (2013) for Libyan banks. Our finding is surprising; however, investors in smaller banks may face relatively greater information asymmetry and thus require higher level of CSRD to deal with agency problems. A significant positive association is found between bank age and the level of CSRD in the banks’ annual reports (0.1497, z-statistic = 4.59). This means that older banks disclose more societal information in their annual reports than younger ones. This finding seems to be consistent with that of Baccouche et al. (2010) and Abdul Hamid (2004). They suggest that legitimacy theory may be an explanation of CSRD by Tunisian banks. Whereas, there is a positive but non-significant association between the age of the bank and its level of CSRD in the websites (-0.0140, z-statistic = -0.66). This Finding is consistent with earlier evidence for North Africa, in Kribat et al. (2013) on Libya. Consequently, the age-related hypothesis is supported for the first model but rejected in the second. In line with the results of (Kribat et al., 2013; Dhouibi and Mamoghli, 2013; Hossain, 2008; Belkaoui and Kaprick, 1976; Abbott and Monsen, 1979; Roberts, 1992; Stanwick and Stanwick, 1998), there is a significant positive association between financial performance and the level of CSRD in the banks’ annual reports (0.0062; z-statistic = 2.09).This result suggests that TLB performing strongly have incentives to signal their CSR to the market. While, there is a negative and significant association between financial performance and the level of CSRD in the banks websites (-0.0184, z-statistic = -1.75). This result was already found by Damak Ayadi (2006) for the French context and Driss and Jarboui (2014) for the Tunisian one. Hence, our results provide evidence that the relationship between financial performance and CSRD was mixed. Consequently, the performance-related hypothesis is supported for the both models. It turned out that the indebtedness level is not an explanatory factor of CSRD in the banks’ annual reports. This is may be due to the fact that a negative but non-significant association was found between the leverage and the banks’ CSRD in the annual reports (-0.0207, zstatistic = -1.05). This finding seems to contradict those found by Khemir and Baccouche (2010) and Driss and Jarboui (2014). However, it appears that the indebtedness level has a [24]

positive and low significant impact on the CSRD in the banks websites (0.3305, z-statistic = 1.85). For this reason, the hypothesis related to the leverage was refuted in the first model but partially confirmed in the second. Our results showed a positive and significant relationship between the State shareholding and the level of CSRD in the banks’ annual reports (0.0294, z-statistic =1.68). This result is not consistent neither with the results of Al-Janadi et al. (2013), who found a negative association between the State shareholding and the level of CSRD, neither with the findings of Naser et al. (2006), who found no relationship. Whereas our result is in accordance with the result found by Eng and Mak (2003). It implies that the banks where the State is a dominant shareholder disclose more CSR information in their annual reports than other banks do. However, the State shareholding was found to have a positive but non-significant impact on the level of CSRD in the banks websites (0.0023, z-statistic = 0.03). This result shows that banks in which the State is a dominant shareholder do not disclose more CSR information in their websites than other banks do. Consequently, the State ownership-related hypothesis is supported for the first model but rejected in the second. We found a positive but non-significant association between foreign shareholding and the CSRD in the annual reports (0.0127, z-statistic = 0.71) and the websites (0.0292, z-statistic = 1.31). Similarly, a positive and significant relation is found by Dhouibi and Mamoghli (2013) between foreign ownership and voluntary disclosure in Tunisian Bank’s reports. Our result is not consistent with the results obtained by Barako et al. (2006) and Hanniffa and Cooke (2002 & 2005). It shows that the TLB do not use CSR reporting as a means to attract foreign investors. Therefore, the foreign shareholding-related hypothesis was refuted in both models. The type of auditor’ control variable appears to have no effect on the level of banks’ CSRD, whether in the annual reports (-0.2143, z-statistic = -1.16) or websites (3.6239, z-statistic = 0.79), which means that the banks audited by large audit firms do not disclose CSR information more than those audited by small audit firms. This insignificant result seems to confirm that of Baccouche et al. (2010) and Dhouibi and Mamoghli (2013). Overall our findings provide evidence that: age, State ownership and financial performance affect positively CSRD in the annual reports of the TLB. Besides, financial performance (the indebtedness level) affects negatively (positively) CSRD in the websites of the TLB. Moreover, bank size, foreign shareholding and the type of auditor are not related to banks CSRD.

