The Influences of Home Country Institutional
Quality and Internationalization on Firm Performance
in Emerging Markets: Is CSR the Missing Link?
A quantitative study on Latin American firms
Author: Mirjam Terlaak Poot Student number: s2021587 Email: m.terlaak.poot@student.rug.nl
Word count: 11,997 words Supervisor: dr. R.W. de Vries Co-supervisor: dr. C.H. Slager
Faculty of Economics and Business University of Groningen
Duisenberg Building, Nettelbosje 2, 9747 AE Groningen, The Netherlands P.O. Box 800, 9700 AV Groningen, The Netherlands
http://www.rug.nl/feb
i
ABSTRACT
Despite the growing importance of emerging markets in the global economy, little is known about the influences of home country institutional quality, firm internationalization and corporate social responsibility (CSR) on the financial performance of firms from Latin America. Based on institutional theory, this thesis develops a theoretical framework that proposes that CSR forms a partially mediating mechanism in the impacts of quality of institutions and international expansion on financial performance. It empirically tests the relationships among 159 firms from Argentina, Brazil, Chile, Colombia and Mexico by conducting ordinary least squares regressions. Although the raised hypotheses are not supported, of which likely explanations are data related, this thesis adds to existing literature by providing evidence for a positive and significant effect of institutional quality on economic performance on the Latin American firm-level.
ii
ACKNOWLEDGEMENTS
iii
TABLE OF CONTENTS
Abstract ... i
Acknowledgements ... ii
Table of Contents ... iii
List of Figures ... iv List of Tables ... iv List of Acronyms ... v 1. Introduction ... 1 2. Literature Review ... 4 2.1 Institutional Theory ... 4
2.2 CSR, Legitimacy and Institutional Quality in Emerging Markets ... 6
2.3 Internationalization of EMNEs and CSR ... 8
2.4 CSR and Financial Firm Performance ... 9
2.5 The Partially Mediating Effect of CSR ... 10
2.6 Conceptual Model ... 12 3. Methodology ... 14 3.1 Sample ... 14 3.2 Variables ... 14 3.3 Test Models ... 18 3.4 Robustness Tests ... 19 4. Results ... 21 4.1 Descriptive Statistics ... 21 4.2 Regression Analyses ... 23 5. Discussion ... 28 6. Conclusion ... 34 6.1 Research Contributions ... 34 6.2 Managerial Implications ... 35
6.3 Limitations and Future Research ... 35
iv
LIST OF FIGURES
Figure 1 – Conceptual model ... 13
LIST OF TABLES
Table 1 – Sample distribution by country and industry ... 15 Table 2 – Descriptive statistics ... 21 Table 3 – Pearson correlations ... 22 Table 4 – The effects of home country institutional quality and internationalization on
CSR ... 23 Table 5 – The relationship between CSR and firm performance ... 24 Table 6 – The mediating effect of CSR on the relationships between home country
institutional quality and financial performance and between firm
v
LIST OF ACRONYMS
CSR Corporate social responsibility
DMNE Developed market multinational enterprise
E20 20 emerging market countries
EM Emerging market
EMNE Emerging market multinational enterprise ESG Environmental, social and corporate governance
FDI Foreign direct investment
GDP Gross domestic product
GRI Global Reporting Initiative
HC3 Heteroscedasticity consistent version 3
HCCM Heteroscedasticity consistent covariance matrix LOF Liability of foreignness
MNE Multinational enterprise OLS Ordinary least squares
ROA Return on assets
ROE Return on equity
UNCTAD United Nations Conference on Trade and Development VIF Variance inflation factor
1
1. INTRODUCTION
The number of multinational enterprises (MNEs) coming from emerging markets (EMs) is on the rise. In 2018, nearly a third of the firms listed in the global Fortune 500 originated from emerging economies, against merely 7% in 2005 (Casanova & Miroux, 2018). Emerging market multinational enterprises (EMNEs) are also increasing sources of outward foreign direct investment (FDI). Although there has been some stagnation and drops in FDI outflow due to the global economic crisis, EMs comprised 20% of worldwide outward FDI in 2015 compared to 2% in 2000 (Casanova & Miroux, 2016). A recent example of May 2019 is the major $2bn acquisition of British Avon Products by the Brazilian cosmetics company Natura, the parent of Body Shop, making it the largest direct cosmetics selling company worldwide (Fontanella-Khan & Schipani, 2019).
EMs differ from developed countries in terms of economic and institutional development. Economically speaking they are characterized by a low average per capita income, relatively high gross domestic product (GDP) growth, economic reforms towards liberalization and increased integration into the global market (Marquis & Raynard, 2015). From an institutional point of view, emerging economies lack a well-established institutional framework which hampers market efficiency and raises transaction costs, such as search costs due to information asymmetries (Meyer, Estrin, Bhaumik, & Peng, 2009). Institutional theory deals with the effect of (lack of) quality of formal institutions on economic performance (North, 1990). It forms an important cornerstone in explaining differences between emerging and developed countries (Peng, Wang, & Jiang, 2008). One of these differences concerns the institutional distance (Kostova, 1996) that EMNEs face when internationalizing. This hinders the establishment of legitimacy in the foreign country, for instance due to negative stereotypes (Kostova & Zaheer, 1999). Another difference relates to the liability of emergingness, which refers to additional barriers that EMNEs have to deal with only because of originating from an EM (Madhok & Keyhani, 2012). Examples of such barriers are weak market-supporting institutions and low legitimacy requirements in the home country compared to developed countries. This makes adhering to global standards and competing with global rivals coming from developed markets more challenging (Liou & Lamb, 2018; Peng et al., 2008).
2
Skouloudis, 2016; Pan, Chen, & Ning, 2018). These studies also indicate that more investigation is needed on this topic, especially in EM regions.
Research shows a positive effect of CSR on firms’ financial performance (Hou, Lu, & Hung, 2019; Olaniyi, Jalloh, & Abdusalam, 2015; Orlitzky, Schmidt, & Rynes, 2003; Russo & Fouts, 1997; Tsoutsoura, 2004). However, in the case of Latin America, an EM region, mixed results have been found (Amini & Dal Bianco, 2017). The present thesis seeks to further analyze this relationship, but also includes home country institutional quality and internationalization as distinct factors determining the amount of CSR initiatives among EMNEs. Recent research on the effect of EMs’ home country institutional quality on firm performance measured by productivity found mixed results and called for a more in-depth analysis (Bhaumik, Dimova, Kumbhakar, & Sun, 2018). In addition, scholars found that internationalization among Latin American firms has a positive effect on their financial performance, but they also indicated that more research is needed on this matter (Cuervo-Cazurra, Ciravegna, Melgarejo, & Lopez, 2018). Better institutional quality leads to more stakeholder pressure for CSR (Pan et al., 2018) and the internationalization of EMNEs results into a higher need to invest in CSR to gain legitimacy (Beddewela & Fairbrass, 2016; Liou & Lamb, 2018; Reast, Maon, Lindgreen, & Vanhamme, 2013). Therefore, it is expected that CSR to some extent positively influences the relationships between home country institutional quality and firm performance as well as between internationalization and firm performance.
