1
Home-Country Corruption Levels and its influence
on the Corporate Social Performance of MNEs
Abstract
In the field of international business the three lenses concept, and their influence on Multinational Enterprises (MNEs) and their environment is becoming increasingly important. As MNEs are becoming more influential because of globalization and affect many different countries with their behavior, knowledge on MNEs and the influence and interaction of CSR, corruption, and politics is essential. Therefore this study examines the interaction of two lenses, namely the influence of home-country corruption levels on an MNE’s social performance. Using the ASSET4 database and the Transparency International Corruption Index, a regression analysis was performed that resulted in evidence for the negative relationship between home-country corruption levels and CSP. When world-region of an MNEs home-country was taken into account, MNEs from Europe and North-America were found to have a certain social performance level, despite their home-country corruption levels.
Marloes den Os
10175563
29-06-2015
year 2014/2015
Supervisor: Dr. D.A. Waeger
9 Impact of the home-country on companies’ social or environmental performance
2
Statement of Originality
This document is written by Student Marloes den Os, who declares to take full responsibility
for contents of this document.
I declare that the text and the work presented in this document is original and that no sources
other than those mentioned in the text and its references have been used creating it.
The Faculty of Economics and Business is responsible solely for the supervision of
completion of the work, not for the contents.
3
Contents
Abstract………...………..1
1 Introduction……….………..…....5
2 Theoretical Framework………….………..……8
2.1 Social performance………..……8
2.2 Corruption…..………..……9
2.3 Integrating corruption, CSP, and MNEs………...……….11
3 Conceptual Framework……….……….…12
3.1 Home-country corruption levels and social performance………..…12
3.2 Industries and social performance…..………...…13
3.3 CSP and the different world regions…..………...……….…14
4 Methodology……….………..………...…..…17
4.1 Research design……….17
4.2 Research sample and data collection….………...….17
4.2.1 Dependent variable……...………..18
4.2.2 Independent variable……….………..……18
4.2.3 Control variables………...………..……19
4.2.4 Moderation variables…………...………..…….19
5 Results……….………...21
5.1 Sample statistics……….21
5.2 Data statistics..………...………22
5.3 Correlation results………...………...23
5.4 Regression results………...………..………25
5.4.1 Home-country corruption levels…………...………..25
5.4.2 Moderation by industry sector……….………...25
5.4.3 Moderation by world region………...………26
6 Discussion……….………...29
6.1 The link between corruption and CSR………...29
6.2 Theoretical implications………...….30
6.3 Managerial implications………...……31
6.4 Limitations and suggestions for future research……….31
7 Conclusion………..33
4
1 Introduction
The field of international business has seen a significant rise in interest the past decade. Nowhere this is more clear, than the rapidly growing number of new journals and articles dealing with interconnectedness of trade, Multinational Enterprise (MNE) strategy, and MNE performance (Dunning and Lundan, 2008). There are an increased number of studies concerning the relationship between MNEs and their role in society, which is influenced by the changes taking place in the global business climate. Globalization, the increasing volume of cross-national trade and investment, as well as recent corporate scandals, point out the importance of issues relating to politics, corruption, and Corporate Social Performance (Rodriguez et al. 2006). Politics, corruption, and CSP are generally referred to as the ‘three lenses’ on the relationship between MNEs and their environment. Each of these three topics provides a unique ‘lens’ that makes us understand how MNEs influence dimensions of their global economic and political environments (Luo, 2006; Rodriguez et al. 2006). Research on these three lenses isn’t well developed yet. A troubling trend is the development of three separate but parallel literatures on politics, corruption, and Corporate Social Responsibility. Each of the lenses has its own literature; however this literature rarely acknowledges the existence and influence of one of the other lenses. Corruption, CSR, and politics are all three examples of the non-market environment and activities (Rodriguez et al. 2006). The interconnectedness of these three lenses in literature is important for the expansion of international business research as the nonmarket environment is not just an exogenous constraint on MNEs, but is also sensitive for manipulation through MNE non-market activities such as lobbying, bribery, and CSR. In other words, not only does the environment influence MNEs; MNEs also influence their environment (Boddewyn, 1988).
