Abstract
Corporate social responsibility (CSR) in developing countries has recently received increasing attention, and scholars have pointed to the strong contextuality of CSR in the respective regions. Regarding the latter, however, sub-Saharan Africa has been scrutinized only marginally by academia. Moreover, empirical research on the impact of the institutional context has been scant, despite its attributed importance for CSR. Our article seeks to fill a part of this research gap by investigating CSR website reporting of 211 companies in seven sub-Saharan countries. The study’s aim is twofold: First, we identify to what extent sub-Saharan companies report on CSR and which contents they disclose. Second, by building on institutional theory, we investigate how the socio-economic and political environments influence CSR reporting. For this purpose, we examine the impact of country-level and company-level determinants. We find that the sample African companies’ CSR efforts focus strongly on local philanthropy and therefore differ substantially from Western CSR approaches. Furthermore, we evidence that GDP and level of governance standard positively affect CSR reporting. Our study contributes to the literature by empirically evidencing the contextuality of CSR in Africa and by explaining how specific country- and company-level determinants contribute to or hamper the development of CSR in developing countries.
Corporate social responsibility (CSR) in Africa is still largely an exploration of “terra incognita.” While a substantial number of studies has addressed the state of CSR in other developing regions such as Latin America (e.g., Haslam, 2004; Logsdon et al., 2006; Paul et al., 2006; Peinado-Vara, 2004, 2006; Prieto-Carrn, 2006; Schmidheiny, 2006) and especially Asia (e.g., Baughn, Bodie, & McIntosh, 2007; Birch & Moon, 2004; Chambers, Chapple, Moon, & Sullivan, 2003; Chapple & Moon, 2005; Higgins & Debroux, 2009; Kolk et al., 2010; Ramasamy & Woan Ting, 2004; Welford, 2004; for an extensive overview of the literature, see Fifka, 2012, 2013), studies on Africa have remained scarce (Dartey-Baah & Amponsah-Tawiah, 2011; Kivuitu et al., 2005; Kolk & Lenfant, 2010; Phillips, 2006; Visser, 2006a). A notable exception are the studies on CSR in South Africa as the continent’s most developed economy (e.g., Antonites & De Villiers, 2003; Dawkins & Ngunjiri, 2008; De Villiers, 1995, 1998, 1999, 2003; De Villiers & Barnard, 2000; De Villiers & Van Staden, 2006, 2010; Hamann, Agbazue, Kapelus, & Hein, 2005; Mitchell & Hill, 2009). There exists also a considerable body of studies on Nigeria focusing on the ambiguous consequences of the country’s richness in oil (e.g., Amaeshi, Adi, Ogbechie, & Amao, 2006; Eweje, 2006; Idemudia & Ite, 2006; Ojo, 2008; Omeje, 2006; Wheeler, Fabig, & Boele, 2002).
This general gap of research on CSR in Africa, which has been attested by several studies (Baskin, 2006; Rizk, Dixon, & Woodhead, 2008; Visser, 2006a), can be attributed to various reasons. First, African companies have not been considered in internationally comparative studies due to their relatively small size in comparison to multinational enterprises from other continents. Even studies with large samples such as the Fortune Global 250 (Kolk & Perego, 2010) or the Fortune Global 500 (Morhardt, 2010) did not include African companies for the simple reason that there were none on these rankings. More importantly, as Kolk and Lenfant (2010) have appropriately observed, it is difficult to do
research in countries where governance and institutional structures are in flux, and where violent conflicts are prevalent, with different groups fighting for scarce resources and/or government power, with factors involving ethnicity or religion playing a role as well at times. (p. 242)
Finally, the poor socio-economic and political conditions do not only make research on CSR as such more difficult. They also result in a considerably weaker development of CSR in Africa than in other developing regions, as existing studies suggest (Baskin, 2006; Visser, 2006a). Thus, there are lesser “points of contact” for investigating how CSR is practiced by companies in Africa.
In regions where governance and institutional structures are weak, such as sub-Saharan Africa (Alence, 2004), the private sector remains one of the institutions that are able to contribute to an improvement of social, economic, and environmental conditions (B. Campbell, 2012). This potential impact has been discussed particularly for the areas of socio-economic development (International Business Leadership Forum [IBLF] & World Business Council for Sustainable Development [WBCSD], 2004; Nelson & Prescott, 2003) and poverty alleviation (IBLF, 2002; Prahalad, 2004; Prahalad & Hammond, 2002; WBCSD, 2004). These envisaged corporate contributions are often seen to represent a core element of CSR in Africa (Visser, 2006b). Against this background, it needs to be examined how companies themselves convey this role as development agents in their corporate communications and what they perceive to be the focus of their CSR efforts. Moreover, there are strong indications by theoretical and empirical works that this understanding and communication varies across differing institutional contexts (Aguilera & Jackson, 2003; Granovetter, 2000; Guerreiro, Pereira, & Frezatti, 2006; Jackson & Apostolakou, 2010; Matten & Moon, 2008).
These initial considerations provide the basis for the two research questions pursued in this study: First, we seek to answer the question to what extent companies in sub-Saharan Africa undertake CSR reporting, and which contents they include in their disclosure efforts. This allows conclusions on the understanding of CSR displayed by corporations in sub-Saharan Africa and on the areas of CSR they emphasize. Second, our article examines to what degree reporting is influenced by the region’s institutional environment to generate insights into the contextuality of CSR in the African context.
Similar to previous studies on CSR in Africa (Hamann et al., 2005; Idemudia & Ite, 2006; Kolk & Lenfant, 2010), we also assume that CSR is highly contextual. Therefore, we build on institutional theory to investigate the contents reported by the 211 companies in our sample. Building on an earlier study by Chapple and Moon (2005) on CSR website reporting, we also seek to provide an encompassing investigation of potential contents to be disclosed. Based on a directed content analysis of corporate websites, we examine to what extent and on which instruments sub-Saharan companies report for implementing CSR (such as codes of conduct), dimensions of CSR (community involvement, employee relations, production processes), and channels of corporate citizenship (e.g., sponsoring and volunteering). Through this comprehensive approach, we seek to fill a part of the large research gap on CSR in Africa.
To generate an understanding of the relationship between companies and their institutional environments, we take a multi-level approach by considering both, country-level and company-level determinants in our study. More precisely, in logistic regression analyses, we examine the impact of three selected determinants - GDP, life expectancy, and corruption - on CSR reporting in seven sub-Saharan countries: Kenya, Botswana, Ghana, Tanzania, Uganda, Nigeria, and Zambia. As many existing studies have found company-level determinants to have an impact on CSR (Chapple & Moon, 2005; Gamerschlag, Möller, & Verbeeten, 2011; Haniffa & Cooke, 2005; Reverte, 2009; Wanderley, Lucian, Farache, & De Sousa Filho, 2008), we also include four company-level determinants frequently found to be of relevance (Fifka, 2013; Hackston & Milne, 1996): company nationality, size, industry affiliation, and degree of internationalization.
The contributions we make are threefold: We broaden the understanding of the relationship between companies and their institutional environment in general (Adams, 2002; Adams & Harte, 1998; Brammer, Jackson, & Matten, 2012) and take one of the first steps in responding to calls for an investigation of this relationship in the African context in particular (Rivera-Santos, Holt, Littlewood, & Kolk, 2015). We do so by examining descriptively and empirically how socio-economic and political determinants affect CSR reporting. In this context, we provide the first encompassing study of CSR reporting, examining which aspects of CSR are emphasized in the reporting by companies from seven countries in sub-Saharan Africa. This allows conclusions on the understanding of CSR in this region. Finally, this exploratory analysis contributes to managerial literature on Africa, which has found to be scarce (Zoogah, 2008, 2014; Zoogah & Nkomo, 2013). We particularly follow the recommendation by Zoogah and Nkomo (2013, p. 12) to look through an “institutionalist lense” as a vital approach to better understand management practices in Africa.