[25]

TABLES 7 & 8 ABOUT HERE

4.3. Results of the robustness test In the counting models, the estimated coefficients are not directly interpreted as it is in the linear regression models but it requires further analysis. This implies the use of the marginal effect method to measure the impacts of a one-unit increase in an independent variable on the level of CSRD. TABLES 9 ABOUT HERE

Conclusion The importance of CSRD seems to be growing all over the world, and the same appears to be the case in Tunisia, which represents an excellent example of low information environment in MENA region. This study has given an idea of how emergent countries, especially Tunisia, perform CSR reporting and how the banking sector in particular, does this. More specifically, it examines the determinants of CSRD of TLB giving the crucial role played by financial institutions in financing activities. It makes three important contributions to corporate disclosure literature. First, it extends prior research on CSRD to the banking industry, which is usually excluded from previous research on CSR reporting. Second, it adds to the limited number of studies that have examined CSRD in emergent markets. Third, it based on two corporate information supports, namely annual reports and websites. In this study, we follow Branco and Rodrigues (2006 & 2008) and we analyze four areas of CSR information, namely: employee related issues, environmental issues, products and costumers issues, and community involvement issues. This study shows that Tunisian banks do disclose CSR information. On the basis of the content analysis, it is clear that the distribution of the level of CSRD among the four categories made us conclude that “human resources” is the most reported category in the annual reports of the TLB. Whereas, it turned out that, in the websites of these banks, the most explained and detailed category is “the involvement in the community”. On the other hand, it appears that the environment is the least disclosed category, whether in the annual reports or the websites. The analysis of CSR information, on the form basis, assumes that the banks disclose CSR information essentially under a narrative form. In this regard, the banks rarely use a quantitative monetary form, whether in their annual reports or in their websites. Therefore, on the basis of the descriptive analysis, we can assert that the Tunisian banks seem

[26]

to attribute greater importance to annual reports as disclosure media than to websites to disclose CSR information. The results of the multivariate analysis help us conclude that the level of CSRD in the websites of the TLB is explained by the leverage and the financial performance. Furthermore, we document that only the variables: bank age, financial performance and State shareholding, are the determinants of the level of CSRD in the annual reports of the TLB. These results suggest that banks with higher age, performance and State shareholding attribute greater importance to CSRD as part of their reputation management strategies when compared with banks with lower age, performance and State shareholding. These banks show greater concern to improve the corporate image through CSRD. Our findings are in line with previous empirical studies conducted in other emergent markets (e.g., Abdul Hamid, 2004 in Malaysia; Eng and Mak, 2003 in Singapore). Overall, our study contributes to the growing literature concerning the determinants of CSRD in emergent markets. It explores the effect of banks characteristics and various ownership structures on CSRD and it confirm that it play a crucial role in determining the level of CSRD of banks reporting policy in an emergent market namely Tunisia. Furthermore, we focus on one of the most important markets on the North Africa region. Our findings have several implications for different groups. For instance, for investors, it provides evidence that bank age and the presence of high level of financial performance and State ownership tend to improve the level of CSRD in the banks’ annual reports. Furthermore, enhanced CSRD allows investors to impose market discipline more effectively. For banks managers, they should strive to take into account the CSR issues in their operations and thereby report them systematically and as thoroughly as possible. In addition, increased transparency through the CSRD should enable a bank to access capital markets more efficiently and reduce the risk taking. In fact, banks transparency is a key element of a safe and sound banking system. For Tunisian regulators, our findings provide evidence for them to optimize banks ownership structure. In this way, a strong State ownership will tend to improve the level of CSRD of Tunisian banks. However, our research is not free from limitations. In fact, the first limitation is related to the use of the method of manual content analysis which suffers from being subjective. The second limitation is connected to our sample reduced size and its non-random selection manner which can prevent the generalization of our results. Despite these limitations, our study provides interesting insights into the determinants of CSRD in an emergent market and helps think of [27]

new research avenues into the topic of CSR reporting. In fact, it may be interesting to carry out a study on the economic consequences of bank disclosure, since poor performance in the banking sector might be attributed to weak disclosure practices. We can also suggest using means of CSRD other than the annual reports and websites, such as brochures, ads... In addition, introducing other explanatory variables, such as, the board composition, the listing status, the existence of a unit in charge of the CSR within the bank will be relevant. Also, the study of CSRD for Islamic banks seems to be very interesting and need a special focus. Finally, we suggest the use of a disclosure proxy that captures the quality (not only the quantity) of CSR information disclosed.

References Abbott, W. F., & Monsen, R. J. (1979). “On the measurement of corporate social responsibility: self-reported disclosures as a method of measuring corporate social involvement”, Academy of Management Journal, vol 22, no 3, 501-515 Abdul Hamid, F.Z (2004). “Corporate social disclosure by banks and finance companies: Malaysian evidence”, Corporate Ownership and Control, vol 1, no 4, 118-129 Act n° 2005-96 (the October 18, 2005) on the strengthening of the financial security Act n° 1996-112 (the December 30, 1996) on the Tunisian firms' accounting system Adams C A, (2002), “Internal organizational factors influencing corporate social and ethical reporting: Beyond current theorizing”, Accounting Auditing and Accountability Journal, vol 15, n°2, 223–250 Adams, R.B., & Mehran, H., (2005). “Corporate performance, board structure and its determinants in the banking industry”, Federal Reserve Bank of New York Staff Reports, no. 330 Akano A Y, Jamiu O O, Yaya.O A. & Oluwalogbon O T, (2013). “Corporate social responsibility activities disclosure by commercial banks in Nigeria”, European Journal of Business and Management, vol 5, no 7, 173-185 Alberti-Alhtaybat L V, Hutaibat K & Al-Htaybat K, (2012), « Mapping corporate disclosure theories », Journal of Financial Reporting & Accounting, vol 10, n°1, 73-94 Al Janadi, Y. Abdu Rahman, R & Haj Omar, N. (2013). “Corporate governance mechanisms and voluntary disclosure in Saudi Arabia”, Research Journal of Finance and Accounting, vol.4, no 4, 25-35