Corruption has drawn attention lately with cases like the Siemens bribery scandal in 2008, which was agreed to with a settlement 1.6 billion, and the more recent Imtech fraud cases in Poland (the Guardian, 2013; Financial Times, 2013). On top of that, contacts between less corrupt and more corrupt countries are intensifying because of globalization (Habib, Zurawicki, 2002). This results in MNEs with a non-corrupt background entering geographically corrupt areas of the global economy. Before the globalization of business in the 1980s and 1990s, research on corruption and its relationship to firms’ activities was almost non-existent. Research on corruption from the past decade has answered some fundamental questions about corruption, but also addressed the importance of many new questions (Rodriguez et al. 2006). A particular important unexplained issue of corruption is the incidence of corruption (Svensson, 2003). Government corruption is suggested as one of the underlying concepts that are vital to the understanding of the effects of MNEs on the environment (Djankov et al., 2005).
5 Another concept that has been discussed within literature for decades is Corporate Social Responsibility or Corporate Social Performance (Egri and Ralston, 2008). There exists an emerging stream of literature that primarily examines the CSP of MNEs (e.g. Jamaili 2010; Tan and Wang 2011; Yang and Rivers 2009). One important reason for the growing interest in the field of MNEs and CSP, is the growing power of MNEs in the global economy (Scherer and Palazzo, 2011; Habermans, 2001). However, literature concerning CSP and MNEs is still very embryonic. This is in part because CSP is quite difficult to define, especially in the context of MNEs (Rodriguez et al. 2006). Up until now empirical research about CSP has been predominantly quantitative analyses of primary data, and has focused mainly on the links between CSR and financial performance as well as the relationship between MNEs’ internal culture and characteristics and its external social performance (Jackson and Apostolakou, 2010). Also many studies have included corruption as a dependent variable in their research. For example, Luo’s (2006) study researches the interaction between CSP and corruption, using corruption as the dependent variable. On top of that, empirical CSP studies are almost never conducted using databases and there have been very little studies concerning CSR including companies from more than two different countries (Egri and Ralston, 2008). One study that does research the cross-national influence of external factors on the CSP of MNEs using secondary data is that of Ioannou and Serafeim (2012). However, in this study the influence of absence of corruption is tested as being part of a country’s political system. Taking the three lenses concept and scarcely available literature on the interaction of the lenses in mind, I believe that research on the influence of corruption on the CSP of MNEs will contribute to the understanding of the interaction of these two specific lenses. Therefore in this study, I will focus on answering the question: to what extent do corruption levels in a MNE’s home-country influence that MNE’s Corporate Social Performance?
This study will investigate if MNEs are affected in their Corporate Social Responsibility behavior by home-country corruption levels. I expect that the industry in which MNEs operate might have a moderating effect on CSP. I also expect that home-country continent has a moderating effect on the CSP of MNEs. In this study I will use as a sample the companies from the Global Fortune 500 list. Furthermore, I will use data from DataStream and the ASSET4 database developed by Thompson Reuters. I will also use data from the Transparency International Corruption Index. This study will contribute to existing literature on the interaction of corruption and CSR in relation to MNEs, and provide readers with a deeper understanding as to what extend home-country corruption levels affect the social performance of MNEs. Readers will also learn if within certain industries the relationship between corruption levels and CSP is stronger than within other industries, and if within certain continents the relationship between corruption levels and CSP is stronger than within other continents.
6 Following this introduction, there will be a literature review where I will explore existing literature on social performance, corruption, and their interaction. Then, in the conceptual framework, I will draft the hypotheses for this study. In the methodology section the research design is explained, and the research sample and data are thoroughly discussed. This section is followed by the results of the data analysis. In the results section correlation and regression tables with the outcomes of the statistical analysis are presented, and answers are given to all proposed hypotheses. Finally, in the discussion section, the results will be explained and discussed using existing literature. The limitations of this study and suggestions for future research are also presented in this section. I will finalize with a short conclusion where I will summarize this study and shortly answer my research question.