To pursue the two research questions stated above, we start by applying institutional theory to describe the socio-economic and political environment of sub-Saharan Africa and to develop the research hypotheses. Afterward, we explain the selection of our sample and the research methodology, ere we conduct the descriptive and the regression analyses. After presenting the results, we discuss the implications of our findings and develop recommendations for future research.
The Institutional Environment for CSR in Sub-Saharan Africa and Related Hypotheses
Companies are embedded in institutional frameworks and thus belong to certain socio-economic and political systems (Jones, 1999; Whitley, 1999). More precisely, institutional theory argues that companies face certain direct and indirect institutional pressures that have an impact on their values, norms, organizational structures, and actions. Therefore, also the understanding and implementation of CSR varies across different institutional contexts (Aguilera & Jackson, 2003; Granovetter, 2000; Guerreiro et al., 2006; Jackson & Apostolakou, 2010; Matten & Moon, 2008). As DiMaggio and Powell (1983) remark, the institutional environment forms “a context, in which individual efforts to deal rationally with uncertainty and constraints often lead, in the aggregate, to homogeneity in structure, culture, and output” (p. 147). Regarding Africa in particular, researchers have argued that CSR is rooted in the respective local institutions (Amaeshi et al., 2006; Hamann et al., 2005; Idemudia & Ite, 2006). Hence, the term institutions has to be perceived broadly in this context, and it refers not only to governmental and bureaucratic institutions but also to prevailing codified and noncodified norms of behavior and values (Fifka & Pobizhan, 2014).
However, institutional theory as an approach to investigate CSR-related developments is rather new and has only emerged in the mid-2000s (Aguilera, Rupp, Williams, & Ganapathi, 2007; J. L. Campbell, 2007; Matten & Moon, 2008). Barley (2007) thus argues that management literature still lacks a broader understanding of the relationship between companies and their institutional environments. This research gap also becomes evident in studies on CSR reporting. As most studies perceive company characteristics to be the prevalent determinants (Brammer et al., 2012), the impact of organizational determinants on reporting has been intensively studied (Adams, 2002; Fifka, 2013; Gray, Kouhy, & Lavers, 1995; Gray, Javad, Power, & Sinclair, 2001). The analysis of the influence of country-level determinants on CSR reporting is more complex, as it requires the isolation of contextual determinants and the relations between them (Adams, 2002; Adams & Harte, 1998; Brammer et al., 2012). Based on institutional theory, we acknowledge that country-level determinants can also be of substantial influence for companies’ CSR disclosure. We follow an approach taken in previous studies (Chambers et al., 2003; Chapple & Moon, 2005; Jamali & Mirshak, 2007) that consider country- and company-level determinants to act as complements, as both can affect CSR disclosure. Accordingly, we differentiate between two groups of determinants of CSR reporting in our study: country-level determinants (socio-economic and political determinants, such as GDP, life expectancy, and corruption), and company-level determinants (company nationality, size, industry membership, and internationalization). To address this aspect of multi-level analysis in our study, we conduct a complementary analysis of the country-level and company-level determinants, which we discuss in the following section.
Country-Level Determinants
Referring to institutional theory, we assume that there is a relationship between the sub-Saharan African companies’ institutional context and their CSR practices and reporting (Aguilera & Jackson, 2003; Granovetter, 2000; Guerreiro et al., 2006; Jackson & Apostolakou, 2010; Matten & Moon, 2008). As the institutional environment comprises a socio-economic and a political dimension (Jones, 1999; Whitley, 1999), a large variety of factors can be utilized for its specification. Based on previous findings (Alence, 2004; Globerman & Shapiro, 2002; Lalountas et al., 2011; Meadowcroft, 2007; Seleim & Bontis, 2009), we selected determinants that have been shown to positively affect governance and CSR measures taken by firms. Companies located in more economically and socially advanced countries, as well as in countries with higher levels of transparency and lower levels of corruption, are more likely to acknowledge the importance of CSR reporting. Furthermore, companies in stable institutional environments also tend to value the benefits of adequate CSR reporting, such as more reliability, trust, accountability, a better reputation, and legitimacy from stakeholders (Chambers et al., 2003). We thus opted for examining country-level determinants related to economic development (so did Chambers et al., 2003; Chapple & Moon, 2005; Williams, 1999), social development (Chambers et al., 2003; Chapple & Moon, 2005; Nzekwu, 2007), and the level of governance standards (Chapple & Moon, 2005; Hodess, Banfield, & Wolfe, 2001; Van Der Laan Smith, Adhikari, & Tondkar, 2005; Williams, 1999).
Regarding economic development, sub-Saharan Africa is still the poorest region in the world and the only one where poverty has increased since 1990. Even though real per capita income has slightly increased in the last two decades, the number of people living in absolute poverty has augmented by 100 million. Moreover, the fiscal situation has become more unstable, as current account deficits have widened across the region. Additional problems have recently arisen for the so-called “frontier markets” in sub-Saharan Africa, such as Nigeria, Ghana, and Zambia. This term describes the equity markets of developing countries that are small and not easy to access, but where investments can be made, nevertheless. Due to a change in investor sentiment, capital inflows have been reduced. Thus, access to capital for governments and business has become more difficult (the World Bank, 2014).
Concerning the impact of economic development on CSR reporting, we follow the argument that “CSR is a function of economic wealth” (Chambers et al., 2003, p. 3). Several studies have been able to show that CSR increases with economic development (Buhr & Freedman, 2001; Welford, 2004). Welford (2005) and Baughn et al. (2007) have used GDP as a potential proxy for economic wealth to demonstrate this correlation. Welford (2005), moreover, investigated reporting as one of the elements comprising CSR. In line with these studies, we consider reporting as an expression of CSR and assume that it increases with higher economic development. Thus, we formulate Hypothesis 1a as follows:
Social development represents a further dimension of a country’s development status. In this regard, we suppose that higher levels of social development (including higher education levels, higher standards of living, and higher life expectancy) result in higher demands for CSR, as suggested by previous studies (Chambers et al., 2003; Chapple & Moon, 2005). Hence, more socially developed countries represent more stable contexts for the development of CSR (Lam, 2014). Steurer and Konrad (2009) have demonstrated a positive relationship between social development and CSR. In turn, Vormedal and Ruud (2009, p. 220), following Kolk et al. (2001) and Kolk (2003), posited that the level and quality of CSR disclosures are linked to “societal, political and legal characteristics and developments.” The absence of such socio-political drivers reduces the extent and quality of reporting (Vormedal & Ruud, 2009). The existence of such drivers in turn, as Trotman and Bradley (1981) noted, creates pressure for companies to disclose CSR-related information. Thus, we assume that social development positively influences CSR reporting:
The level of governance standards represents a political determinant that can affect the extent of CSR (Cormier & Magnan, 1999; Liu & Anbumozhi, 2009; Meadowcroft, 2007; Mitchell & Hill, 2009). For instance, the level of governance standards is expressed through the enforcement of the rule of law, government policies, freedom of speech, freedom of press, freedom of assembly, democratic practices, and higher levels of transparency inducing less corruption (Alence, 2004; Plumptre & Graham, 1999; United Nations Development Programme [UNDP], 1997). Governments of countries that are more advanced and stable in the establishment of such societal rules are more likely to regulate businesses with respect to their economic, social, and environmental compliance (Scherer & Palazzo, 2011). With regard to CSR reporting in specific, Neu, Warsame, and Pedwell (1998) evidenced that higher levels of regulatory governance lead to more extensive CSR reporting. Also in the African context, such a relationship has been shown (Rahaman, 2000). Therefore, we hypothesize that in sub-Saharan Africa, a high level of governance standards positively influences CSR reporting:
Company-Level Determinants
Regarding the company-level determinants, we investigate the impact of company nationality, size, degree of internationalization, and industry affiliation for two major reasons. First, two large meta-analyses of close to 200 studies (Fifka, 2013; Hahn & Kühnen, 2013) have demonstrated that these factors have a strong impact on CSR reporting. Another determinant that has frequently been examined is profitability, but the results are incoherent regarding its impact on reporting. Second, information availability has been another key factor for our selection, as we could obtain data for all companies for the four variables chosen. This would not have been possible for other company-level variables that have been studied before, such as the age of assets, ownership structure, corporate governance systems, and media exposure (for an overview, see Fifka, 2013). Nevertheless, these factors might be of potential influence, which posits a limitation to our study.