[28]

Baccouche, C. Erraies O, & Mzoughi K,

(2010). “Les déterminants de la publication

volontaire d’informations sociales : cas des entreprises tunisiennes ”, working paper, ISCAETunis Barako, D.G., Hancock, P., & Izan, H.Y. (2006). “Factors influencing voluntary corporate disclosure by Kenyan companies”, Corporate Governance: An International Review, vol 14, no 2, 107-125 Barako, D.G & Brown, A.M. (2008). “Corporate social reporting and board representation evidence From the Kenyan banking sector”, Journal Manage Governance, vol 12: 309-324 Ben Rhouma, A., & Cormier, D. (2007). “Déterminants de la communication sociale et environnementale des entreprises françaises”, working paper, UQAM Belkaoui, A., & Karpik, P. (1989). “Determinants of the corporate decision to disclose social information”, Accounting, Auditing and Accountability Journal, vol 2, no 1, 36-51 Branco M. C, & Rodrigues L.L, (2006). “Communication of corporate social responsibility by Portuguese banks a legitimacy theory perspective”, Corporate Communications: An International Journal, vol 11, Iss 3, 232-248 Branco M. C, & Rodrigues L. L, (2008). “Social responsibility disclosure: a study of proxies for the public visibility of Portuguese banks”, The British Accounting Review, vol 40, no 2, 161-181 Brown N., & Deegan C., (1998). “The public disclosure of environmental performance information: a dual test of media agenda setting theory and legitimacy theory”, Accounting and Business Research, vol. 29, no 1, 21-41 Cabagnols, A. & Le Bas, C. (2006). “Les déterminants du comportement de Responsabilité sociale de l’entreprise: une analyse économétrique à partir de nouvelles donnée d’enquête”, working paper, Lyon 2 University-France Chakroun, R. (2012). “Confrontation between Supply and Demand for Voluntary Information in the Annual Reports in an Emerging Country: The Tunisian Case”. International Journal of Accounting and Financial Reporting. vol 2, no 2, 333-366 Chavent, M., Ding, Y., Fu, L., Stolowy, H., & Wang, H. (2006). “Disclosure and determinants studies: An extension using the divisive clustering method (DIV)”, European Accounting Review, vol 15, no 2, 181-218 Clarke, J. & Gibson-Sweet, M. (1999). “The use of corporate social disclosures in the management of reputation and legitimacy: a cross sectoral analysis of UK top100 companies”, Business Ethics: A European Review, vol. 8, no. 1, 5-13

[29]

Cormier, D., & Magnan, M., (1999). “Corporate environmental disclosure strategies: determinants, costs and benefits”, Journal of Accounting, Auditing and Finance, vol 14, 429451 Coulson, A.B., & Monks, V., (1999). “Corporate environmental performance considerations within bank lending decisions”, Eco-Management and Auditing, vol 6, no 1, 1–10 Cowen, B., Ferrari, L.B., & Parker, L.D. (1987). “The impact of corporate characteristics on social responsibility: A typology and frequency based analysis”, Accounting, Organization and Society, vol 12, no 2, 111-122 Crawford P E & Williams C C, (2010), “Should corporate social reporting be voluntary or mandatory? Evidence from the banking sector in France and the United States”, Corporate Governance: The International Journal of Business in Society, vol 10, no 4, 512-526 Damak Ayadi, S. (2004). “Comment justifier la responsabilité sociale des entreprises? ”, Entreprise Ethique, vol 20, 108-113 Damak Ayadi, S. (2006). “Analyse des facteurs explicatifs de la publication des rapports Sociétaux en France”, Comptabilité, Contrôle et Audit, vol 12, no 2, 93-116 Deegan, C. (2002). “The legitimizing effects of social and environmental disclosures a theoretical foundation”, Accounting, Auditing and Accountability Journal, vol 13, no 3, 281310 Deegan, C. & Gordon, B. (1996). “A study of the environmental disclosure practices of Australian Corporations”, Accounting and Business Research, vol 26, no 3, 187–199 Dhouibi R & Mamoghli C, (2013), “Determinants of Voluntary Disclosure in Tunisian Bank’s Reports”, Research Journal of Finance and Accounting, vol 4, No 5, 80-94 Douglas A, Doris J & Johnson B, (2004) "Corporate social reporting in Irish financial institutions", The TQM Magazine, vol. 16, Iss 6, 387 - 395 Driss, R., & Jarboui, A. (2014). “Social disclosure: what are the main explanatory factors? An empirical test in the Tunisian context”, International Journal of Accounting and Economics Study, vol 2, no 1, 34-39 Eng, L.L., & Mak, Y.T. (2003). “Corporate governance and voluntary disclosure”, Journal of Accounting and Public Policy, vol 22, 325-345 Ernst & Ernst. (1978). “Social responsibility disclosure: Survey of Fortune 500 Annual Reports”. Cleveland Ohio Freeman, R. E. (1984).”Strategic management: A stakeholder approach”, Pitman, Boston Fry E.L., & Hock R.J. (1976). “Who claims corporate responsibility? The biggest and the worst”, Business and Society Review, vol. 18, 62-65 [30]