7
2 Literature review
2.1 Social Performance
Corporate Social Performance is a concept that has gained a lot of attention and influence in the past decades. The rise of this concept has to do with several factors. One factor, which was already mentioned in relation to the three lenses concept, is globalization. Globalization can be defined as the process of intensification of cross-border social interactions due to declining costs of connecting distant locations through communication and the transfer of capital, goods, and people. This process leads to the growing transnational interdependence of MNEs (Scherer and Palazzo, 2011). A second factor that contributes to the rise in power of MNEs and their CSR policies is the decline of nation-state authority in the corporate field (Habermans, 2001). The main cause behind the expansion of CSR is found in the erosion of the division of labor between business and government. Consequently, MNEs have become more widespread globally and also practice a transnational influence. One could say actions and regulations of large MNEs are almost as influential as those of a small state, because it applies to many different countries and people. An acknowledged definition of CSR by Wood (1991) is the “configuration of the principles of social responsibility, processes of social responsiveness, and policies, programs, and observable outcomes as they relate to the firm’s societal relationships”. This definition has been completed by the ‘stakeholder view’ in many studies (Clarkson, 1995). They argue that businesses are not responsible toward society in general, but only toward their ‘stakeholders’. Firms committed to CSR try to minimize its negative impacts and maximize its positive impacts on stakeholder issues (Maigan, and Ralston, 2002).
As CSR has risen in importance the past decades, numerous studies were dedicated to this concept and its relationship with other international business concepts. Many different outcomes about CSP were found from these theoretical and empirical studies. For instance, research rooted in neoclassical economics argued that costs of firms are unnecessarily raised by CSR, as it puts the firm in a position of competitive disadvantage compared to its competitors (Jensen, 2002). Also, some studies have argued that employing in CSR results in significant managerial benefits, and not financial benefits to the firm’s shareholders (Brammer and Millington, 2008). Other studies have emphasized the positive effects of CSR. Some scholars argue that CSR can have a positive impact by providing better access to valuable resources (Waddock and Graves, 1997). Other effects of CSR include that it can positively impact the attracting and retaining of higher quality employees, allowing for better marketing of products and services, creating unforeseen opportunities, and contributing toward gaining social legitimacy (Greening and Turban, 2000; Fombrun, 1996; Fombrun et al., 2000; Hawn et al., 2011). Furthermore, many have argued that CSR may function as advertising. CSR can increase demand for products and services, and reduce consumer price sensitivity, and even eventually result in the development of intangible assets (Cheng et al., 2014). Goss and Roberts
8 (2011) found that firms with the worst CSP scores pay seven to 18 basis points more on their bank debt compared to firms with higher CSP, resulting in financial benefits for firms practicing CSR. Also, Cheng et al. (2014) found that a high CSP leads to lower capital constraints.
Swanson (1995) suggests that firms have three types of motivations to perform well in terms of CSR. The first motivation is the utilitarian perspective. This view suggests CSR can be used as an instrument to achieve its performance objectives, defined in terms of profitability, return on investment, or sales volumes. The second motivation is the negative duty approach. The stakeholders have certain norms defining appropriate behavior, and therefore businesses are forced to adopt certain social responsibility initiatives. The third motivation is the positive duty view. The positive duty view suggests that businesses are self-motivated to have a positive impact on their environment because CSR principles are part of the corporate culture. In the duty view CSR is a fundamental concept for operating, in contrast to the utilitarian view, where CSR is a means to achieve an objective (Swanson, 1995). Other motivations MNEs have to engage in CSR are discussed by Preston and O’bannon (1997). In their research they state that CSP is found to be positively associated with future financial performance. Also management theorists argue that there is a high correlation between good management practice, and CSP, simply because attention to CSP domains improves the relationships with a MNEs’ stakeholders resulting in a better overall performance (Hannan, 1984).