Nationality is among the most researched company-level determinants, and there is a seemingly endless list of studies that demonstrated an influence of nationality on CSR reporting (e.g., Adams, Hill, & Roberts, 1998; Baughn et al., 2007; Buhr & Freedman, 2001; Chapple & Moon, 2005; Gjolberg, 2009; Kolk, 2003; Kolk et al., 2001). The meta-analysis by Fifka (2013) identified 35 studies that evidenced such impact. Likewise, Fortanier, Kolk, and Pinkse (2011, p. 665) pointed to a “strong country-of-origin effect” which previous studies demonstrated for the CSR reporting of multinational companies (MNCs). However, Fortanier et al.’s (2011) investigation showed a convergence in reporting across countries due to the increasing application of global standards. Country of origin in this case refers to company nationality, which shall be illustrated by an example. 7up, a company of U.S. nationality, is also listed on the Nigeria Stock Exchange. Thus, we consider it as a foreign company in our study, although it owns a subsidiary in Nigeria. As these companies of Western origin are more likely to face more scrutiny from their international stakeholders and are more familiar with CSR disclosure (Leppelt et al., 2013), we assume that they are also more likely to report on their CSR initiatives. We thus hypothesize that European or North American company nationality positively affects the CSR reporting of the sub-Saharan African sample companies and formulate Hypothesis 2a as follows:
Aside from nationality, company size is another organizational variable that was frequently demonstrated to influence CSR reporting. A large number of empirical studies evidenced that CSR reporting positively correlates with company size (Branco & Rodrigues, 2008; Cullen & Christopher, 2002; Haniffa & Cooke, 2005; Hörisch, Johnson, & Schaltegger, 2014; Reverte, 2009; Yao, Wang, & Song, 2011). Legitimacy theory, in particular Watts and Zimmermann’s (1986) political cost hypothesis, argues that larger companies face more public scrutiny than smaller ones. As larger companies possess higher market power, they are more likely to encounter stakeholder pressures expressed through public resentment, consumer boycotts, and government regulation (Stiglbauer, Kühn, & Häußinger, 2014). Moreover, as implementing CSR and CSR instruments such as reporting requires substantial knowledge and resources (Siltaoja, 2014), larger companies tend to disclose more CSR information, because they have more resources and capabilities than small and medium-sized enterprises (SMEs) to do so (Embong, Mohd-Saleh, & Hassan, 2012). Based on these considerations and previous findings, we provide Hypothesis 2b:
Industry affiliation is another company determinant that has been shown to influence CSR reporting by many studies (Adams et al., 1998; Gray et al., 1995; Kolk, 2003; Kolk et al., 2001; Reverte, 2009; for an overview, see Fifka, 2013). The findings from these studies prove the existence of structural differences between sectors, which was also pointed out by earlier theoretical (Porter, 1979; Rumelt, 1991) and empirical works (Pedersen & Thomsen, 1998; Fairfield, Ramnath, & Yohn, 2009). These differences affect companies’ CSR reporting (Aerts, Cormier, & Magnan, 2006; Husillos-Carqués, Gonzalez, & Alvarez Gil, 2011; Parsa & Kouhy, 2008; Sotorrío & Sánchez, 2010; Wanderley et al., 2008). Moreover, most studies confirm that companies operating in “polluting sectors,” such as the mining, oil, and chemical industries, publish more CSR-related information than those from the finance and service industries, because they find themselves under more public scrutiny (Brammer & Pavelin, 2006; Jenkins & Yakovleva, 2006; Line et al., 2002; Patten, 1991, 2002). To test whether this also holds true for our sub-Saharan sample companies, we formulate Hypothesis 2c:
Based on the assumption that globalization fosters the implementation of CSR (Chapple & Moon, 2005), we argue that the very nature of internationalization positively affects CSR. First, MNCs apply CSR as a means to establish their reputation as a good corporate citizen in the host country (Esen, 2013). Second, MNCs use CSR to encounter scrutiny from their international stakeholders concerning the establishment of uncritical supply chain relations with local suppliers (Leppelt et al., 2013). Third, MNCs regard CSR disclosure as a good channel to communicate with international global business watchdogs (including NGOs and supranational organizations such as the Organisation for Economic Co-Operation and Development [OECD] and the World Bank; MacIntosh et al., 2004). Empirical studies further evidence that the international listing status represents a significant determinant of CSR reporting (Cooke, 1989; Hossain, Perera, & Rahman, 1995; Hossain, Tan, & Adams, 1994; Reverte, 2009; Robb, Single, & Zarzeski, 2001). In our study, we define international companies or MNCs to include those companies listed on foreign stock exchanges. By contrast, domestic companies refer to those companies only listed on the domestic stock exchange. Overall, we hypothesize that MNCs listed on foreign exchanges are more likely to report on CSR than domestic companies that are exclusively listed within their respective home country. These arguments lead us to Hypothesis 2d:
Method
Selection of Countries
With the aim of drawing a comprehensive and representative picture of the status quo of CSR reporting in sub-Saharan Africa, we selected seven countries - Nigeria, Kenya, Botswana, Ghana, Tanzania, Uganda, and Zambia - for various reasons. They have in common that they face severe long-term challenges that call for CSR initiatives: widespread poverty, inadequate physical and social infrastructure, limited nonagricultural economic development, insufficient access to education, and environmental degradation. Moreover, the countries chosen also have a common history. All of them were colonies of the British Empire, with Botswana and Tanzania formerly also being a German protectorate for a short period of time. Aside from a joint institutional heritage due to British colonial rule, this legacy is of practical relevance for our study. In all countries, corporate communications are provided extensively in English, which serve as the basis of our research.
Data gathering from website analysis seemed feasible for the seven countries in question, as their stock markets work relatively well with a considerable number of listed companies. Other stock markets in the region must be judged as nonoperational, due to a very limited number of listed companies: Cameroon (2 companies), Rwanda (4), Malawi (8), and Swaziland (10). Thus, we excluded these countries. Listing is relevant for our approach, because research suggests that listed companies are better reporters (Haddock, 2005; Haniffa & Cooke, 2005), which makes the provision of relevant company information needed to determine the status quo of CSR more likely. Second, legitimacy theory assumes that reporting is more prominent in big companies, because they face more political and public pressure than small companies (Watts & Zimmermann, 1986) as well as higher analyst coverage (Brennan, Jegadeesh, & Swaminathan, 1993). As empirical studies have shown (Embong et al., 2012; Siltaoja, 2014), bigger companies tend to disclose more CSR-related information than SMEs do, because they have more resources, substantial knowledge, and experience available to do so. This enabled us to collect the relevant data for the full sample. Moreover, the focus on big listed companies facilitates controlling for the effect of size on CSR reporting (Kolk, 2003; Owen, 2007).