Gray R., Kouhy R., & Lavers S. (1995). “Corporate social and environmental reporting: A review of the literature and longitudinal study of UK disclosure”, Accounting, Auditing and Accountability Journal, vol. 8, no 2, 47-77 Global Reporting Initiative (GRI) (2006). Sustainability Reporting Guidelines 3. Available at: http://www.globalreporting.org. Guide of the annual report of the Tunisian enterprises, (2009), Arab Institute of Business Leaders Guthrie, J., & Parker, L.D. (1989). “Corporate social reporting: A rebuttal of legitimacy theory”, Accounting and Business Research, vol 19, Issue 76, 343-352 Hackston, D., & Milne, M. (1996). “Some determinants of social and environmental disclosure In New-Zealand companies”, Accounting, Auditing and Accountability Journal, vol 9, no 1, 77-108 Haniffa R M & Cooke T E, (2002), “Culture, corporate governance and disclosure in Malaysian corporations”, ABACUS, vol 38, no 3, 317-347 Haniffa, R.M., & Cooke, T.E. (2005). “The impact of culture and governance on corporate social reporting”, Journal of Accounting and Public Policy, vol 24, 391- 430 Henault, G. & Lemoine, G. (2008). “Entrepreneuriat et développement durable en Francophonie”, Edition des archives contemporaines-savoirs francophones. Paris Hill, C.W.L, & Jones, T.M. (1992). “Stakeholder-agency theory”, Journal of Management Studies, vol 29, no 2, 131-154 Ho S S M & 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 Hogner, R. H. (1982). “Corporate social reporting: eight decades of development at US Steel”, Research in Corporate Performance and Policy, vol 4, 243-250 Hossain, M., (2008). “The extent of disclosure in annual reports of banking companies: The case of India”. European Journal of Scientific Research, vol 23, No 4, 659-680 Hossain M, & Reaz M, (2007). “The determinants and characteristics of voluntary disclosure by Indian banking companies”, Corporate Social Responsibility and Environmental Management, 14, Issue 5, 274–288 Hossain M, Perera, M.H.B, & Rahman AR. (1995). “Voluntary disclosure in the annual reports of New Zealand companies”, Journal of International Financial Management and Accounting, vol 6, no 1, 69–87

[31]

Htay S N N, Ab Rashid H M, Adnan M A & Mydin Meera A K, (2012), “ Impact of corporate governance on social and environmental information disclosure of Malaysian listed banks: Panel data analysis”, Asian Journal of Finance and Accounting, vol 4, n°1, 1-24 Htay S N N, Mohd Said R & Salman S A, (2013), “Impact of corporate governance on disclosure quality: Empirical evidence from listed banks in Malaysia”, International Journal of Economics and Management, vol 7, n°2, 242-279 Huang X B & Watson L (2015), “Corporate social responsibility research in accounting”, Journal of Accounting Literature, vol 34, 1-16 Jensen, M. & Meckling W, (1976). “Theory of the firm: managerial behavior, agency costs and ownership structure”, Journal of Financial Economics, vol 3, 305-360 Jeucken, M.H.A & Bouma, J.J., (1999). “The changing environment of banks”, Greener Management International, vol 27, 21–35 Joseph, K.A. (2008). “Corporate social responsibility in Nigerian banking system”, Society and Business Review, vol 3, no 1, 57-71 Khan, M.H. (2010). “The effect of corporate governance elements on corporate social responsibility (CSR) reporting: Empirical evidence from private commercial banks of Bangladesh”, International Journal of Law and Management, vol 52, no 2, 82-109 Khlif H, Guidara A & Souissi M, (2015),"Corporate social and environmental disclosure and corporate performance", Journal of Accounting in Emerging Economies, Vol 5, Iss 1, 51 – 69 Khemir S & Baccouche C, (2010), “Analysis of the determinants of corporate social responsibility disclosure in the annual reports of Tunisian listed firms”, Research in Accounting in Emerging Economies, vol 10, 119–144 Kribat M, Burton B & Crawford L, (2013), « Evidence on the Nature, Extent and Determinants of Disclosures in Libyan Banks’ Annual Reports », Journal of Accounting in Emerging Economies, vol 3, No 2, 88-114 Lipunga A. M. (2013).”Corporate social responsibility reporting by commercial banks in annual reports: evidence from Malawi”. International Journal of Business and Social Research, vol 3, no 9, 88-101 Masud M.A.K, & Hossain M.S, (2012). “Corporate social responsibility reporting practices in Bangladesh: A study of selected private commercial banks”, Journal of Business and Management, vol 6, no 2, 42-47 Mostovicz E I, Kakabadse A & Kakabadse N K, (2011), “The four pillars of corporate responsibility: ethics, leadership, personal responsibility and trust”, Corporate Governance: The International Journal of Business in Society, vol 11, No 4, 489-500 [32]