Although there is an upward trend of CSR in Latin America and five of the E20 countries (20 emerging market countries) are situated on that continent (Casanova & Miroux, 2018), limited literature has been dedicated to this region compared to other EM areas (Amini & Dal Bianco, 2017; Visser, 2008). Therefore, the focus area of this thesis concerns firms from the five Latin American EMs: Argentina, Brazil, Chile, Colombia and Mexico. Nevertheless, the rationales behind the raised hypotheses could also be applied to developed market contexts.
The current thesis addresses the following research question: does CSR to some extent
mediate the impacts of home country institutional quality and firm internationalization on the financial performance of Latin American firms? It empirically examines the research question
among 159 firms originating from the aforementioned five countries by conducting OLS regression analyses.
3
the international business literature: CSR and internationalization among Latin American firms. It adds to gaining understanding of the impacts of the growing number of EMNEs in the worldwide economy and how they behave in terms of socially and environmentally responsible practices. This is necessary, since currently there is a gradual shift from developed country multinational enterprises (DMNEs) to EMNEs as major players dominating the global marketplace (Guillén & García-Canal, 2009). Therefore, universal problems, such as climate change and unfair labor practices, will increasingly need to be addressed by EMNEs in the form of CSR. Secondly, scholars have indicated that more research is needed on the effects of EM’s institutional quality on CSR (Halkos & Skouloudis, 2016; Pan et al., 2018) and on financial firm performance (Bhaumik et al., 2018). This thesis aims to fill this gap. It is relevant for policymakers to be aware of the importance of institutions in advancing the economic performance of firms. Especially EMNEs could gain much from an improved level of institutions in their home country, since the general institutional quality in emerging markets is low compared to developed countries (Meyer et al., 2009). Thirdly, to the best of my knowledge a similar analysis on the partially mediating effect of CSR on the relationship between internationalization, institutional quality and firm performance in EMNEs has not been conducted. This thesis proposes a theoretical framework that explains how CSR could function as a mechanism between home country institutions and financial performance, as well as between foreign expansion and corporate performance. Although the focus of this thesis is on EMNEs, the rationales behind the raised hypotheses are of a general nature and therefore also applicable to DMNEs.
4
2. LITERATURE REVIEW
2.1 Institutional Theory
Institutions are commonly known as “the rules of the game in a society” (North, 1990: 3) that structure human interaction through formal and informal rules. They directly impact the economic performance of a nation by either reducing or incrementing transaction costs for the exchange or production of goods and services (North, 1990). A distinction can be made between formal and informal institutions (Peng et al., 2008). Examples of formal institutions include legal forces and market institutions. Regulatory institutions, such as national governments, are empowered to determine rules, monitor them and impose sanctions if individuals do not comply with them (Scott, 2008). The role of market-supporting institutions, such as specialist intermediaries, is that they “ensure that property rights are respected, that people can be trusted to live up to their promises, that externalities are held in check, that competition is fostered and that information flows smoothly” (McMillan, 2008: 3). They can for instance decrease information asymmetries by providing knowledge to firms about business partners and their potential conduct. In this manner risks related to partners can be reduced (Meyer et al., 2009). Informal institutions, on the other hand, are rather underlying constructs, such as culture or norms that determine interpersonal relationships (Peng et al., 2008).
In EMs transaction costs tend to be high because of underdeveloped formal institutions and as a result imperfect markets, which gives rise to issues such as corruption (Liou & Lamb, 2018; Meyer et al., 2009). When EMs improve their institutional quality, this leads to more efficient markets, better contract enforcement and increased levels of trust (Kafouros & Aliyev, 2016). EMNEs internationalizing to developed countries face different and better institutional environments than in their home country, such as dissimilar regulatory, cognitive and normative institutions. This is commonly referred to as the institutional distance between a home and host country (Kostova, 1996). The home country is where the EMNE’s headquarters are located and the host country where its subsidiaries operate.
5
environment influences the MNE because of challenges related to legitimacy and liability of foreignness (Kostova et al., 2008).
The first application describes how organizational activities are disseminated, adopted and institutionalized when the MNE expands internationally (Kostova et al., 2008). The MNE’s subsidiaries are embedded in a situation described as institutional duality, which refers to the idea that a subsidiary needs to adhere to both mandates by the parent company, that on its turn is influenced by the home institutional context, and institutional practices of the host environment. As a result, the subsidiary faces two different types of isomorphic pressures (Kostova & Roth, 2002). The concept of multiple embeddedness takes this idea even a step further by also considering the fact that the MNE’s headquarters need to cope with heterogeneous, local institutional contexts of all the different host countries (Meyer, Mudambi, & Narula, 2011). The parent thus faces a wide variety of isomorphic pressures, which relates to the second application of institutional theory about isomorphic forces.
DiMaggio and Powell (1983) identified three distinct forces that contribute to the adaptation of institutional requirements and to isomorphic change. Coercive isomorphism concerns formal pressures, enforced by the government or other organizations that have more authority than the MNE, and to informal pressures, such as societal expectations. Mimetic isomorphism occurs when a firm shows imitative behavior of other local successful companies. Lastly, normative isomorphism is driven by professionalization and entails the establishment of practices that are deemed appropriate by the environment (DiMaggio & Powell, 1983). Ultimately, adhering to these forces enhances the legitimacy and profitability of the firm (Kostova & Roth, 2002).
The third application in the international business literature further specifies the relationship between a subsidiary and the foreign institutional context. Every subsidiary faces a liability of foreignness in a host environment because of increased costs compared to domestic firms (Zaheer, 1995). Liability of foreignness negatively impacts a subsidiary’s legitimacy due to stereotypes and distinct legitimacy expectations, such as the engagement in CSR that targets local communities (Kostova & Zaheer, 1999).
6
authorities is by strategic isomorphism, which is a form of mimetic isomorphism and entails adjusting to strategies implemented by other firms (Deephouse, 1996). CSR has also been mentioned as a tool to increase legitimacy (Beddewela & Fairbrass, 2016). Beddewela and Fairbrass (2016) found that coercive and normative isomorphic forces were clear influential factors in the amount of CSR practices conducted by MNEs in Sri Lanka with the aim to enhance their legitimacy in the host context. Ozdora-Aksak and Atakan-Duman (2016) studied legitimacy among Turkey’s largest firms and revealed relationships between the business area of CSR and the different isomorphic pressures. They found that when the CSR practice focused on the core business of the firm, it either dealt with an economic or legal responsibility through coercive or mimetic isomorphism. If the CSR activity was of a philanthropic nature, it had a discretionary responsibility and the related pressure was normative isomorphism. When the CSR initiative implied both core business and discretionary elements, the activity implied an ethical responsibility forced by mimetic and normative isomorphism. The next section further explores the relationship between CSR and legitimacy in an emerging market context.
2.2 CSR, Legitimacy and Institutional Quality in Emerging Markets
The international business literature has widely examined CSR and provides a broad variety of definitions (Carroll, 1999). It makes a distinction between CSR concentrating on either governmental, environmental or social matters (Hawn & Ioannou, 2016). Some scholars of institutional theory underline the role of the government in advancing CSR activities (Albareda, Lozano, & Ysa, 2007; Whitford & Provost, 2019). Other studies highlight the environmental responsibility of the firm in addition to reaching corporate objectives (Lioui & Sharma, 2012; Nie, Wang, & Meng, 2019). Lastly, there are studies emphasizing the social component by considering CSR as social initiatives that firms undertake beyond legal requirements (McWilliams & Siegel, 2001). The present thesis takes an integrative perspective on CSR by incorporating firms’ actions related to governmental, ethical, environmental, social and socioeconomic issues to the benefit of internal and external stakeholders and the local community (Otubanjo, 2013; Venturelli, Caputo, Leopizzi, Mastroleo, & Mio, 2017).