Although all countries are suffering from an unfavorable economic environment and weak institutions (Alence, 2004; Visser, 2006b), they have recently shown differences in economic development. Nigeria and Kenya represent two large and strong economies in sub-Saharan Africa (with GDP growth rates of 6.6% and 4.6% in 2012, respectively), and Ghana, Zambia, and Tanzania also experience a promising economic development (7.9%, 7.3%, 6.9%, respectively). In contrast, Uganda and Botswana are smaller countries with comparatively weaker economic development (3.4% and 3.7%, respectively; the World Bank, 2013). As it is a major aim of our study to examine the impact of economic development on the institutionalization of CSR, we were purposefully looking for countries of the same region with different levels of GDP.
Regarding the political environment, no country currently has to face large-scale violent conflicts or civil wars, which might have distorted the results, although religious and ethnic strives are common. Moreover, the political institutions have undergone very different development processes in the countries in question. Botswana has been a stable democracy since its independence in 1966. By contrast, Nigeria and Ghana had long periods of civil wars and military rule (1960s to 1980s), followed by times of fragile democracy (since the 1990s). Although Uganda and Tanzania have democratic governments, they represent rather authoritarian one-party states. Kenya, Ghana, and Zambia, in contrast, were able to re-instate multi-party democracies in the early 1990s.
The different levels of political stability are reflected by the corruption perceptions index 2012 and the human development index (HDI). The corruption perceptions index measures the perceived levels of public-sector corruption in 176 countries around the world. Whereas Botswana ranked highest on position 30 with a score of 65, all other countries did not achieve more than 50 out of 100 points (Table 1). The maximum score is 100 points, indicating that a country is fully free from corruption. Countries scoring lower than 30—Kenya and Nigeria in our case—must be regarded as highly corrupt (Transparency International, 2012).
Socio-Economic and Political Determinants.
Note. FDI = foreign direct investment.
A higher number indicates a lower level of corruption. The maximum score is 100.
A higher number indicates a higher social development. The maximum score is 1.
The HDI in turn is a measure of social development that includes three basic dimensions, namely health, education, and a decent standard of living in a composite measure, whose maximum value is 1. As Table 1 shows, the seven countries of our study have relatively similar human development levels. However, Botswana (0.634) and Ghana (0.558) can be considered medium human development countries, as their level is above 0.5. All other countries belong to the low human development countries (UNDP, 2013).
Sample
We selected the seven countries described above also because of their functioning capital markets in comparison with other sub-Saharan countries. Numerous studies showed that listing on the stock market has a positive effect on reporting CSR-related information (Haddock, 2005; Hossain et al., 1995; Llena et al., 2007; Stanny & Ely, 2008). This effect has also been evidenced for developing countries (Haniffa & Cooke, 2005), and even for Ghana as one of our countries of study in specific (Rahaman, 2000).
Thus, we included all companies listed on the Nigeria Stock Exchange, the Nairobi Securities Exchange, the Botswana Stock Exchange, the Ghana Stock Exchange, the Dar es Salaam Stock Exchange Tanzania, the Uganda Securities Exchange, and the Lusaka Stock Exchange Zambia, as of November 1, 2013. This created an initial sample of 232 companies. Due to the difficult availability of in-print information, we referred to the companies’ websites for data collection. As 21 websites were either under construction or not accessible, the final sample comprised 211 companies.
Research Approach and Data Collection
To determine the extent and content of CSR reporting, we conducted a directed content analysis based on research categories previously used by Chapple and Moon (2005) with the aim of applying these categories to sub-Saharan Africa. In particular, we investigated the websites for the following categories: (a) existence of a CSR section on the company website, (b) extent of CSR reporting, (c) implementation of CSR, (d) types of CSR, and (e) dimensions and channels of CSR. By doing so, we added the research categories “implementation of CSR” and “types of CSR” to Chapple and Moon’s (2005) research categories to provide a more encompassing picture on the nature of CSR in sub-Saharan Africa and how it is implemented. In addition, we included “CSR wording” in the category on the existence of a CSR section on the web page, in which we examine the terms companies use to describe their CSR approach. Finally, we slightly modified the channels of CSR, which describe how companies engage in CSR by adding “CSR partnerships” as an increasingly popular form to the five channels identified by Chapple and Moon (2005) that are explained below. We will now describe the individual categories in detail.
The first research category scrutinized the presence of information related to CSR on the companies’ websites. After locating each company’s website, we checked it for CSR information. For doing so, we applied a broad definition of CSR, as a narrow definition might have led to an exclusion of relevant information. Thus, we considered all statements that listed elements such as CSR, social responsibility, sustainability, corporate citizenship, social investment, philanthropy, and any information related to the triple bottom line (Elkington, 1998), namely the economic, environmental, or social dimension. To know which CSR wording companies use most frequently, we collected all respective keywords mentioned on the websites.
In case, the companies reported on CSR, we considered the second research category, namely the extent of CSR reporting, to determine the degree of attention paid to CSR reporting. Based on previous studies (Chapple & Moon, 2005; Chaudhri & Wang, 2007; Wanderley et al., 2008), we distinguished between three categories: minimal CSR reporting (1 to 2 pages), medium CSR reporting (3 to 10 pages), and extensive CSR reporting (more than 10 pages).
In the third research category, we examined which instruments of CSR implementation the sample companies are reporting to use. Based on Wanderley et al. (2008), we examined whether a separate CSR report, the publication of a code of ethics, a code of conduct, a separate section dedicated to environmental responsibility, and a specific section on stakeholder relations were used to implement CSR.
The fourth research category comprises the types of CSR, which are displayed by the companies in their CSR reporting. The underlying concept was developed by the United Nations (UN; 2007) for developing countries and differentiates between three mutually exclusive types of CSR in business practice: minimum, median, and maximum CSR standard. Whereas the minimum CSR standard refers to the prevention of harm for the sake of fulfilling legal obligations, the median CSR standard implies a positive contribution “to sustainable development by addressing company’s social and environmental impacts, and potentially also through social or community investment” (UN, 2007, p. 1). The maximum CSR standard in turn promotes “the active alignment of internal business goals with externally set societal goals (those that support sustainable development)” (UN, 2007, p. 1).
Finally, to analyze the focus areas of CSR, the fifth research category differentiates between three dimensions of CSR reported: community involvement, socially responsible production processes, and socially responsible employee relations (Chambers et al., 2003; Chapple & Moon, 2005; Moon, 2002). The community involvement dimension refers to philanthropic CSR activities that are mostly separate from the company’s core business: agriculture, infrastructure and local economic development, community development, arts and culture, religion, education and training, youth and children, sports, welfare (including poverty and emergency relief), health and disability, housing, environment, and conservation. The production processes dimension and the employee relations dimension correspond to the maximum standard of CSR, because these two dimensions are concerned with the strategic link between the company’s main business activities and its CSR initiatives. Therefore, the production processes dimension concentrates on companies’ compliance with socially responsible business conduct in their supply chains and in their on-site operations. More precisely, this type of compliance comprises environmental issues, health and safety, human resources, and ethical aspects. The employee relations dimension includes employee welfare, compliance with labor rights, and employee codetermination. Although the classification of CSR activities into these three dimensions is neither mutually exclusive, nor collectively exhaustive, it helps to differentiate between the three broad types of CSR activities (Chapple & Moon, 2005). With the intention of classifying the different CSR activities into philanthropic and sponsorship initiatives and those more institutionalized in the business, such as CSR partnerships, employee volunteering programs, and foundations, we also examined the channels to operationalize CSR as a part of the fifth research category.