2.2 Corruption
The most common definition of corruption is: misuse of public power or authority for private gain (Svensson, 2005). According to Mauro (1995), corruption has been blamed for the failures of ‘developing countries’ to develop. Recent empirical research has confirmed that there is a positive link between higher perceived corruption and lower investment and growth within a country (Mauro, 1995). There is difficulty in measuring levels of corruption in different countries. Economists and political scientists recently have begun to analyze corruption levels through indexes. Such ratings are by definition subjective; however there are good reasons to be interested in what kind of patterns they reveal. Cross-national ratings tend to be highly correlated with each other. Also they tend to be highly correlated over time (Treisman, 2000). Empirical work confirms that subjective evaluations of corruption appear to influence investment decisions, growth, and the political behavior within a country (Mauro, 1995). Research on corruption faces empirical and theoretical obstacles. On the empirical hand, corruption is very difficult to measure, as corruption is illicit and secretive. Individuals will go to great lengths to hide it, and therefore it is often very difficult to gain the necessary information to create legitimate and reliable data. Fortunately, in recent years some
9 progress has been made to deal with this obstacle. The current generation of studies has focused on collecting and reporting objective information about corruption (Banerjee et al., 2012).
Corruption is a reflection of a country’s legal, economic, cultural and political institution. Corruption can be a response to either beneficial or harmful rules. On the one hand, corruption can appear in response to benevolent rules. Individuals will pay bribes to avoid penalties for harmful conduct when the monitoring of rules is not very strict. On the other hand, corruption can also arise when inefficient institutions or bad politics are put in place, where officials will collect bribes from individuals seeking to get around them (Djankov, 2003). Corruption exists in many different types. The most commonly known types of corruption are lobbying and bribery. Lobbying and bribery are often mistaken for the same; however there are a few differences. The difference between bribery and lobbying is that bribery is firm specific, although potential external effects may arise for other firms. Lobbying affects all firms in the sectors, as well as future entrants. The second difference is that changes made through lobbying tend to be more permanent, as there will be costs to re-enacting the original as a bureaucrat who was bribed before, is very likely to ask for bribes in the future. Thirdly, decisions about bribes are made by individual public officials who consider their private costs and benefits, while the benefits of income from lobbying is weighted against the cost to the government of a rule change (Harstad and Svensson, 2004).
In previous research measures of corruption were best predicted by GDP per capita, measures of regulatory barriers, and cultural characteristics (Djankov et al., 2002). Kwok and Tadesse (2006) researched the effects of MNEs on the environment of corruption. They found that corruption levels were significantly lower in countries with high Foreign Direct Investment flows in the past. This means that in countries where many MNEs are actively participating in the market, corruption levels are lower. Also found in previous research on cross-border and domestic takeovers, is that corruption markedly reduces target premiums. This means that when an MNE wants to acquire a company, they will pay less for this company if it has a home-country with high corruption levels, as opposed to low corruption levels. Weitzel and Berns (2006) estimated that one-point deterioration in a country’s ranking on the Transparency International Corruption Perceptions Index is, on average, associated with a reduction of 21% for local target premium. These results hold for both foreign and domestic acquirers. This suggests that corruption is not only a barrier for foreign firms to invest, but is also very costly to local firms. So, high corruption levels have been found to negatively affect foreign direct investment. Therefore, a study examining how corruption influences the behavior of different MNEs, including different home-countries, is a very valuable contribution to literature.
10
2.3 Integrating corruption, CSP and MNEs
Institutional theory has established that different types of institutions exert influence on MNEs decision making. MNEs are embedded within broader social structures (Campbell, 2007). Moreover, CSR activities are performed in relation to the social context surrounding an MNE, which is stated in the three lenses concept. This means MNEs and their social performance are influenced by institutions such as the government in their home-country (Jackson & Apostolakou, 2010). In the previous paragraph was stated that corruption is a reflection of a country’s legal, economic, cultural and political institution (Djankov, 2003). In other words, previous research suggests that corruption as a reflection of a countries government is likely to have an influence on the social performance of MNEs. In their analysis of previous research done on CSP, Egri and Ralston (2008) found that very little research included more than 50 different MNEs. Also none of this research has been done cross-national or with the use of databases.