Mohadeo J D, Oogarah-Hanuman V & Soobaroyen T, (2011), “A longitudinal study of corporate social disclosures in a developing economy”, Journal of Business Ethics, vol 104, 545-558 Naser, K., Al- Hussaini, A., Al-Kwari, D., & Nuseibeh, R., (2006). “Determinants of corporate social disclosure in Developing countries: The case of Qatar”, Advances in International Accounting, vol 19, 1-23 Ness, K. E., & Mirza, A.M. (1991). “Corporate social disclosure: A note on a test of agency theory”, British Accounting Review, vol 23, Issue 3, 265-282 Neu, D., Warsame, H., & Pedwell, K. (1998). “Managing public impressions: environmental Disclosures in annual reports”, Accounting, Organization and Society, vol 23, no 3, 265-282 Oxibar, B. (2005).”La diffusion d’information sociétale : outils de mesure et déterminants : une comparaison multi supports”, working paper, Paris Dauphine University Paulet E, (2011). "Banking ethics", Corporate Governance: The International Journal of Business in Society, Vol. 11 Iss: 3, 293 - 300 Parker, L. D. (1982). “Corporate annual accounting: A mass communication perspective”, Accounting and Business Research, vol 12, no 48, 279-286 Patten D M, (1991), “Exposure, legitimacy, and social disclosure”, Journal of Accounting and Public Policy, vol 10, 297-308 Patten D.M. (1992). “Intra-industry environmental disclosures in response to the Alaskan oil spill: A note on legitimacy theory”, Accounting, Organizations and Society, vol 17, no 5, 471475 Patten, D.M., & Crampton, W., (2004). “Legitimacy and the internet: an examination of corporate web page environmental disclosures”, Advances in Environmental Accounting and Management, vol 2, 31–57 Peters S, Miller M & Kusyk S, (2011), “How relevant is corporate governance and corporate social responsibility in emerging markets?”, Corporate Governance: The International Journal of Business in Society, vol 11, no 4, 429-445 Rabah Gana M., & Dakhlaoui M. (2011). “Societal information disclosure and the cost of equity: The case of Tunisian companies”, Global Journal of Management and Business Resaerch, vol. 11, Issue 7 Roberts R W, (1992), “Determinants of corporate social responsibility disclosure: An application of Stakeholder theory”, Accounting Organizations and Society, vol 17, no 6, 595612

[33]

Scholtens B. (2009). “Corporate social responsibility in the international banking industry”, Journal of Business Ethics, vol 86, 159–175 Sethi, P, (2013).”Social reporting by Indian banks and foreign banks: a comparative study”, Journal of Business and Management. vol 15, no 3, 45-53 Sharma, N (2011). “CSR practices and CSR reporting in Indian banking sector”, International Journal of Advanced Economic and Business Management, vol 1, no 2, 58-66 Stanwick. P.A., & Stanwick S.D (1998). “The relationship between corporate social performance and organizational size, financial performance and environmental performance: an empirical examination”, Journal of Business Ethics, vol. 17, no 2, 195-204 Suchman, M. C. (1995). “Managing legitimacy: strategic and institutional approaches”, Academy of Management Review, vol. 20, no 3, 571-610 Tarna, K. (1999). “Reporting on the environment: current practice in the financial services Sector”, Greener Management International, vol 26, 49-64 Tsang, E.W.K., (1998). “A longitudinal study of corporate social reporting in Singapore: The case of the banking, food and beverages and hotel industries”, Accounting, Auditing and Accountability Journal, vol 11, Issue 5, 624–635 Unerman J, (2000), « Methodological issues: Reflections on quantification in corporate social reporting content analysis », Accounting Auditing and Accountability Journal, vol 13, no 5, 667–680 Ullmann, A. A. (1985). “Data in search of a theory: a critical examination of the relationship among social performance, social disclosure and economic performance”, The Academy of Management Review, vol 10, no 3, 540-557 Upadhyay-Dhungel K & Dhungel, A (2012). “Corporate social responsibility reporting practices in the banking sector of Nepal”, Banking Journal .vol 3, Issue 1, 61-78 Watts, R. L. & Zimmerman J.L. (1978). “Towards a positive theory of the determination of accounting standards”, The Accounting Review, vol 53, no 1, 112-134 Wilmshurst T D & Frost G R, (2000), “Corporate environmental reporting: A test of legitimacy theory”, Accounting Auditing and Accountability Journal, vol 13, no 1, 10–26 Williams S M & Ho Wern Pei C A, (1999), “Corporate Social Disclosures by Listed Companies on Their Web Sites: An International Comparison”, The International Journal of Accounting, vol 34, no 3, 389-419 Yeshmin F, (2012). “Visualization of corporate social responsibility information of commercial bank in Bangladesh”, World Journal of Social Sciences, vol 2, no 5, 114 – 127