To obtain the information on these five research categories, we applied a directed content analysis including the websites of the 211 sample companies. The limitation of this research approach is twofold. On the one side, companies—to improve their image—might actually report CSR-related activity that they have not undertaken (Husted & Allen, 2006; Logsdon & Wood, 2005). While the investigative work by auditors, journalists, and NGOs who seek to uncover such alteration of information reduces the chance of misleading disclosure by companies, there is a certain chance of deceptive CSR reporting (Bell DeTienne & Lewis, 2005; Debeljak, Krkac, & Busljeta Banks, 2011; Stoll, 2002). Especially in regions such as sub-Saharan Africa with weak media and civil society institutions (Alence, 2004; Visser, 2006b), external control mechanisms are rather underdeveloped (Jackson & Rosberg, 1982). Indeed, studies show that CSR disclosure might not reflect actual CSR practice (Adams, 2004; Font, Walmsley, Cogotti, McCombes, & Häusler, 2012; Kolk & Van Tulder, 2006). On the other side, companies might undertake CSR activities that they do not report (Browne & Nuttall, 2013). Habisch, Patelli, Pedrini, and Schwartz (2011) pointed out that this limitation is not worrisome, as companies have a strong self-interest to report on what they actually do. However, some companies might undertake CSR activities, but lack the means or knowledge of how to report, and thus, actual practice remains undisclosed.
Despite these potential limitations, many studies have relied on content analysis to examine CSR practices and CSR reporting (e.g., Chen & Bouvain, 2009; Fifka & Pobizhan, 2014; Habisch et al., 2011; Holder-Webb, Cohen, Nath, & Wood, 2009; Steurer & Konrad, 2009). Also in the African context, this research methodology of using CSR reporting as a proxy for CSR practice has been applied before (Amaeshi et al., 2006; Dawkins & Ngunjiri, 2008; Wahba, 2008).
To guarantee the validity and reliability of the research results, the following two assumptions hold: (a) All websites are independent of each other, and (b) the research categories are coded into a nominal scale (Stemler, 2001). By verifying the presence or absence of each research category, two independent coders applied a binary scoring method. In particular, a 1 was assigned for each category observed to be present on the examined websites, and a 0 for each category judged to be absent (M. J. Campbell & Swinscow, 2009). The intercoder reliability measured in Cohen’s κ and Krippendorff’s α ranged between .731 and 1 for each research criteria among all seven countries. These results indicate a good (0.75 > κ > 0.6) to very high/excellent (κ > 0.75) consistency of the results (Bakeman & Gottman, 1997; Milne & Adler, 1999). Finally, we moderated the results of the two coders and aggregated the results at the individual issue level and the overall level.
Modeling
The five research categories of CSR that we developed serve two purposes: to qualitatively describe the contents disclosed by companies in sub-Saharan Africa, and to quantitatively examine the impact of country-level as well as company-level determinants on CSR reporting. In this quantitative analysis, we refer to the following CSR categories as dependent variables: the dichotomous variable existence of CSR section on the website (CSRsec), and the five dichotomous variables related to CSR implementation (CSRimpl), namely CSR report (CSRrep), code of ethics (CoE), code of conduct (CoC), section on the environment (ENV), and section on stakeholder relations (STAKE).
The three country-level determinants previously discussed—GDP, corruption, and life expectancy as part of the HDI—represent the country independent variables in our quantitative analysis (see Table 2). The first indicator refers to economic development. To measure the level of wealth of the respective countries, we opted for GDP. We could not include foreign direct investment (FDI) as a measure of wealth, as it resulted in strong correlation with GDP and corruption. We logarithmized the variable GDP to reduce skewness and kurtosis (Backhaus, Erichson, Plinke, & Weiber, 2011) and thus obtained lnGDP. To turn the second latent indicator, namely social development, into a measurable construct, we considered life expectancy (LIFE). We chose this proxy, because we assume that life expectancy is positively related to physical well-being (Chapple & Moon, 2005). We could not include adult literacy as a measure of political participation or the HDI, because both too strongly correlate with GDP. The third indicator considers the level of governance standards as a driver for the institutionalization of CSR. To operationalize the level of governance standards, we apply the corruption perceptions index (CORR). We could not consider other proxies, such as the Bribe Payers Index and the Opacity Index, because they do not cover sub-Saharan countries in their measure. The socio-economic development indicators GDP and life expectancy were obtained from the World Bank (2013), and the corruption perceptions index from Transparency International (2012).
Descriptive Statistics for the Dependent and Independent Variables.
Note. CSRsec = CSR section; CSRrep = CSR report; CoE = code of ethics; CoC = code of conduct; ENV = section on the environment; STAKE = section on stakeholders; lnGDP = gross domestic product; LIFE = life expectancy; CORR = corruption; N_EU/NA = company’s nationality Europe/North America; lnSIZE = company size; IND = industry sector dummies for (financial) services vs. manufacturing; ISTOCK = degree of internationalization.
Regarding the measurement of the four independent company-level determinants (see Table 2), we first identified the nationality of the sample companies and classified them into three groups: African continent nationality (N_AFR), European continent nationality (N_EU), and Northern American nationality (N_NA). However, African nationality was omitted in the regression models, because this dummy represented the control dummy. Second, to investigate whether company size (SIZE) positively affects CSR reporting, we operationalized company size using the number of employees mentioned in the annual reports. We also logarithmized the number of employees and obtained lnSIZE. Third, we created dummy variables to operationalize the industry affiliation (IND): Whereas the dummy for the service sector includes all companies operating in the service and financial service sectors, the dummy for the manufacturing sector comprises companies from manufacturing, utilities, and exploitative industries. Fourth, we determined the degree of internationalization (ISTOCK) by identifying whether the company is listed on foreign stock markets or only on the domestic stock exchange, as previous studies did before us (Hossain et al., 1994, 1995; Reverte, 2009; Robb et al., 2001).
To test the two research hypotheses, we conducted regression analyses and investigated to what extent the individual independent variables affected companies’ CSR reporting. As our dependent variables are dichotomous and therefore S-shaped, we could not apply linear regression models. Furthermore, as the assumptions of normal distribution and homoscedasticity of the residuals were violated (Backhaus et al., 2011), we chose to perform logistic regression analyses. According to the Eicker–Huber–White robust treatment of errors, we clustered the standard errors according to the seven sample countries. In doing so, we sought to keep the assumption of zero correlation across groups as with fixed effects, while also allowing any within-group correlation (Wooldridge, 2006).
By applying the rule of thumb of Friedrichs (2006), which states that a sample of 10 companies is appropriate per predictor variable, we kept the number of variables low to reduce the risk of multicollinearity between the regressor variables (Loderer & Martin, 1997). After operationalizing the dependent and independent variables, we developed the following logistic regression models to analyze the determinants of sub-Saharan African companies’ CSR reporting and CSR implementation:
where, the dependent variable CSRsec refers to existence of CSR section on the company website, and CSRimpl refers to reported implementation of CSR in the form of (a) CSR report, (b) code of ethics, (c) code of conduct, (d) section on the environment, and (e) section on stakeholders.
The independent variables are lnGDP, LIFE (life expectancy), CORR (corruption), N_EU/NA (company’s nationality Europe/North America), lnSIZE (company size), IND (industry sector dummy for [financial] services vs. manufacturing), ISTOCK: degree of internationalization.
Results
Descriptive Findings
The first major research aim of our study was to determine if there is (a) a CSR section on the company website, and (b) to what extent companies report on CSR. Moreover, we examined (c) what instruments of CSR implementation are mentioned on the website, (d) the types of CSR displayed, and (e) the different dimensions and channels of CSR that are conveyed. In the following, we discuss our findings regarding each of these categories.