So there are relatively few studies that address the effect of home-country elements on CSP and even fewer that address the relationship of the three lenses. As mentioned in the introduction, one study that does address the influence of national-level elements on CSP is that of Ioannou and Serafeim (2012). They have done a cross-national research on the influence of external factors on social performance of MNEs, including 42 different countries and. This study included the measurement of the influence of corruption as part of a country’s political system on social performance. However, the data used to create the variable for corruption measured the absence of corruption within a country, instead of national-level corruption levels. Also the data were measured using data from only 7 different databases (World Bank, 2015). In this study I will use Transparency International’s measure of perceived corruption levels per country. This measure draws from 13 different databases and measures perceived corruption per year and assigns scores accordingly to the perceived corruption levels of the countries measured. Besides the more accurate measurement of corruption, the index by Transparency international is also a very widely used source that is generally seen as a very reliable source (e.g. Uslaner, 2008; Lambsdorff, 1999).
This study will contribute to existing literature and research on corruption, CSP, and MNEs. My research will contribute to the existing literature on the interaction of the three lenses, as this study will explain more about the relationship between corruption and CSR. In particular this study will add to the understanding of the effect of home-country corruption levels on the social performance of MNEs. This study will also contribute to existing CSR literature because the study is cross-national and for the analysis of corruption and social performance secondary data is used.
11
3 Conceptual Framework
3.1 Home-country corruption levels and social performance
MNEs have become more aware of widespread corruption in business sectors. Increasingly they are taking action against corruption levels (Logan et al., 1997). MNEs attending to corruption levels mainly take the form of ethical codes of conduct with which their foreign employees must comply. This is a way to organizationally respond to the increasing levels of country corruption (Doh et al. 2003). MNEs operating in a corrupt environment will refrain from undertaking certain actions that can help improve citizenship and benefit the local community. In a highly corrupt environment, MNEs executives will evaluate that such contributions will be embezzled (Giddens, 1984). In other words, MNEs who are founded in countries with high corruption levels might be less inclined to undertake social activities as they are discouraged to do so by their environment. On top of that, in countries with high corruption levels MNEs are inclined to undertake unethical practices such as child labor and bribery (Ioannou and Serafeim, 2012).
MNEs often deal directly with governments, because they have to bargain contracts or lobby activities. As a result state efficiency and bureaucracy, as well as the system of values and beliefs of the ruling party or coalition, affect the level of CSP that MNEs achieve (Rodriguez et al, 2005). Therefore I expect that MNEs originating from countries with lower levels of corruption will achieve higher CSP. Firstly because in environments of high corruption MNEs are more likely to engage in unethical practices to reduce costs or increase their market share. And secondly, countries with high corruption levels don’t provide incentives to MNEs for engaging in social responsible behavior (Ioannou and Serafeim, 2012). Corruption makes it difficult for MNEs to operate ethically and also brings along high transaction costs for MNEs who are dealing with corrupt environments. As stated in the literature review, CSR is known to positively influence MNEs in many areas. A MNEs motivation to participate in CSP can be intrinsic, extrinsic, or CSR can be used as an instrument to achieve certain performance objectives. Therefore, I expect that in corruption levels in firm’s home-country will have an impact on MNEs corporate social performance. I expect that high corruption levels in a MNEs home-country will result in low social performance. This leads to the following hypothesis
hypothesis 1: High corruption levels in an MNE’s home-country will negatively affect an MNE’s social performance
12
3.2 Industries and social performance