7
the government, society and consumers, can enhance legitimacy outside the firm. External legitimacy refers to the extent to which these stakeholders approve the activities of a firm (Zheng, Luo, & Maksimov, 2015). Examples of external CSR initiatives are sustainability reports, public claims or philanthropy (Dai, Du, Young, & Tang, 2018; Zheng et al., 2015). As opposed to external CSR, internal CSR initiatives address the employees and the owners of the firm, such as corporate policies on saving energy or the incorporation of CSR in existing business processes (Hawn & Ioannou, 2016; Zheng et al., 2015). Ultimately this gives rise to legitimacy within the firm. Internal legitimacy entails the degree to which employees view the company’s mission and actions as correct (Zheng et al., 2015).
The establishment and maintenance of legitimacy in a host country is a particularly relevant issue for EMNEs, since they come from markets with underdeveloped institutions (Liou & Lamb, 2018), a common characteristic of EMs (Meyer et al., 2009). EMNEs are “international companies that originated from emerging markets and are engaged in outward FDI, where they exercise effective control and undertake value-adding activities in one or more foreign countries” (Luo & Tung, 2007: 482). Institutional voids of emerging markets are for instance reflected by a lack of quality and safety regulations, an inadequate enforcement of laws and corrupt business practices (Luo & Tung, 2007; Madhok & Keyhani, 2012; Meyer et al., 2009). The large institutional distance between EMs and developed countries, the lack of knowledge of the host environment, the absence of a local network to facilitate conducting business and the presence of negative stereotypes result into greater liability of foreignness (Eden & Miller, 2004; Kostova & Zaheer, 1999). Besides, EMNEs need to adhere to stricter business standards and regulations than at home (Liou & Lamb, 2018). The concept of ‘liability of emergingness’ integrates the additional hurdles that EMNEs need to defeat, simply by originating from an emerging economy (Madhok & Keyhani, 2012). Consequently, EMNEs face a greater challenge in establishing legitimacy compared to their developed market counterparts (Liou & Lamb, 2018).
8
quality. Pressures from different stakeholders therefore result in companies conducting more CSR (Pan et al., 2018). These institutional forces are examples of formal pressures. Therefore the current thesis understands institutional quality in terms of formal institutions as explained above.
Activities undertaken by stakeholders as well as the influence that stakeholders have form an essential predictor for whether firms conduct CSR and the types of CSR actions that they engage in (Aguinis & Glavas, 2012). This suggests that there is a direct effect of the institutional quality on CSR initiatives undertaken by firms. Prior literature also indicates that more research is needed on this topic, especially in EM regions (Halkos & Skouloudis, 2016; Pan et al., 2018), such as Latin America (Marano & Kostova, 2016). The present thesis therefore hypothesizes the following:
Hypothesis 1. The institutional quality of the home country is positively related to the amount of CSR engagement among EMNEs.
2.3 Internationalization of EMNEs and CSR
Internationalization is the process “through which a firm expands the sales of its goods or services across the borders of global regions and countries into different geographic locations or markets” (Hitt, Ireland, & Hoskisson, 2007: 251). Compared to DMNEs, EMNEs follow rapid ways of internationalization to distant countries (Guillén & García-Canal, 2009) which go against more traditional paths of expansion, such as the Uppsala model (Johanson & Vahlne, 1977). According to the springboard model, EMNEs utilize their overseas expansion in a systematic and reiterative fashion for two reasons: to gain essential resources required to compete more efficaciously against domestic and global players and to diminish their vulnerable position with respect to the weak institutional and market environment in their home country (Luo & Tung, 2007).
9
Attig et al. (2016) demonstrated with a large study among firms in 44 countries that internationalization is significantly and positively linked to CSR activities because of increased responses to stakeholder demands. 62% of the observations came from three developed countries, namely the United States, the United Kingdom and Japan, which could question the generalizability of the study. Muller (2018) showed a different picture among 1,056 US MNEs, namely that the effect of CSR on firm performance depends on the level of internationalization. The costs of conducting CSR outweigh benefits when the firm is at lower levels of internationalization, whereas the opposite is true at higher levels of internationalization (Muller, 2018). The effect of internationalization on CSR activities among Latin American MNEs has not been studied in depth in the literature (Cuervo-Cazurra et al., 2018). Since EMNEs try to adhere to existing global standards that are higher than in their home country (Marquis & Raynard, 2015), it is expected that internationalization of Latin American firms has a positive impact on their amount of CSR incentives:
Hypothesis 2. Firm internationalization is positively related to CSR engagement among EMNEs.
2.4 CSR and Financial Firm Performance
Many studies have outlined the positive impact of CSR on firms’ financial performance (Busch & Friede, 2018; Hou et al., 2019; Olaniyi et al., 2015; Orlitzky et al., 2003; Russo & Fouts, 1997; Tsoutsoura, 2004). Examples of strategic CSR benefits are cost reduction, a greater demand for a firm’s products and increased competitive advantage compared to other companies (Baron, 2001; Russo & Fouts, 1997). Some studies have argued for the negative effects of CSR on corporate performance (Friedman, 1970), such as a decreased focus on shareholders’ value maximization by inducing higher costs (Becchetti, Ciciretti, & Hasan, 2009) and a nonoptimal resource allocation (Friedman, 1970). El Ghoul, Guedhami and Kim (2017) analyzed 2,445 firms from 53 countries and found that CSR enhanced firm value to a greater extent in countries with weaker functioning market institutions than in developed countries. CSR for instance contributed to gaining access to financing when well-established equity and credit markets were lacking (El Ghoul et al., 2017).
10
developed countries, CSR positively affects the financial performance of companies as it encourages development. In middle-market economies the effect is even greater, because of enhanced technological capabilities. When the nation has industrialized, however, the positive impact vanishes (Amini & Dal Bianco, 2017). A shared feature of EMs is their rapid industrialization (Marquis & Raynard, 2015), implying that they are at a further stage of industrialization in which CSR still has a positive effect on firm performance, as suggested by Amini and Dal Bianco (2017). This thinking results in the following hypothesis:
Hypothesis 3. CSR has a positive effect on EMNEs’ financial performance.
2.5 The Partially Mediating Effect of CSR
Research among developing countries showed that institutional quality does not have the same impact on firm performance across and within nations and that more research is needed on this matter (Bhaumik et al., 2018; Dollar, Hallward‐Driemeier, & Mengistae, 2005). Scholars have confirmed that institutions affect firm performance also through factor inputs and have suggested that the influence of institutions on firm performance needs to be reexamined (Bhaumik et al., 2018). Adopting the ordonomic logic, firms can employ morality, that is, CSR, as a factor of production (Hielscher, 2011). The ordonomic perspective on the role of businesses in society argues that companies need to take social responsibility in their daily operations, in the public field of determining rules and in the political area of formulating these rules (Will & Hielscher, 2014). This implies that CSR can gradually become integrated, in other words institutionalized, in the national business system of a country (Matten & Moon, 2008). The effect of CSR therefore plays an important role in explaining the relationship between home institutional quality and firm performance.