[34]

Yeung, S, (2011). “The role of banks in corporate social responsibility”, Journal of Applied Economics and Business Research. vol 1, no 2, 103-115 Zappi, G (2007). “Corporate responsibility in the Italian banking industry: creating Value through listening to stakeholders”, Corporate Governance: The International Journal of Business in Society, vol 7, Issue 4, 471-475 Zéghal D. & Ahmed S.A. (1990), “Comparison of social responsibility information disclosure media used by Canadian firms”, Accounting, Auditing and Accountability Journal, vol.3, no 1, 38-53

[35]

Table 1: Independent variables measurement and expected signs Abbreviated name

Variable full name

Variable description

Predicted sign

Measurement proposed by the previous literature Total balance sheet : Hackston & Milne (1996), Branco & Rodrigues (2008) Turnover : Belkaoui & Kaprick (1989), Hackston & Milne (1996) Market value : Hackston & Milne (1996) Number of branches : Branco & Rodrigues (2006 & 2008), Akano et al. (2013) Number of years since the setting up of the business : Baccouche et al. (2010), Hossein & Reaz (2007), Roberts (1992), Hamid (2004) Return on equity (ROE) : Patten (1991), Roberts (1992), Oxibar (2001 & 2005), Damak-Ayadi (2004) Return on assets (ROA) : Patten (1991), Damak-Ayadi (2004), Branco & Rodrigues (2008) Net Income / Trurnover : Damak-Ayadi (2004) Long-term debts / Total assets : Oxibar (2003 & 2005), Driss & Jarboui (2014) Total debts / Total assets : Belkaoui & Kaprick (1989), Branco & Rodrigues (2008), Khemir & Baccouche (2010) Dummy variable; 1 if the State is shareholder and 0 if not : Baccouche et al. (2010) Percentage of shares owned by the state : Al Janadi et al. (2013) Percentage of shares owned by Foreign shareholders : Baccouche et al. (2010), Haniffa & Cooke (2002 & 2005), Branco & Rodrigues (2006), Al Janadi et al. (2013) Dummy variable: 1 if the bank is audited at least by a Big 4 and 0 if not: Baccouche et al. (2010)

BKSIZE

Bank Size

Number of brunches

bank

+

BKAGE

Bank Age

+/-

ROE

Economic Performance

Number of the operating years since the setting up of the business Return on Equity = Net Income / Equity

LEV

Leverage

Debt ratio = Total debts / Total assets

+/-

STATE

State Ownership

Percentage of shares held by the State

+

FORGN

Foreign Ownership

Percentage shares held foreigners

of by

+

AUDIT

Type Auditor

Dummy variable: 1 if the bank is audited at least by a Big 4 and 0 if not

+/-

of

+/-

[36]

Table 2: The overall level of CSR disclosure in the annual reports 2007

2008

2009

2010

2011

2012

202 192 327 358 NS 11.08 10.53 17.94 19.64 % CSRDIS_AR = The level of CSR disclosure in the annual reports. NS = The number of sentences.

344 18.87

400 21.94

CSRDIS_AR

Overall disclosure level 1823 100

Table 3: The disclosure level of CSR categories in the websites and annual reports Categories Category A Category B Category C Category D

Annuel Reports NS 36 1297 269 221

Websites

Per cent 1.97 % 71.14 % 14.75 % 12.14%

NS 1 54 34 125

Per cent 0.46 % 25.23 % 15.88 % 58.43 %

Overall disclosure 1823 100% 214 100% Category A: Environmental disclosure. Category B: Human resources disclosure. Category C: Products and customers disclosure. Category D: Community involvement disclosure. NS= The number of sentences.

Table 4: The CSR disclosure level depending on the information form in the websites and annual reports

Information forms Narrative Quantitative and non monetary Quantitative and monetary Overall disclosure NS= The number of sentences.