As Table 3 shows, a relatively large number of the sample companies provide a CSR section on their website. Eighty percent of them do so, which is high in comparison with the results of other studies on developing countries (Chapple & Moon, 2005). We explain the relatively high number of CSR sections on corporate websites by the growing importance of the Internet and the increasing use of website communication as a business communication tool in recent years (Ashbaugh, Johnstone, & Warfield, 1999; Rowbottom & Lymer, 2009). Nevertheless, there are substantial differences across the countries in our sample. While 94% of the Tanzanian companies have a CSR section on their website, only 57% of the companies from Botswana do so. Although all countries score above 50%, the standard deviation of 0.1245 implies that there is no uniform sub-Saharan pattern of CSR reporting. The one-way ANOVA confirms this finding at a significance level of α = .012. In addition, the Scheffé tests reveal that the seven countries differ significantly with regard to the number of companies reporting on CSR on their websites, as there are no homogeneous subsets.
Existence and Extent of CSR Website Reporting, Reported CSR Implementation, CSR Types, and CSR Dimensions in Sub-Saharan African Countries (% of companies).
Note. CSR = corporate social responsibility
Worthwhile to notice is the fact that the countries vary not only with regard to the existence of CSR sections on their websites but also in the wording used to refer to their social and environmental initiatives. The term most frequently used by all sample companies is CSR, followed by the term sustainability. The companies from the countries with the highest number of CSR sections, namely Tanzania, Uganda, and Zambia, primarily refer to CSR. Companies from Botswana and Ghana, on the contrary, mainly use the term social responsibilities.
By examining the extent of the CSR reporting, we aim to contrast those companies that put relatively little emphasis on their CSR website reporting with those that devote medium to extensive attention to it. As Table 3 shows, the overall percentage of companies providing extensive CSR reporting is relatively low (from 14% to 33%). Moreover, one might assume that in countries where larger numbers of companies provide a CSR section, reporting would also be more extensive. However, the Spearman’s rank order correlation does not confirm any statistically significant relationship. Thus, there is no correlation between the number of companies reporting and the extent of reporting.
Our third category of investigation examines how companies report on the organizational implementation of CSR. This refers to the instruments companies use to implement CSR. As depicted in Table 3, only 18% of all sample companies issue a stand-alone CSR report, and even less (14%) provide information on how they handle stakeholder relations. While general ethical guidelines provided in a code of ethics are also not frequently used, codes of conduct providing specific behavioral guidelines are the only instrument used by more than one third of the companies. This finding can be interpreted as an attempt to hold employees accountable to appropriate behavior, especially against the background of widespread corruption.
Regarding the types of CSR, we applied the classification developed by the UN (2007) to examine how the sample companies display CSR in their reporting. Table 3 shows that the seven-country mean for companies displaying the minimum CSR standard is relatively low with 19%. This finding indicates that a relatively small proportion of the companies interpret CSR merely as an attempt to fulfill legal obligations and to prevent against harm. The median CSR standard in turn achieved the highest overall mean of 58%, implying that the majority of sub-Saharan companies perceive CSR as a possibility to make a positive contribution to their social and environmental context in the form of social or community investment. By aligning the core internal business activities with positive implications for the local economies, a mean of 23% of the seven countries pursue the maximum CSR standard. This finding reveals that about a quarter of the companies pursue strategic CSR, which seeks to benefit the company and the community in the long run. Reported examples of the maximum CSR standard we found among the sample companies include microfinance programs for suppliers, equity social institutional accounts, agriculture loans, knowledge transfer for improved crop yields, education and training opportunities for employees to be promoted or to become business owners, free housing close to company plant, consideration of environmental standards in R&D, ecotourism, and so on.
With respect to our last research category, we first examined the three dimensions of CSR: community involvement, socially responsible production processes, and socially responsible employee relations (Table 3). Across the seven countries, 88% to 100% of the companies are engaged in the community, which demonstrates that they put a very strong emphasis on reporting locally oriented involvement activities. This outcome confirms Moon’s (2002) and Chapple and Moon’s (2005) finding that community involvement is the dimension of CSR most often addressed. Employee relations (addressed by 61% on average) and responsible production processes (57%) are given considerably less attention, which underlines that CSR in sub-Saharan Africa is characterized by supporting the local community, as has also been shown by Visser (2006b). This finding is supported by examining the more specific CSR issues that are addressed on corporate websites (Table 4). In all countries, companies are focusing on the locations where they operate, with a variety of themes covered, such as education and training (with a seven-country mean of 80%), health and disability (68%), community development (63%), youth and children (62%), as well as the environment (61%). The issues least frequently mentioned comprise arts and culture (18%), and religion (4%). Moreover, many companies discuss safety issues related to employees on production processes, though environmental safety is also of concern to the sample companies. Concerning the employee dimension, an issue frequently mentioned is employee welfare, indicating that companies claim to pay a sufficient wage and to provide housing for their workers.
Dimensions, Issues, and Channels of CSR.
Note. CI = community involvement; PP = production processes; ER = employee relations.
Due to the emphasis placed on community involvement, philanthropy is the major channel through which CSR is operationalized in all countries of investigation. This is in accordance with Visser’s (2006b) conceptual revision of Carroll’s CSR pyramid for Africa. In the dimensions of employee relations and production processes, companies frequently operate with codes prescribing specific behavior on the side of the employees, but also on the side of the company. To conclude our section on the descriptive findings, we now present a short CSR profile of each country (see Table 4).
Tanzania is the country with the highest number of companies providing a website section on CSR, the highest coverage of the employee relations dimension and the second highest coverage of the production processes dimension. Its community involvement comprises education and training, youth and children, community development, and health and disability. Whereas the production processes dimension primarily concentrates on the environment and human resources, the employee relations dimension mainly refers to employee welfare. The CSR modes deployed by Tanzanian companies offer a wide range from philanthropy, foundations, and sponsorships to CSR partnerships with companies and NGOs, CSR codes, and employee volunteering.
Uganda, the country with the second highest number of companies providing CSR sections on their website, also engages in education and training, community development, welfare, and health and disability. The reporting on production processes is relatively low and focuses mainly on environmental issues. Ugandan companies provide the third highest coverage with respect to the employee relations dimension and are primarily concerned with employee welfare. With the help of CSR partnerships with other companies and NGOs, CSR codes, philanthropy, sponsorships, and employee volunteering, the Ugandan companies are able to implement their CSR approach successfully at various corporate levels.
Although Zambia, Nigeria, and Kenya show similar levels of CSR communication and address similar issues, their CSR modes differ. The main community issues for the three countries are education and training followed by health and disability. Whereas Zambian and Kenyan companies also consider the environment, Nigerian companies address infrastructure and local economic development. One explanation for why these countries seek to tackle health and disability is the fact that HIV/AIDS represents a serious health and development issue for them. All three countries concentrate on the environment and human resources issues with respect to production processes. Nigeria and Zambia mainly address employee welfare in their CSR employee relations dimension, and Kenya emphasizes employee engagement. Beyond philanthropic activities, the three countries frequently apply CSR partnerships with other companies and NGOs, as well as CSR codes. Several Nigerian companies have established foundations, while Kenyan companies frequently run employee volunteer programs.
Ghana and Botswana are the two countries where the number of companies reporting is the lowest. Their community involvement embraces education and training, and community development. Companies in both countries attempt to tackle environment and human resource issues arising in the production process, and try to improve employee welfare in the employee dimension. However, the CSR modes displayed differ substantially. Whereas companies from Botswana use philanthropic initiatives, CSR partnerships with NGOs, and codes of conduct, companies from Ghana additionally implement sponsorships, foundations, and CSR partnerships with companies and NGOs.