Theoretically speaking, CSR could affect the influence of home country institutional quality on firm performance in two manners. It could fulfil the role of a moderator by weakening or strengthening the relationship between institutional quality and financial performance or of a (partial) mediator by explaining the relationship between the two (Baron & Kenny, 1986). Partial mediation is a form of mediation and occurs when (1) the effect of the mediator is significant; and (2) the direct effect of the independent variable on the dependent variable without mediator is significant (Baron & Kenny, 1986).
11
integrated 588 journal articles and 102 books showed that CSR has merely been studied as a dependent or an independent variable and not as a moderator (Aguinis & Glavas, 2012). Within international business, CSR has thus almost not been examined as a moderator. Secondly, the international business literature found significant, direct effects between the quality of home country institutions and CSR (Halkos & Skouloudis, 2016), due to increased stakeholder demand for CSR (Pan et al., 2018) and institutional forces, such as standards and regulations, in favor of CSR (Aguinis & Glavas, 2012). CSR on its turn positively impacts firm performance (Busch & Friede, 2018; Orlitzky et al., 2003). A second-order meta-analysis that combined 25 prior meta-analyses and examined nearly one million observations established a direct and positive relationship between CSR and firm performance (Busch & Friede, 2018). A positive, direct effect between home country institutional quality and firm performance has however also been established in the literature, for instance for firms originating from Asian emerging economies (Dollar et al., 2005) and for domestic and foreign subsidiaries from varying parent companies worldwide (Gugler, Mueller, Peev, & Segalla, 2013). These significant results of direct influences of institutional quality on CSR, CSR on firm performance and lastly, quality of institutions on financial performance, support the choice of analyzing CSR as a partial mediator. CSR as a moderator, on the other hand, would imply an indirect effect and thus contradict the findings of most prior research on CSR. Hence, this thesis explores the partially mediating effect of CSR.
A recent study on the effect of institutional quality on firm performance in developing countries and EMs only included one Latin American country, namely Brazil. Because of the large inter-country differences found in the effect of institutions on financial performance, it called for a revision of this impact (Bhaumik et al., 2018). The present thesis tests this relationship for five Latin American EMs and includes CSR as a partial mediator. To summarize:
Hypothesis 4. CSR has a partially mediating effect on the positive relationship between home country institutional quality and EMNEs’ financial firm performance.
12
credibility and legitimacy shortage in the perspective of host country stakeholders because of the liability of emergingness (Madhok & Keyhani, 2012). CSR is shown to ameliorate a firm’s reputation (Aguinis & Glavas, 2012; Brammer & Pavelin, 2006; Brammer & Millington, 2005; Wang & Qian, 2011), which in turn improves organizational legitimacy (Zheng et al., 2015). CSR engagement increases firm performance because of enhanced legitimacy (Wang & Qian, 2011). Consequently, CSR is important in the internationalization-firm performance relationship. Nevertheless, research has also demonstrated that a direct influence between firm internationalization and financial performance exists (Cuervo-Cazurra et al., 2018). Hence, the current thesis argues that CSR functions as a partial mediator between internationalization and financial firm performance. This results into the following hypothesis:
Hypothesis 5. CSR engagement by EMNEs has a partially mediating effect on the positive relationship between firm internationalization and EMNEs’ financial firm performance.
2.6 Conceptual Model
Figure 1 summarizes the independent variables, mediator, dependent variable and the raised hypotheses into one conceptual model. Although the current thesis investigates the hypotheses among Latin American EMNEs since Latin America is an understudied region (Marano & Kostova, 2016), the theoretical framework behind the hypotheses could also be generalized to DMNEs or EMNEs from other regions.
13
level of CSR activities conducted by firms because of the aforementioned reasons. Lastly, Hypothesis 5 proposes that CSR functions as a partial mediator between the positive effect of firm internationalization on firm performance.
FIGURE 1
Conceptual model of the effects of home country institutional quality and firm internationalization on CSR and firm performance.
14
3. METHODOLOGY
3.1 Sample
In order to test whether CSR plays a partially mediating role in the home institutional quality-firm performance and internationalization-quality-firm performance relationships, quality-firms from Latin American EMs have been selected: Argentina, Brazil, Chile, Colombia and Mexico. These five countries are considered emerging economies in that region (Casanova & Miroux, 2018). Firstly, data on CSR is collected from the database ASSET4 from Thomson Reuters by including all companies with headquarters in the five Latin American countries, yielding 212 firms. Hereafter, financial data for these firms is retrieved from Orbis, a database by Bureau van Dijk. This procedure results in 194 firms being left. Missing data about the year of incorporation is added by drawing on the company website. When data on the control variables research and development expenses (R&D expenditure) and Cash are missing, they are set equal to zero (Servaes & Tamayo, 2013). Lastly, the data is cleaned from missing values for the remaining regression variables. The final sample consists of 159 firms. The CSR and financial data are cross-sectional, except for financial performance which includes time series.
Table 1 provides an overview of the number of firms per Latin American country in the sample and the industry that they operate in. Brazil is the most represented country with nearly half of the total number of companies. Chilean and Mexican firms constitute both around a fifth of the sample. Only a minor number of Argentinean and Colombian companies are included, which can be explained by the general lack of data on emerging market firms (Bekaert & Harvey, 2002). As Table 1 depicts, the companies cover a wide variety of industries. Most firms are part of the utility industry (18.9%), followed by the retail sector (11.9%) and food and tobacco manufacturing (10.1%).
3.2 Variables
Dependent variable: firm performance
15 TABLE 1
Sample distribution: by country.
Country N Percentage Argentina 9 5.7 Brazil 75 47.2 Chile 34 21.4 Colombia 8 5.0 Mexico 33 20.8 Total 159 100.0
(Table 1 continued) Sample distribution: by industry.
Industry N Percentage
Agriculture, Horticulture & Livestock 2 1.3
Banking, Insurance & Financial Services 8 5.0
Business Services 1 0.6
Chemicals, Petroleum, Rubber & Plastic 8 5.0
Communications 5 3.1
Computer Software 2 1.3
Construction 8 5.0
Food & Tobacco Manufacturing 16 10.1
Industrial, Electric & Electronic Machinery 1 0.6
Leather, Stone, Clay & Glass products 5 3.1
Media & Broadcasting 2 1.3
Metals & Metal Products 9 5.7
Mining & Extraction 5 3.1
Property Services 8 5.0
Public Administration, Education, Health & Social
Services 2 1.3
Retail 19 11.9
Textiles & Clothing Manufacturing 1 0.6
Transport Manufacturing 3 1.9
Transport, Freight & Storage 13 8.2
Travel, Personal & Leisure 3 1.9
Utilities 30 18.9
Waste Management & Treatment 1 0.6
Wholesale 1 0.6
Wood, Furniture & Paper Manufacturing 6 3.8
16
correlated with CSR than market-based measures (Orlitzky et al., 2003), therefore this thesis uses an accounting-based estimate of firm performance in line with other studies on CSR and financial performance (Russo & Fouts, 1997; Tsoutsoura, 2004). Firm performance is measured by ROA, the ratio of net income to total assets, since this is viewed as an objective measurement of firm performance (Calantone, Cavusgil, & Zhao, 2002). The information on ROA is extracted from the Orbis database.