Annuel Reports NS Per cent 1044 57.27 % 620 34 %

NS 200 11

Websites Per cent 93.45 % 5.14 %

159

8.73 %

3

1.41 %

1823

100 %

214

100 %

[37]

Table 5: Summary of the descriptive statistics Panel A: (Model 1 variables) Variables

N

Minimum

Maximum

Mean

Std. deviation

CSRDIS_AR 66 4 86 27.62121 18.01774 BKSIZE 66 5 185 108.5909 39.35244 ROE 66 -11.87 29.76 9.622121 16.54069 LEV 66 64.55 101.09 89.65135 6.984151 STATE 66 0 66.76 20.54373 27.59613 FORGN 66 0 64.24 30.11726 23.41323 BKAGE 66 19 129 47.95455 27.0637 CSRDIS_AR = The level of CSR disclosure in the annual reports. BKSIZE = Number of bank brunches. ROE = Net Income / Equity. LEV = Total debts / Total assets. STATE = Percentage of shares held by the State. FORGN = Percentage of shares held by foreigners. BKAGE = Number of the operating years since the setting up of the business. AUDIT Frequency Percentile 0 16 24.24 1 50 75.76 AUDIT = 1 if the bank is audited at least by a Big 4, 0 if not.

Panel B: (Model 2 variables) Variables N Minimum Maximum Mean Std. deviation CSRDIS_WS 11 0 95 19.45455 27.9441 BKSIZE 11 19 189 124.4545 45.57711 ROE 11 1.5 163.03 25.32273 46.43749 LEV 11 83.11 101.62 92.01464 5.184594 STATE 11 0 67.4 19.59273 27.92948 FORGN 11 5.64 64.24 31.27591 23.08872 BKAGE 11 25 130 51.45455 28.11179 CSRDIS_WS = The level of CSR disclosure in the websites. BKSIZE = Number of bank brunches. ROE = Net Income / Equity. LEV = Total debts / Total assets. STATE = Percentage of shares held by the State. FORGN = Percentage of shares held by foreigners. BKAGE = Number of the operating years since the setting up of the business. AUDIT Frequency Percentile 0 3 27.27 1 8 72.73 AUDIT = 1 if the bank is audited at least by a Big 4, 0 if not.

[38]

Table 6: Summary of correlation analyses Panel A: Pearson correlation matrix (N = 66)

CSRDIS_AR BKSIZE ROE LEV STATE FORGN BKAGE CSRDIS_WS

CSRDIS_AR 1.0000 0.4293 0.1880 0.2242 -0.4122 0.1072 -0.1892 0.2587

BKSIZE 1.0000 0.1355 0.0757 -0.2052 -0.2836 0.0928 0.1341

ROE

1.0000 -0.0305 -0.0745 -0.0835 0.0890 0.1067

LEV

1.0000 -0.2477 -0.1689 -0.0882 0.1009

STATE

FORGN

BKAGE

1.0000 -0.4515 -0.2437 -0.4745

1.0000 -0.1594 0.4169

1.0000 -0.1136

CSRDIS_WS

Panel B: VIF tests Variables LEV BKSIZE STATE FORGN BKAGE ROE

Model 1 VIF 2.77 2.76 1.84 1.76 1.31 1.08 Mean VIF = 1.92

1/VIF 0.360738 0.362593 0.544087 0.568340 0.760936 0.925598

[39]

Model 2 VIF 1/VIF 2.27 0.439784 2.19 0.455973 1.89 0.529313 1.80 0.556764 1.72 0.579865 1.13 0.881229 Mean VIF = 1.84

1.0000

Table 7: Results of the first model multivariate regression Panel A: Poisson regression Variables Coeff Z Sig BKSIZE 0.100605 11.12 0.000*** BKAGE 0.0038103 2.03 0.043** ROE -0.0351836 -5.82 0.000*** LEV -0.0118179 -8.60 0.000*** STATE -0.0025541 -2.20 0.028** FORGN -0.0106882 -8.98 0.000*** AUDIT 0.05623 0.70 0.481 Deviance = 325.2178 Sig = 0.0000*** Chi-2 of Pearson = 325.3045 Sig = 0.0000*** BKSIZE = Number of bank brunches. ROE = Net Income / Equity. LEV = Total debts / Total assets. STATE = Percentage of shares held by the State. FORGN = Percentage of shares held by foreigners. BKAGE = Number of the operating years since the setting up of the business. AUDIT = 1 if the bank is audited at least by a Big 4, 0 if not *, **, *** significant at the 10%, 5%, 1% level respectively.

Panel B: Hausman test Chi –2 P-value

104.46 0.0000***

Panel C: Negative binomial regression Variables Predicted signs Coeff Z Sig BKSIZE + -0.0067 -1.28 0.199 BKAGE +/0.1497 4.59 0.000*** ROE +/0.0062 2.09 0.037** LEV +/-0.0207 -1.05 0.294 STATE + 0.0294 1.68 0.092* FORGN + 0.0127 0.71 0.476 AUDIT +/-0.2143 -1.16 0.245 Log Likelihood = -185.02751 Chi 2 = 39.07 Sig = 0.0000*** BKSIZE = Number of bank brunches. ROE = Net Income / Equity. LEV = Total debts / Total assets. STATE = Percentage of shares held by the State. FORGN = Percentage of shares held by foreigners. BKAGE = Number of the operating years since the setting up of the business. AUDIT = 1 if the bank is audited at least by a Big 4, 0 if not *, **, *** significant at the 10%, 5%, 1% level respectively.