Overall, the CSR issues and modes deployed vary among the sample countries. With respect to the CSR modes, it is evident that the CSR mode most often applied is philanthropy, which does not represent a systematic implementation of CSR in the companies’ core business. This finding also corroborates Chapple and Moon’s (2005) as well as Welford’s (2004) finding that CSR in developing countries is primarily of charitable nature. Companies in Tanzania and Uganda not only show the largest number of reporting companies, they also employ systematic modes of CSR most frequently, including foundations, CSR partnerships, codes of conduct, employee relations codes, and employee volunteering to enable knowledge transfer. All companies deploy CSR codes to ensure good governance practices and the consideration of human rights and ethics in internal and external supply chains.
Empirical Findings
To evaluate to what degree the differences in the descriptive findings can be attributed to differences in the socio-economic and political environments of the countries in our sample, we conducted logistic regression analyses to determine the impact of country- and company-level determinants on CSR reporting.
Therefore, we first estimated the impact of the country-level determinants ((a) GDP, (b) life expectancy, (c) and corruption), and the company-level determinants ((d) company nationality, (e) company size, (f) industry affiliation, and (g) degree of internationalization) on the CSR section on the website (CSRsec). Second, we ran the same logistic regression model but calculated the impact on the five dependent variables, namely instruments of CSR implementation (CSRimpl) reported. With respect to the descriptive statistics, the dispersion of most independent variables is on an acceptable level. As summarized in Table 5, no binary Pearson correlation for the independent variables exceeds the value of 0.5 (Cheng, Luckett, & Mahama, 2007; Hair, Black, Babin, & Anderson, 2010; Peng & Beamish, 2008). As all variance inflation factors (VIFs) are below the critical value of 5, and all tolerance values are above 0.25 (Hair et al., 2010; Menard, 2002; Roberts, 1992), we can exclude multicollinearity to be a problem in our study.
Correlation Coefficients Among Independent Variables.
Note. lnGDP; LIFE = life expectancy; CORR = corruption; N_EU/NA = company’s nationality Europe/North America; lnSIZE = company size; IND = industry sector dummies for (financial) services vs. manufacturing; ISTOCK = degree of internationalization.
Significant at the .1 level (two-tailed). **Significant at the .05 level. ***Significant at the .01 level.
As depicted in Table 6, we ran two separate hierarchical logistic regression analyses to assess the relative variance of the country- and company-level determinants and to explain the likelihood of the various dummies for CSR implementation. Column (A) depicts the findings of the logistic regression model for CSR section, and columns (B) to (F) for CSR implementation, namely (B) separate CSR report, (C) existence of code of ethics, (D) existence of code of conduct, (E) section on the environment, and (F) section on stakeholder relations on corporate website. From the hierarchical regression models, it becomes evident that McFadden’s R2 (1973) for the country-level determinants only amount to 1.7% to 6.75%. These low figures indicate that the country determinants do not explain the proportion of total variation in the reported implementation of CSR convincingly (Draper & Smith, 1998). By contrast, McFadden’s R2 for the company-level determinants assume values between 13.85% and 27.86%, which are considerably higher than the country-level determinants. Hence, we derive that the company-level determinants play a more dominant role in assessing the implementation of CSR than the country-level determinants. The regression models that comprise both country- and company-level determinants finally yield McFadden’s R2 between 18.35% and 30.28%. Since all McFadden R2 are close to or above the threshold of 20%, we assure the goodness of fit of our logistic regression models (Backhaus et al., 2011).
Hierarchical Logistic Regression Models.
Note. The dependent variables: Column (A) contains the findings of the logistic regression using CSRsec (CSR section: provision of CSR information on website, yes or no); column (B) shows the results of the logistic regression for CSRrep (CSR report: existence of separate CSR report, yes or no); column (C) encompasses the findings of the logistic regression for CoE (code of ethics: publication of a code of ethics on website, yes or no); column (D) comprises the results of the logistic regression for CoC (code of conduct: publication of a code of conduct on website, yes or no); column (E) illustrates the findings of the logistic regression for ENV (ENV: section on the environment on website, yes or no); column (F) refers to the results of the logistic regression for STAKE (STAKE: section on stakeholders on website, yes or no). The independent variables: lnGDP; LIFE = life expectancy; CORR = corruption; N_EU/NA = company’s nationality Europe/North America; lnSIZE = company size; IND = industry sector dummies for (financial) services vs. manufacturing; ISTOCK = degree of internationalization.
Significant at the .1 level (two-tailed). **Significant at the .05 level. ***Significant at the .01 level.
We also conducted a series of multivariate multiple probit models to estimate several correlated binary outcomes jointly instead of looking at them separately. As the multivariate multiple probit models revealed very similar results, we implemented them as a robustness check of our findings. However, in the following, we only concentrate on the description of the findings of the logistic regression models that consider both country- and company-level determinants simultaneously.
With respect to the country-level determinants (H1a to H1c), we investigated whether the level of economic and social development, and the national level of governance standards affect CSR reporting by the sample companies in the seven sub-Saharan countries.
The findings depicted in Table 6 mostly approve our first research hypothesis. With respect to H1a, we can confirm that the sample countries’ level of wealth (expressed in GDP) significantly positively affects the existence of a CSR section (at α = .017), the publication of a code of ethics (at α = .024), and the existence of a separate section on the environment (at α = .016). As the assumption that higher levels of wealth (measured in GDP) enable companies to invest more resources in CSR holds true for three of the six dependent variables, we partly confirm H1a.
H1b tested whether social development (measured in life expectancy) affects sub-Saharan companies’ CSR reporting. From the regressions, it becomes evident that life expectancy has a highly significant positive bearing on the publication of a code of ethics (at α = .006). Although the direction of the effect is always positive, the remaining dependent variables do not yield significant results. Hence, we cannot approve that higher levels of social development lead to more CSR implementation and reject H1b.
Regarding the governance level, the logistic regressions evidence that the level of corruption as a negative expression of good country governance is highly significantly associated with the existence of a CSR section (at α = .000) and the instruments used for implementing CSR (at α < .007), except for the publication of a CSR report and a section on stakeholders. In accordance with our assumption, this relationship is negative. Hence, there is statistical evidence that higher levels of governance standards positively affect CSR reporting. We thus accept H1c.
To determine whether company-level determinants have an impact on CSR reporting in sub-Saharan Africa (H2a to H2d), we further analyzed whether the companies’ nationality, size, industry affiliation, and degree of internationalization influence the provision of CSR information and the reported implementation of CSR. Whereas European company nationality is significantly positively associated with the provision of a CSR section and publication of a CSR report (at α < .002), the North American nationality variable generates significant positive results for the disclosed implementation of a CSR report, code of conduct, and section on the environment (at α < .027). With respect to H2a, we can thus confirm that companies’ European or North American nationality positively affects the provision of a CSR section on the website and the reported implementation of CSR.
Consistent with H2b, the regression analyses evidence that there is a highly significant, positive impact of company size (measured in number of employees) on CSR reporting (at α < .009), except for the section dedicated to stakeholders. Hence, the bigger a company in size the more likely it provides a section on CSR and its implementation. Based on this finding, we accept H2b and also confirm Watt and Zimmermann’s (1986) political cost theory.
In accordance with previous studies (Brammer & Pavelin, 2006; Jenkins & Yakovleva, 2006; Line et al., 2002), our findings also approve the existence of a significant systematic variation across industry sectors with respect to the dependent variables, as stated in H2c. More precisely, those companies operating in the financial services and services sector show significant lower levels of CSR reporting on the website and the five types of CSR implementation. Hence, membership in these sectors leads to lower levels of reporting (at α < .070), except for the number of companies that report overall, and the publication of a CSR report. By contrast, affiliation to “polluting sectors,” such as manufacturing, utilities, and exploitative industries, is significantly positively associated with CSR reporting. Overall, we accept H2c by arguing that companies from “polluting sectors” more proactively disclose CSR-related information to meet stakeholder expectations and maintain their legitimacy.