Independent variables: home country institutional quality and firm internationalization
Consistent with other studies investigating CSR and institutional conditions (Halkos & Skouloudis, 2016; Tashman, Marano, & Kostova, 2019), home country institutional quality is measured by using the World Governance Indicators (WGI) from the World Bank. The WGI provide information on the regulatory quality of country institutions based on more than 30 data sources (World Bank Group, 2019). It obtains six dimensions on a scale ranging from –2.5 to 2.5: control of corruption, government effectiveness, political stability and absence of violence and terrorism, regulatory quality, rule of law and voice and accountability (Kaufmann, Kraay, & Mastruzzi, 2010). The higher the score, the better a country ranks on a particular dimension. The variable Institutional quality of each home country is calculated by taking the average of the sum of the scores of every dimension.
The most common indicator of firm internationalization is foreign sales to total sales (Attig et al., 2016; Sullivan, 1994). Other measures are foreign assets to total assets or foreign employment. The benefit of using foreign sales over other indicators is that “only sales dispersion constitutes a true performance measure at the output level” (Rugman & Verbeke, 2004: 7). A caveat is that it does not consider the scope of internationalization. Foreign sales comprises the sum of sales from all the foreign geographic segments other than the home country (Attig et al., 2016). Data on total sales and foreign sales per geographic segment for the variable Internationalization is extracted from Orbis.
Mediator: CSR
17
specific domain can offset low scores on another domain (Escrig-Olmedo, Muñoz-Torres, Fernández-Izquierdo, & Rivera-Lirio, 2014). Moreover, the scores vary to a wide extent between providers, due to diverging measuring methods and the use of different definitions of ESG performance (Berg, Kölbel, & Rigobon, 2019). Because of mere access to one provider, this thesis cannot control for issues related to measurement divergence. It uses the ESG ratings from the database ASSET4 from Thomson Reuters that provides information about more than 6,000 public firms worldwide (Thomson Reuters, 2017). ASSET4 is based on over 400 metrics and has the highest number of indicators per category compared to other providers, such as KLD and Sustainalytics (Berg et al., 2019).
Control variables
This thesis includes common firm-level variables to control for the relationship between the two independent variables and firm performance. Firstly, research has shown that firm size is associated with firm performance (Lee, 2009; Younis & Sundarakani, 2019). Larger companies have higher financial performance than smaller firms because of economies of scale and scope and higher market dominance (Scherer, 1973). Firm size is measured as the number of employees per company (Le & Kroll, 2017). Since the database Orbis contained mostly missing values and access to the platform Eikon from Thomson Reuters was hindered, Firm size could eventually not be included in the regression analyses.
Secondly, research has established that the availability of liquidity influences financial performance (Giacomino & Mielke, 1995; Uyar, 2009). When a firm disposes of a higher amount of cash, it is less dependent on external financing sources. As a result, borrowing costs decrease which positively affects the net profitability of a firm (Uyar, 2009). Therefore Cash is included as another control variable. The information on Cash is gathered through Orbis.
18
Lastly, R&D expenditure is incorporated as a control variable. Despite the fact that R&D costs are usually expensed rather than capitalized, they may generate value to the firm (Servaes & Tamayo, 2013). Particularly when firms are internationalizing, the capability to leverage multinationality benefits, such as scale economies, is contingent on the level of innovativeness (Kotabe, Srinivasan, & Aulakh, 2002). R&D expenditure is measured as the ratio of R&D expenses to total sales (El Ghoul et al., 2017).
Besides the firm-level controls, initially three country-level control variables were included, since especially in emerging markets country effects positively impact firm performance to a greater extent compared to developed countries (Goldszmidt, Brito, & de Vasconcelos, 2011). The intended control variables were GDP, population and the unemployment rate of each Latin American country. However, since there was a very high correlation between GDP and population (0.996) indicating multicollinearity1, a composite variable of GDP per capita was created. Composite variables combine two highly correlated independent variables into one in order to solve multicollinearity issues (Berry & Feldman, 1985). As a result, only two country-level control variables remain: Unemployment and GDP
per capita. The information on the country-level controls is obtained from the World Bank.
3.3 Test Models
In order to test the partially mediating effect of CSR, the approach suggested by Baron and Kenny (1986) is followed: 1) the effect of the independent variables on the mediator is analyzed. Together with the control variables, the impacts of home country institutional quality and internationalization on CSR are examined in regression models 1 and 2, testing Hypothesis 1 and Hypothesis 2 respectively:
𝐶𝑆𝑅 = 𝛼0+ ∑ 𝑎𝑘𝐶𝑇𝑅 5 𝑘=1 + 𝛼8𝐼𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 + 𝜀1 (1) 𝐶𝑆𝑅 = 𝛽0+ ∑ 𝛽𝑘𝐶𝑇𝑅 5 𝑘=1 + 𝛽8𝐼𝑛𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 + 𝜀2 (2)
where in (1) CTR comprises the five control variables and institutions represents the home country institutional quality; 2) the influence of the independent variables on the dependent
19
variable is investigated. The controls together with firstly, home country institutional quality, and secondly, internationalization are added in regression models 3 and 4 to examine their effects on firm performance:
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 = 𝛾0+ ∑ 𝛾𝑘𝐶𝑇𝑅 5 𝑘=1 + 𝛾8𝐼𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 + 𝜀3 (3) 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 = 𝛿0+ ∑ 𝛿𝑘𝐶𝑇𝑅 5 𝑘=1 + 𝛿8𝐼𝑛𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 + 𝜀4 (4)
3) the impact of the mediator on the dependent variable is analyzed. Both the control variables and the mediator CSR are introduced into regression model 5 to study their impact on firms’ financial performance, which corresponds to Hypothesis 3:
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 = 𝜁0 + ∑ 𝜁𝑘𝐶𝑇𝑅
5
𝑘=1
+ 𝜁8𝐶𝑆𝑅 + 𝜀5 (5)
4) the mediation effect is examined by adding the respective independent variable, mediator and dependent variable to regression models 6 and 7, testing Hypotheses 4 and 5 respectively:
𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 = 𝜂0+ ∑ 𝜂𝑘𝐶𝑇𝑅 5 𝑘=1 + 𝜂8𝐶𝑆𝑅 + 𝜂9𝐼𝑛𝑠𝑡𝑖𝑡𝑢𝑡𝑖𝑜𝑛𝑠 + 𝜀6 (6) 𝑃𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 = 𝜃0+ ∑ 𝜃𝑘𝐶𝑇𝑅 5 𝑘=1 + 𝜃8𝐶𝑆𝑅 + 𝜃9𝐼𝑛𝑡𝑒𝑟𝑛𝑎𝑡𝑖𝑜𝑛𝑎𝑙𝑖𝑧𝑎𝑡𝑖𝑜𝑛 + 𝜀7 (7)
Partial mediation holds when the outcomes of steps 1), 2) and 4) are significant (Baron & Kenny, 1986).
3.4 Robustness Tests
20
conducted to examine whether the models suffer from heteroscedasticity issues (Koenker, 1981). The inclusion of control variables ensures that the models do not show omitted-variable bias or autocorrelation (Poole & O’Farrell, 1971).