As discussed in the literature section, social performance has become an increasingly important concept for MNEs in the past decade. However, this is not equally so for all industries. For many years, media and governments have been regulating some controversial industries very closely (Durand, and Vergne, 2014). One of these controversial industries is the oil and gas industry. The oil and gas industry is controversial because there has been a persistent and widespread engagement in unethical business practices. These practices have entailed negative social consequences. Negative social and environmental consequences of the oil and gas industry are for example global warming caused by the production and use of oil, and deterioration in local air and water quality around petroleum refineries. Over the past several decades, oil and gas companies have been criticized harshly for issues ranging from environmental violations, human rights abuses, detrimental impact on local communities, to breaches of labor and safety standards (Du, 2012). Because of the general image of the oil and gas industry as portrayed above, this industry is generally seen as having a bad CSR record. Therefore I believe that the oil and gas industry will strengthen the negative relationship between corruption and social performance. Another argument as to why MNEs operating in the oil and gas industry will strengthen the negative relationship between home-country corruption levels and CSR is that the oil and gas industry is particularly vulnerable to problems related to corruption. As mentioned in the theoretical framework, MNEs who need to work closely together with local governments and governmental officials are affected by their corruption levels. As oil and gas companies need to engage with government officials frequently to negotiate, for example drilling rights, they will be likely to be even more affected by the corruption levels in this country. Therefore, companies that are headquartered in highly corrupt countries, and that - on top of that - are also operating in the oil and gas industry, should be particularly poor performers in terms of CSP. Therefore, one could expect that the negative main relationship in hypothesis 1 will be even stronger for countries in the oil and gas industry. This leads to the following hypothesis
hypothesis 2: The negative relationship between home-country corruption levels and MNE’s social performance will be stronger for MNEs operating in the oil and gas industry
As mentioned in the literature review, financial benefits can be a motive for MNEs to engage in CSP. Hull and Rothenberg (2008) have used this positive relationship in their study to test whether this positive relationship between social performance and financial performance was moderated by certain factors. One of the moderation effects they tested was the distinction between firms in differentiated industries versus firms in undifferentiated industries. Undifferentiated industries are industries that include products that are homogeneous, while differentiated industries include heterogeneous
13 products. They found that the relationship between financial performance and social performance was significantly stronger within undifferentiated industries. This means that financial performance as an incentive for MNEs to engage in social performance is stronger in undifferentiated industries than in highly differentiated industries. Hull and Rothenberg suggest that the effect of social performance is stronger for MNEs operating in undifferentiated industries because of the earlier stated positive effects of CSR like advertising, increase in demand, and decrease in consumer price sensitivity. They argue that these positive effects are stronger for undifferentiated firms than for differentiated firms. Other support for the stronger effect of CSR in undifferentiated industries is that they found that for MNEs in undifferentiated industries CSR added significantly to their competitive advantage, while this was not the case for MNEs in differentiated industries (Hull and Rothenberg, 2008). In other words, in differentiated industries, the social performance of an MNE makes little to no difference in their competitive advantage, while in low differentiation industries CSP is a means to a competitive advantage. Consequently, this suggests that companies in undifferentiated industries are more likely to engage in CSR as they have valid incentives to do so. MNEs in undifferentiated industries can achieve a better financial performance and gain a competitive advantage by engaging in CSR and are therefore more likely to engage in CSR activities, despite their home-country corruption levels. When looking at the ICB industry classification, the basic materials industry, and utilities industry classify as undifferentiated industries where almost homogeneous products are sold. Therefore I hypothesize the following
hypothesis 3: The relationship between home-country corruption levels and MNEs’ social performance will be weaker for MNEs operating in the basic materials industry
hypothesis 4: The relationship between home-country corruption levels and MNEs’ social performance will be weaker for MNEs operating in the utilities industry
3.3 CSP and the different world regions
For many scholars, CSR is a well-known and permanent concept within international business research. Since companies have started to market their social and environmental projects, for many consumers the CSR aspect of large MNEs is undeniably a part of their identity. Meanwhile CSR has influenced the globalization process. This influence stimulates the social development in emerging economies, but governments in these developing parts are not fully able to take care of the population and be responsible for its quality of life as in developed parts of the world (Soares Outtes Wanderley