[40]

Table 8: Results of the second model multivariate regression Panel A: Poisson Regression Variables Coeff Z Sig BKSIZE -0.0318 -7.06 0.000*** BKAGE -0.0534 -6.69 0.000*** ROE 1.0277 6.16 0.000*** LEV 0.0426 1.32 0.187 STATE 0.0193 4.52 0.000*** FORGN 0.0232 2.72 0.007*** AUDIT 14.9519 4.58 0.000*** Deviance = 65.187 Sig = 0.0000*** Wald Chi 2 = 18501.27 Sig = 0.0000*** BKSIZE = Number of bank brunches. ROE = Net Income / Equity. LEV = Total debts / Total assets. STATE = Percentage of shares held by the State. FORGN = Percentage of shares held by foreigners. BKAGE = Number of the operating years since the setting up of the business. AUDIT = 1 if the bank is audited at least by a Big 4, 0 if not *, **, *** significant at the 10%, 5%, 1% level respectively.

Panel B: Negative binomial regression Variables Predicted signs Coeff Z Sig BKSIZE + 0.0049 0.36 0.721 BKAGE +/-0.0140 -0.66 0.507 ROE +/-0.0184 -1.75 0.080* LEV +/0.3305 1.85 0.064* STATE + 0.0023 0.03 0.973 FORGN + 0.0292 1.31 0.189 AUDIT +/3.6239 0.79 0.430 Log Likelihood = -35.236 Chi 2 = 13.16 Sig = 0.0684* BKSIZE = Number of bank brunches. ROE = Net Income / Equity. LEV = Total debts / Total assets. STATE = Percentage of shares held by the State. FORGN = Percentage of shares held by foreigners. BKAGE = Number of the operating years since the setting up of the business. AUDIT = 1 if the bank is audited at least by a Big 4, 0 if not *, **, *** significant at the 10%, 5%, 1% level respectively.

[41]

Table 9: Marginal effect of independent variables Variables dy/dx Z Sig BKSIZE 0.0382684 0.35 0.724 BKAGE -0.1434963 -1.69 0.091* ROE 2.56905 1.83 0.067* LEV 0.0179636 0.03 0.973 STATE 0.2273858 1.23 0.220 FORGN -0.1091806 -0.65 0.517 AUDIT 20.32712 0.73 0.464 BKSIZE = Number of bank brunches. ROE = Net Income / Equity. LEV = Total debts / Total assets. STATE = Percentage of shares held by the State. FORGN = Percentage of shares held by foreigners. BKAGE = Number of the operating years since the setting up of the business. AUDIT = 1 if the bank is audited at least by a Big 4, 0 if not *, **, *** significant at the 10%, 5%, 1% level respectively.

[42]

Appendix 1: List of banks Banks AMEN BANK (AB) BANQUE DE TUNISIE (BT) BANQUE INTERNATIONALE ARABE DE TUNISIE (BIAT) UNION BANCAIRE POUR LE COMMERCE ET L’INDUSTRIE (UBCI) BANQUE ATTIJARI DE TUNISIE (ATTIJARI) BANQUE DE L’HABITAT (BH) UNION INTERNATIONALE DE BANQUES (UIB) SOCIETE TUNISIENNE DE BANQUE (STB) ARAB TUNISIAN BANK (ATB) BANQUE NATIONALE AGRICOLE (BNA) BANQUE DE TUNISIE ET DES EMIRATS (BTE) Overall

Characteristic Private bank Private bank Private bank Foreign bank Foreign bank Public bank Foreign bank Public bank Private bank Public bank Foreign bank

Websites www.amenbank.com.tn www.bt.com.tn www.biat.com.tn www.ubci.com.tn www.attijaribank.com.tn www.bh.com.tn www.uib.com.tn www.stb.com.tn www.atb.com.tn www.bna.com.tn www.bte.com.tn 11 banks

Appendix 2: Grid computing CSR disclosure (Branco & Rodrigues, 2006 & 2008) A12345B67891011121314C151617D1819202122-

Environmental disclosure Environmental policies or company concern for the environment Environmental management, systems and audit Lending and investment policies Conservation of natural resources and recycling activities Conservation of energy in the conduct of business operations Human resources disclosure Employee health and safety Employment of minorities or women Employee training Employee assistance/benefits Employee remuneration Employee profiles Employee share purchase schemes Employee morale Industrial relations Product and consumers disclosure Product quality Consumer complaints/satisfaction Provision for disabled, aged, and, difficult-to-reach-customers Community involvement disclosure Charitable donations and activities Support for education Support for the arts and culture Support for public health Sponsoring sporting or recreational projects

[43]