Reverse causality is unlikely in the case of home country institutions, since firm performance does not lead to institutional quality. With regard to internationalization and financial performance, reverse causality could be possible. Increased firm performance could provide the EMNE with additional financial resources to further internationalize. Therefore to prevent reverse causality, a time lag is included for the variables. Data on Firm performance is obtained for the second-last fiscal year and for the last fiscal year. The last financial year is for most companies 2018 and sometimes 2019. The regression analyses firstly incorporate the second-last fiscal year for Firm performance. Hereafter, they include the last fiscal year to examine whether the effect on financial performance becomes visible after an additional time lag. Data on the mediator CSR reflects the last fiscal year minus two years. As for the independent variables Institutional quality and Internationalization, data refers to the last financial year minus three years.
21
4. RESULTS
4.1 Descriptive Statistics
This chapter starts with presenting descriptive statistics and correlations of the data. It discusses the outcomes of the required robustness tests and explains necessary adjustments to the data in order to comply with the normality assumptions. Hereafter, the findings from the regression analyses to test the raised hypotheses are demonstrated and interpreted.
Table 2 provides descriptive statistics of all regression variables and Table 3 shows the bivariate Pearson correlations. The correlation between Institutional quality and GDP per
capita exceeds the threshold of 0.7, namely 0.775 (in absolute values) indicating
multicollinearity (Kennedy, 2003). However, some scholars use a less conservative threshold of 0.8 (Judge, Hill, Griffiths, Lutkepohl, & Lee, 1982). A correlation between 0.7 and 0.8 suggests there could be mild multicollinearity (Tabachnik & Fidell, 2014). Therefore, VIF statistics were estimated between all independent variables to examine whether the presence of multicollinearity would be problematic in the regression models. The VIFs varied in a range between 1.07 and 2.81, which is well below the general threshold of 10 when multicollinearity would be an issue (Hair, Black, Babin, & Anderson, 2010) and less than even more conservative thresholds of 4 or 5 (O’Brien, 2007).
TABLE 2 Descriptive statistics.
Median Mean Std. Dev. Min. Max.
1. ROA in % 3.8 4.6 6.6 –29.5 30.0
2. Institutional quality 0.09 –0.14 0.51 –0.31 1.08 3. Internationalization in % 10.2 0.0 23.7 0.0 97.3
4. CSR 53.1 49.5 19.4 7.9 85.5
5. Cash in $ million 347.9 170.0 940.8 –3,227.0 7,969.0
6. Firm age in years 40.7 35.0 29.9 0.0 151.0
7. R&D expenditure in $ million 4.8 0.0 30.2 0.0 319.0
8. Unemployment in % 8.4 7.8 2.5 3.9 11.6
9. GDP per capita in $1.000 8,813 10,100 2,247 5,871 13,790
22
more similar and solve skewness issues (Altman & Martin Bland, 1996). Consequently, these variables were transformed into logarithms. To overcome the fact that values of zero cannot be changed into a logarithm, a constant of 1 was added to every observation (Ekwaru & Veugelers, 2018). After the log transformations three variables still showed skewed distributions outside the permitted range of –1 and 1: Internationalization had a skewness of 1.66, Cash of –1.32 and
R&D expenditure of 4.58. The data was therefore inspected for outliers, since the presence of
outliers could strongly influence skewness (Brys, Hubert, & Struyf, 2004). Cash did not have any outliers, but for Internationalization and R&D expenditure all non-zero values were outliers. The skewness scores of Cash and Internationalization are close to the allowed range and hence they are not problematic for the normality assumption. R&D expenditure is quite positively skewed, but since it is a merely a control variable, it will be kept in the regression analyses. TABLE 3 Pearson correlations. 1 2 3 4 5 6 7 8 9 1. ROA 1.000 2. Institutional quality [L] .137† 1.000 3. Internationalization [L] –.143† .013 1.000 4. CSR –.066 .259** .150 1.000 5. Cash [L] .107 .021 .216** .304*** 1.000 6. Firm age [L] –.069 –.120 .127 .189* .134† 1.000 7. R&D expenditure [L] –.109 .120 .355*** .115 .161* .071 1.000 8. Unemployment .014 .522** –.172* .244** .000 .007 .062 1.000 9. GDP per capita .014 –.775** –.107 –.329*** -.025 .131 –.113 –.423*** 1.000 Notes: N = 159 for all variables. [L] indicates a log transformation. † p < .10; *p < .05; **p < .01; ***p < .001
(two-tailed test).
23
versions of the HCCM. Since the sample of this thesis is < 250, the standard errors of the version HC3 (heteroscedasticity consistent version 3) are used for the regression models (Long & Ervin, 2000).
4.2 Regression Analyses
Table 4 presents the results of the OLS regression analyses of Hypotheses 1 and 2. Hypothesis 1 anticipates that greater home country institutional quality leads to higher CSR engagement. Model 1 of Table 4 demonstrates the regression of the various control variables on CSR. The
F-value shows that the model is highly significant (p < 0.001). Three control variables, Cash (p
< 0.001), Firm age (p < 0.05) and GDP per capita (p < 0.001), proved to be significantly related to CSR. Interestingly, GDP per capita is negatively related to CSR, implying that when GDP per capita increases, the level of CSR initiatives decreases, ceteris paribus. The model explains 21.2% of the variation in CSR.
TABLE 4
The effects of home country institutional quality and internationalization on CSR. Model 1 Model 2 Model 3
Control variables Cash [L] 3.219*** (0.892) 3.221*** (0.896) 3.108*** (0.898) Firm age [L] 9.572* (4.014) 9.491* (4.027) 9.145* (4.067) R&D expenditure [L] 0.840 (2.691) 0.913 (2.710) –0.153 (2.857) Unemployment 88.957 (59.130) 97.780 (63.851) 102.857 (63.337) GDP per capita –2.525*** (0.713) –2.802** (1.014) –2.413** (0.750) Independent variables Institutional quality [L] -2.702 (7.650) Internationalization [L] 1.878 (2.303) R2 0.241 0.242 0.244 Adjusted R2 0.212 0.212 0.214 F-value 13.034*** 10.652*** 11.065***
Notes: N = 159 for all variables. Dependent variable = CSR. [L] indicates a log transformation.
Numbers in parentheses are robust standard errors with HC3 adjustment. † p < .10; *p < .05; **p <
24
Model 2 of Table 4 regresses the five control variables and Institutional quality on CSR and is significant at the 0.1% level. Cash (p < 0.001), Firm age (p < 0.05) and GDP per capita (p < 0.01) are significantly related to CSR. Institutional quality is however not significantly linked to CSR, hence Hypothesis 1 has to be rejected. This is also reflected by the same adjusted R2 scores for Models 1 and 2, indicating that the inclusion of home country institutional quality into the model does not explain additional variation in CSR. Model 3 of Table 4 shows the regression of Hypothesis 2, which explores the positive effect of firm internationalization on CSR activities. The model is again highly significant (p < 0.001) and has an explanatory power of 21.4%. Three controls, Cash (p < 0.001), Firm age (p < 0.05) and GDP per capita (p < 0.01) show a significant relationship with CSR. Internationalization is however not significantly related to Firm performance and therefore Hypothesis 2 is not accepted. Compared to Model 1, the adjusted R2 of Model 3 shows that there is a slight increase of 0.2% in the explained variation by Internationalization entering into the regression. Model 3 has the highest explanatory power of the models presented in Table 4, namely 21.4%.