14 et al. 2008). Developed nations such as the majority of the countries in Europe and the USA, implement actions that stimulate CSR development by MNEs. For instance, the European Commission declared 2005 as the year of CSR in countries of the European Union (Luetkenhorst, 2004). In these developed countries, a foundation for CSR in the form of institutions, standards and systems is in place (Kemp, 2001). In developed countries civil society is organized well, governments support CSR and press has taken on the role of watchdog for ethical and unethical performance of MNEs (Jamali, 2007). In the world CIA fact book, Europe rank as the highest developed continent and North-America ranks as second most developed continent. Therefore I believe that MNEs founded in Europe or North-America will have a better social performance in place as they are forced to comply with several regulations concerning CSR and on top of that stimulated to engage in socially responsible activities by their host-governments. This means MNEs from these continents will have a high social performance, despite their home-country corruption levels. This leads to the following hypotheses
hypothesis 5: The negative relationship between home-country corruption levels and MNEs’ social performance will be weaker for MNEs founded in Europe
hypothesis 6: The negative relationship between home-country corruption levels and MNEs’ social performance will be weaker for MNEs founded in the North America
A region that is scorn for its high corruption levels is Asia. Several economies in Asia, such as China and India, have been growing steadily for the past few years, but so have the corruption levels (Srirak Plilat, 2014). The 28 Asian countries included in Transparency Internationals’ corruption index in 2013 account for nearly 61% of the world’s population, but the majority has high levels of corruption. 18 out of 26 score have a corruption index score lower than 40 on a scale where 0 is highly corrupt and 100 is very clean. The level of perceived corruption in China is only rising, despite their efforts to battle corruption with public campaign called 'tigers and flies' (New York Times, 2014). Assuming hypothesis 1 is true; MNEs who have their home-country located in Asia should have an especially poor social performance score. This leads to the following hypothesis
Hypothesis 7: The negative relationship between home-country corruption levels and MNEs’ social performance will be stronger for MNEs founded in Asia
15 Africa is ranked as least developed continent in the world CIA fact book. Contrary to high developed countries, CSR initiatives have not encountered high interests in developing countries. Factors that prevent the development of CSR in developing countries are that civil society is not well organized in developing countries, the government doesn’t strongly promote CSR, MNEs do not face strong, constant regulation concerning CSR, and press has not taken the role of watchdog yet (Jamali, 2007). On top of that the development of a country is found to stagnate when high levels of corruption are present (Lawal, 2007). This suggests that low developed countries have high levels of corruption. Evidence for the presence of high corruption levels is found in the Transparency International corruption index where Africa is listed behind Asia as the continent with the highest corruption levels in the world. On top of that, developing countries are associated negatively with social performance (Jamali, 2007). To conclude, Africa is the continent with the most developing countries, low developed countries are associated with high levels of corruption and low social performance levels, and as a continent Africa is second corrupted. This leads me to believe, that Africa will have moderating effect that strengthens the negative relationship between corruption and social performance. Therefore I hypothesis the following
hypothesis 8: The negative relationship between home-country corruption levels and MNEs’ social performance will be stronger for MNEs founded in Africa
16
4 Methodology
4.1 Research Design
The overall design of the study is a cross-sectional design. This is a type of observational study that involves the analysis of data collected at one specific point in time. These data are specifically collected from a population, or representative subset. Cross-sectional designs involve the use of a cross-sectional regression, which I will use to test my hypotheses (Olsen and George, 2004).
To conduct this study, I will be using secondary data. This means I will be using existing data to answer my research question. The advantages of using secondary data is that it greatly reduces time and costs. This is really convenient as this thesis has a limited timeframe and no budget. Disadvantages of using secondary data are however that I will have no control over the study population, design, or measurements (Grady, et al., 2013). For the data collection I will need access to computers that have access to several large databases. I choose this type of research because in order to give a valid answer to the research question ‘how do corruption-levels in company's’ home-country affect the company's social performance?’ I will need a large research sample containing MNEs from different world regions and different industries.