Although H2d assumed that international companies tend to report CSR-related information more frequently on their websites, the findings reveal that this only holds true for the general provision of a CSR section (at α = .026) and the publication of a code of ethics (at α = .009). As the degree of internationalization does not influence the other four types of CSR implementation, we do not accept H2d.
Overall, we partly confirm the first group of hypotheses with respect to the country-level determinants of CSR by evidencing that economic development (expressed in GDP) and the level of governance standards (measured by the countries’ corruption level) have a considerable bearing on the sample companies’ CSR reporting. We also approve that company-level determinants have an impact on CSR reporting. We thus accept the second group of hypotheses, because there is evidence that company-level determinants, specifically nationality, company size, and membership to “polluting sectors,” influence CSR reporting by sub-Saharan African companies. Although McFadden’s R2 is on an acceptable level when considering country and company determinants simultaneously, we assume that there are additional country- and company-level determinants, which affect reporting.
Discussion and Conclusion
The aim of this study was to examine what contents sub-Saharan companies communicate in their CSR website reporting and to what degree country- and company-level determinants affect their reporting. As our analysis shows, both groups of determinants are of influence. Regarding country-level determinants, our study demonstrates that the socio-economic and political context is relevant for CSR reporting. There are significant differences among the sub-Saharan countries in our sample with respect to the extent of CSR disclosure and the contents reported. This contextual impact matches the findings of other studies on sub-Saharan Africa (Rivera-Santos et al., 2015; Visser, 2006a).
We can conclude from our findings that businesses in sub-Saharan Africa take on responsibilities that are strongly related to local communities, focusing on issues such as education and training, health and disability, community development, as well as the well-being of young people and children. Thus, we approve Schmidheiny’s (2006) assumption that in developing countries, more emphasis is placed on social issues than on environmental and ethical issues. The mode through which the social problems are addressed is primarily charitable in nature. Thus, CSR in our sample is more of a philanthropic than of a strategic nature, while a systematic implementation of CSR is lacking. This confirms the findings of Baskin (2006), Jamali and Mirshak (2007), Gugler and Shi (2009), and Visser (2006a, 2008) who also demonstrated that CSR in developing countries tends to be less formalized.
Despite the impact of country-level determinants, it becomes evident from the hierarchical regression models that their impact on CSR reporting is smaller than the impact of company-level determinants. Like other studies before us, we also find an effect of company nationality, size, and industry affiliation. Our results show that European and North American company nationality positively affects the disclosure of CSR information significantly. Moreover, larger companies and companies from “polluting sectors” display stronger efforts to report on CSR and its implementation. By contrast, smaller companies and companies operating in the services sector tend to report less frequently on CSR.
Limitations
Although our findings clearly demonstrate the impact of country- and company-level determinants, our study has some limitations. First, it must be pointed out that CSR reporting might deviate from actual CSR behavior, as companies might over- or understate CSR-related information. This potential disconnection between reporting and actual CSR practice has been pointed out by some studies (Adams, 2004; Font et al., 2012; Kolk & Van Tulder, 2006). Second, the content analysis we conducted was based only on website reporting and did not consider other media used for CSR communication, such as annual reports, nonfinancial reports, and other potential media. An enlargement of the media analyzed could provide a broader base of information. Third, our sample comprised 211 companies listed on stock exchanges, representing some of the biggest companies in their respective countries, which are powerful market actors. It could thus be criticized that the study is not representative for reporting in general, as it does not include disclosure by SMEs. This limitation, however, is also inherent to most studies on CSR, as they have focused on large companies (Chapple & Moon, 2005; Chen & Bouvain, 2009; Kolk, 2008; Maignan & Ralston, 2002). In turn, research on CSR reporting by SMEs is scarce until today, as Fifka (2013) concluded from his meta-analysis. Fourth, there are some limitations with respect to the operationalization of our independent variables. Despite the existence of more precise measures to assess the degree of internationalization, such as the ratio of foreign to total assets or foreign to total sales (Kolk & Fontanier, 2013), we chose to operationalize internationalization by verifying whether a company is listed on an international stock exchange or not, because we could not obtain the required financial figures of the sample companies. With respect to the industry sector, we only distinguished between service and manufacturing companies and did not include further subsectors, such as automotive, chemicals and pharmaceuticals, electronics, oil, and so on. A more differentiated examination of the degree of internationalization and industry affiliation could create further insights.
Contributions
Despite these limitations, the findings of our study contribute to the literature in several ways. First of all, they demonstrate that the examination of differences in CSR across countries or regions should consider country- as well as company-level determinants. Both groups of factors are complementary as well as influential. Due to the impact of contextual variables, Western understandings of CSR cannot simply be applied to developing regions, such as sub-Saharan Africa. As our study has shown, the notion of CSR displayed by sub-Saharan African companies strongly focuses on philanthropic engagement in the local community, whereas environmental and broader social issues are of lesser importance. These findings have implications for future research on the determinants of CSR and CSR reporting, as the impact of the contextual environment has not been investigated precisely so far. While many studies in the past found differences in CSR and CSR reporting across countries, they simply attributed these differences to the differing socio-economic and political environments (Jones, 1999; Whitley, 1999), without undertaking a statistical analysis of specific determinants that represent the contextual environment (Baskin, 2006; Hartman, Rubin, & Dhanda, 2007; Welford, 2004). We can conclude from our study that such an examination of context-specific determinants is necessary to assess their impact on CSR and CSR reporting. Our investigation further contributes to CSR research on a geographic area that has only received little attention so far, in spite of the role that CSR could potentially play for the socio-economic development in countries, where governments are traditionally weak and often not able to contribute to social and economic progress (Alence, 2004; Visser, 2006b).
Avenues for Future Research
This last aspect provides one of several opportunities for future research. It would be worthwhile to attempt an examination of how companies in sub-Saharan Africa make an impact on socio-economic development through CSR. Although CSR impact analysis is difficult—Gjolberg (2009, p. 10) even spoke of “measuring the immeasurable”—because of the many determinants that influence development, a survey of stakeholders on how they perceive the impact of companies’ CSR efforts could present a potential research method. This would also allow better conclusions on judging the effectiveness of the measures that are reported by the companies. Likewise, surveys or interviews of managers would present a viable method to study the motives for why sub-Saharan companies disclose CSR and what difficulties they see in the implementation of a more encompassing reporting.
Moreover, the study of CSR reporting in sub-Saharan Africa should be expanded to SMEs to provide a more complete picture. Although it must be assumed that only little disclosure is undertaken, it would be especially interesting to see what types of SMEs report and what kind of CSR information they disclose. In this context, it would be worthwhile to look at website reporting, as it is less costly than the production and publication of stand-alone reports. Thus, chances that SMEs will report on CSR on their website than in a printed stand-alone report seem to be much higher.
Finally, our study on the impact of socio-economic and political determinants on CSR reporting is only a starting point for avenues of future research into the contextual setting of CSR in developing regions. The investigation of further determinants that represent the contextual environment could provide valuable insights into drivers and impediments for CSR in the respective countries. From these findings, implications on how business and political actors can promote the socio-economic development role of CSR could be drawn.
Footnotes
Acknowledgements
We are especially grateful to the editors, Ans Kolk and Miguel Rivera-Santos, for their constructive and continuous input, and also would like to thank the anonymous reviewers for their comments and suggestions on how to improve the manuscript.
The article was accepted during the editorship of Duane Windsor.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.