Table 5 introduces the regression results of Hypothesis 3, which proposes that higher CSR engagement leads to better financial performance. Model 1 regresses the control variables
TABLE 5
The relationship between CSR and firm performance.
Model 1 Model 2 Control variables Cash [L] 0.550* (0.285) 0.663** (0.312) Firm age [L] –1.510 (1.860) –1.176 (1.976) R&D expenditure [L] –2.048 (2.327) –2.019 (2.315) Unemployment 9.060 (19.222) 12.167 (19.661) GDP per capita 0.080 (0.264) –0.009 (0.263) Independent variable CSR –0.035 (0.031) R2 0.035 0.043 Adjusted R2 0.003 0.005 F-value 1.111 1.046
Notes: N = 159 for all variables. Dependent variable = Firm performance. [L] indicates a
25
on Firm performance and Model 2 adds CSR as an independent variable. The F-values of Models 1 and 2 indicate that both models are not significant and hence should not be interpreted. As a result, Hypothesis 3 is not supported. A repetition of the regression analyses with a time lag of two years between CSR and Firm performance was also insignificant, although the first model (p < 0.10) with merely controls yielded significant results for two control variables (see Table A1, Appendix A).
Table 6 demonstrates the OLS regression analyses testing Hypotheses 4 and 5. Hypothesis 4 anticipates that while home country institutional quality positively impacts financial firm performance, that positive relationship is partially mediated through CSR. Model 1 of Table 6 shows the regression of the control variables and Institutional quality on Firm
performance. The model is significant (p < 0.01) and has an explanatory power of 5.8%. Also
the controls Cash (p < 0.10) and GDP per capita (p < 0.05) show significant relationships. The
TABLE 6
The mediating effect of CSR on the relationships between home country institutional quality and financial performance and between firm internationalization and financial performance.
Model 1 Model 2 Model 3 Model 4
Control variables Cash [L] -.545† (0.294) 0.648† (0.331) 0.633 (0.284) 0.732* (0.311) Firm age [L] –1.274 (1.860) –0.972 (1.957) –1.193* (1.879) –0.902 (1.983) R&D expenditure [L] –2.260 (2.391) –2.231 (2.380) –1.314 (2.770) –1.318 (2.762) Unemployment –16.496 (19.608) –13.381 (19.977) –1.219 (20.671) 2.065 (21.413) GDP per capita 0.882* (0.421) 0.793† (0.418) –0.003 (0.282) –0.080 (0.275) Independent variable Institutional quality [L] 7.811** (2.706) 7.724** (2.631) Internationalization [L] –1.389 (1.079) –1.328 (1.083) Mediator CSR –0.032 (0.023) –0.032 (0.031) R2 0.094 0.101 0.050 0.057 Adjusted R2 0.058 0.059 0.013 0.013 F-value 2.940** 2.657* 1.885† 1.775†
Notes: N = 159 for all variables. Dependent variable = Firm performance. [L] indicates a log transformation.
26
independent variable Institutional quality is significant too at the 1% level, indicating that an increase in home country institutional quality leads to a growth in firm performance, ceteris paribus. An additional test that included Firm performance of one year later also proved to be significant at the 10% level (see Model 1 of Table A2, Appendix A). However, Institutional
quality no longer was significant in that regression. Model 2 of Table 6 presents the results of
the effect of Institutional quality on Firm performance while controlling for CSR. The overall model is significant (p < 0.05) and explains 5.9% of the variance in Firm performance. Cash and GDP per capita show low significance (p < 0.10) and Institutional quality high significance (p < 0.01). Again an extra test was run to examine whether the model was still significant when
Firm performance was two years after the CSR implementation (see Model 2 of Table A2,
Appendix A). This overall regression showed a low significance at the 10% level. Institutional
quality was not significant in that model.
In order for partial mediation to hold, firstly the effect of Institutional quality on CSR needs to be significant. Model 2 of Table 4 already showed that this is not the case. Besides, the effect of CSR on Firm performance while controlling for Institutional quality has to be significant. Model 2 of Table 6 reveals that CSR is not significantly linked to Firm performance while including Institutional quality into the regression. Thirdly, the influence of Institutional
quality on Firm performance when adding CSR to the regression analysis needs to show a
significant relationship. Model 2 of Table 6 confirms that this relationship is highly significant (p < 0.01). Overall, the models show that two of the three assumptions for partial mediation do not hold and consequently Hypothesis 4 needs to be rejected.
Hypothesis 5 suggests that the positive relationship between firm internationalization and firm performance is partially mediated by CSR. Model 3 of Table 6 includes both the five controls and the independent variable Internationalization. The model is slightly significant at the 10% level. An additional test with a time lag of three years instead of two years between
Internationalization and Firm performance was also significant at the 5% level (see Model 1
of Table A3, Appendix A). That regression had an explanatory power of 6.7% compared to 1.3% of Model 1. Firm age is the only significant variable in Model 3 of Table 6 (p < 0.05). There is no significant direct effect between Internationalization and Firm performance.
27
with Firm performance. The inclusion of Firm performance one year later in an extra test appeared to be significant as well (p < 0.05, see Model 2 of Table A3, Appendix A). That additional model explained 6.5% of the variation in Firm performance, which is higher than the 1.3% explained in Model 4 of Table 6.
28
5. DISCUSSION
The present thesis examined whether CSR partially mediates the influences of home country institutional quality and firm internationalization on Latin American emerging markets’ firm performance. The previous chapters developed a theoretical framework leading to several hypotheses related to this research question. The hypotheses were empirically investigated among firms in Argentina, Brazil, Chile, Colombia and Mexico. This chapter discusses the meaning of the rejected hypotheses in light of institutional theory.
At first, it was expected that the quality of home country institutions positively contributes to the amount of CSR initiatives conducted by EMNEs. Contrary to the expectations based on theoretical assumptions and limited earlier empirical research on this topic (Halkos & Skouloudis, 2016; Pan et al., 2018), the regression analyses did not find supportive evidence for this. This implies that in the case of Latin American EMs, the country’s institutions do not seem to function as an indicator for the level of CSR implemented by firms from these countries. This could be caused by the fact that the measures for institutional quality used in this thesis only encompass factors related to regulatory quality and do not include other components, for instance concerning market support from institutions. The overall regulatory quality scores did not vary to a great extent between the countries. The difference between the lowest (Argentina:
–0.31) and the highest score (Chile: 1.08) was only 1.39, although the overall range of the measure is 5 points. This might provide a biased indication that the institutions of the countries under scrutiny are relatively similar in quality, whereas the incorporation of other institutional factors, such as the ones suggested by Campbell (2007), might show a different picture.
29
(positively) and the engagement of independent organizations (positively and negatively, depending on the measure) showed highly significant relationships with CSR activities. Therefore, the inclusion of these measures into the institutional quality score of this thesis could have provided a more holistic picture and might have yielded different results. It needs to be noted that as opposed to the outcomes of this thesis, Halkos and Skouloudis (2016) found a significant and positive relationship for regulatory effectivity on CSR. Similar to this thesis, their regulatory effectiveness variable also incorporated the WGI together with some other measures. The conflicting findings between this thesis and Halkos and Skouloudis’ (2016) research could be explained by the fact that Halkos and Skouloudis (2016) tested 86 countries worldwide with widely varying institutional contexts, whereas this thesis focused on EMs from one specific region where differences in institutions are much more of a subtle nature.
Secondly, it was hypothesized that firm internationalization would lead to greater CSR engagement among firms. The results showed that this hypothesis had to be rejected. However, the findings of the present thesis need to be interpreted with a certain degree of caution, since the number of firms with foreign sales in the sample was only 34 (21.4%). The actual number could be a bit higher, because not all firms provided information on their sales per geographic segment. This can be attributed to the general lack of financial data on emerging market firms (Bekaert & Harvey, 2002). As a result, the relatively low number of internationalized firms in the sample restricts the generalizability of the results to all Latin American EMNEs.
30
were again insignificant. As explained above, this discrepancy in findings might be caused by the low number of Brazilian firms (N = 13) with foreign sales in the sample.
Hereafter, it was anticipated that increased CSR engagement induces greater firm performance among EMNEs from Latin America. Despite ample research confirming such a positive relationship for developed countries (Russo & Fouts, 1997) and for Asian EMs (Cheung, Tan, Ahn, & Zhang, 2010), the empirical analysis of this thesis did not provide supportive evidence for this claim. A previous study on several Latin American countries conducted by Amini and Dal Bianco (2017) reported both positive and insignificant results. Therefore, it could be that the implementation of CSR is region specific and that the theoretical rationales that apply to developed markets and Asian EMs cannot be generalized to Latin America. In line with this reasoning, Fisher, Mahoney and Scazzero (2016) for instance found that the total CSR scores of firms significantly differed across certain regions. Namely, Africa, Europe and South America revealed the highest CSR scores compared to the rest of the world, whereas Latin America and Asia-Pacific scored the lowest. North America was in between the two groups. Interestingly, that study also showed a significant difference between the overall CSR scores of Latin American and South American firms (Fisher et al., 2016). The present thesis did not draw a distinction between Latin and South America and considered the whole continent below the US as Latin America. A differentiation between Latin and South American companies could be an interesting topic for further research.
Besides the importance of regional differences, there is an ongoing debate about the role of divergence in CSR measurement methods in explaining inconclusive results. Hussain, Rigoni and Cavezzali (2018) analyzed the relationship between CSR and financial performance and established that the CSR measure causes results to be heterogeneous. More specifically, they revealed that the ESG parameters2 were not significantly linked to any measure, market or accounting-based, of financial performance. Another type of CSR measure relying on guidelines issued by the Global Reporting Initiative (GRI) yielded significant results for both financial performance measurement methods (Hussain et al., 2018). As outlined in the methodology section, it is also known that ESG ratings vary between providers (Berg et al., 2019). This shows that the use of a specific dataset can form an essential predictor in the
31
outcome of the analysis. Therefore, a different manner of measuring CSR in this thesis could have yielded other findings.
Subsequently, it was hypothesized that CSR functions as a partial mediator in the positive relationship between the institutional quality of a home country and firm performance, because of more institutional forces and stakeholder demands for CSR when institutional quality rises. The regression analyses did not confirm such a partially mediating role. However, in line with the expectations, the direct effect of the quality of the country’s institutions on financial performance appeared to be highly significant and positive for the Latin American EMNEs. A study performed by Goldszmidt et al. (2011) analyzed country effects on firm performance by combining institutional and economic conditions and found that the positive influences were greater for EMs than for developed countries. Yet, separate country effects of Argentina, Brazil, Chile and Mexico disclosed that for these nations specifically the overall results were insignificant (Goldszmidt et al., 2011). The present thesis thus showed that by examining institutional factors independently, the findings are positive and significant for five Latin American emerging economies. Previous studies had already demonstrated that institutions contribute to the national economic performance and income of Latin American countries (Grier & Maldonado, 2015; Vianna & Mollick, 2018). The results of this thesis confirmed that the impact is also significant at the firm-level.
32
The measure for institutional quality of this thesis comprises elements of institutional certainty. Hence, following the reasoning and findings by Cuervo-Cazurra et al. (2018), institutional quality could change the effect of internationalization on financial performance. Logically, one would expect that a rise in quality of institutions positively adds to this relationship. A higher quality of institutions in EMs is shown to stimulate firms to expand overseas, whereas a low quality institutional environment functions as an impeding factor (Sun, Peng, Lee, & Tan, 2015). Therefore, an additional regression analysis was conducted to examine whether institutional quality positively moderates the relationship between internationalization and firm performance (see Table C1, Appendix C). Institutional quality as independent variable remains highly significant (p < 0.01), also when controlling for internationalization. Yet, the outcome of the interaction term is insignificant. This could again be explained by the low number of international firms in the sample of this thesis.
To conclude, the insignificant findings for the five hypotheses do not coincide with the theoretical assumptions made in the literature review and are incongruent with similar research conducted in developed countries or other EM regions. Overall, the explanations for these insignificant results are of a twofold nature. The first reason is data related and the second involves the importance of intraregional differences. With regard to the issue of data, this thesis needed to draw on existing databases. In the case of Latin America this is problematic, because generally there is a lack of available company data for these countries. Therefore, the reliability of the data might be questionable. Although the overall sample size of this thesis was sufficient to conduct the regression analyses, the number of international firms was rather low, which might have affected the analyses with respect to firm internationalization. Besides, there is inconsistency in how CSR is measured and what definitions of ESG are used between different providers and rating agencies (Berg et al., 2019). This causes certain CSR measures to yield significant outputs in relation to financial firm performance, whereas others do not (Hussain et al., 2018). Finally, although institutional quality was shown to significantly affect firm performance, there was no meaningful relationship with CSR. This could be attributed to the mere inclusion of regulatory effectiveness as an indicator for institutional quality in this thesis. The incorporation of additional factors, for instance competitiveness, and informal institutional forces, such as NGOs, could have rendered the measure more multidimensional.
33
34
6. CONCLUSION
The objective of this thesis was to analyze in the context of emerging markets whether CSR plays a partially mediating role in the relationship between home country institutional quality and firm performance as well as between firm internationalization and financial performance. Drawing on 159 firms from five Latin American EMs, results from regression analyses demonstrated that CSR did not function as a mediator in these two relationships. The findings indicated that home country institutional quality was positively and significantly related to financial performance. The level of firm internationalization did not show a meaningful effect on firm performance, which contradicts prior research on Latin American companies (Cuervo-Cazurra et al., 2018). Moreover, a repetition of the regressions on the effect of financial performance two years after the CSR initiatives also did not reveal significant outputs. These findings conflict with earlier meta-analyses on the CSR-firm performance relationship (Busch & Friede, 2018; Orlitzky et al., 2003). Since the outcomes of the investigation conducted in this thesis did not correspond with the theoretical expectations and similar research, it is plausible to assume that the underlying reason of the insignificant findings is data related. The limitations and future research section will further elaborate on this, but firstly the research contributions and managerial implications of this thesis are presented.
6.1 Research Contributions