4.2 Research Sample and data collection
As a research sample I have chosen to use the Global fortune 500 companies. The Global Fortune 500 is a list of the 500 largest MNEs in the world. The time period I will use in my study is the year 2013. I have chosen 2013 because I will be able to find the most recent and complete data for this year. 2014 would be more recent; however some data about this year might not be available yet. The data sources I will use for the secondary data collection are DataStream Professional, World scope and Asset4. DataStream is a database produced by Thomson Reuters. DataStream contains basic and financial information on companies (https://forms.thomsonreuters.com/datastream/). The Asset4 database is also developed by Thomson Reuters, and part of the DataStream Professional package. The Asset4 contains data on the ‘pillars’ of CSR: social, environmental, and corporate governance performance of companies. The three pillars of CSR are measured using several variables. Environmental performance is measured using resource construction, emission reduction, and product innovation. Social performance is measured using employment quality, health & safety, training & development, diversity & opportunity, community, and product responsibility. Corporate governance is measured using board structure, compensation policy, board functions, shareholder rights, and vision & strategy. For this study I will only use the Social Performance data as my outcome variable. As control variables for MNE size and profitability I will use data from World scope. From the DataStream, I collected data on International Securities Identification Number (ISIN) company identifier code, company name, Industry Classification Benchmark (ICB) Industry codes.
17
4.2.1 Dependent variable
As mentioned earlier, previous studies have encountered trouble while constructing a truly representative measure of CSP. Reasons for this are the multidimensionality of the theoretical construct, and because measurements of a single aspect of CSP provide a limited perspective on the firm’s performance in the more general social and environmental sense (Ioannou and Serafeim, 2012). By the multidimensionality of the CSP construct, they mean that social performance has multiple characteristics that need to be measured within one single measurement. Therefore, in this study I will use the global ESG dataset from the Tomson Reuters ASSET4. For this database, specially trained research analysts collected 900 evaluation points per firm, and according to guidelines all the primary data used were objective and publicly available. Data used for these evaluation points typically include CSR and annual reports, stock exchange filings, and other sources. The analysts transform this gathered ESG data into consistent units that allows them to make a quantitative analysis of this qualitative data. For social factors these data include employee turnover, injury rate, accidents, training hours, women employees, donations, and health and safety controversies. Based on these data points the ASSET4 index offers a comprehensive platform for establishing customizable benchmarks for the assessment of corporate social performance. All the 900 data points are used to come up with the several variables including the social performance score. For every year, firms will receive a new score based on the data gathered in that year. Therefore, just like the global fortune 500 and corruption index, we will also use the data from 2013 as they will be both recent and complete.
4.2.2 Independent variable
As you can see in table 1 the data for MNE home-country will be gathered from World scope. As mentioned above, the home-country corruption level data will be gathered through Transparency International’s website (http://www.transparency.org/cpi2013/results). Transparency International is the ‘global coalition against corruption’ and made a corruption index. By mapping corruption scores per country they aim to reduce corruption all over the world. They do this as they view corruption as the abuse of entrusted power for private gain, it hurts everyone who depends on the integrity of people in a position of authority. Each year they measure the perceived corruption levels in all the countries in the world.
The procedure I will follow to connect the right home-country corruption score to the right MNE will be as follows. Via World scope, every Global Fortune 500 company will be matched with the right home-country. Then I will copy the home-country list and replace a home-country with the corresponding perceived corruption level score that I will get from the Transparency International
18 website. I will select the perceived corruption level scores from 2013, as the rest of the data also are from the year 2013. I will continue replacing the home-country with the corresponding corruption level scores until every firm is matched with the right perceived country corruption level score.
4.2.3 Control variables
To make sure the dependent variable, social performance, is really affected by home-country corruption levels I will use control variables. These control variables will make sure that the relative impact of home-country corruption levels on social performance of MNE’s is found. In this study I will use a control variable for MNE size and MNE profit. MNE size is gathered as the variable total assets from World scope. The variable of MNE profit is acquired as return on assets, also from World scope.
4.2.4 Moderation variables
To test all hypotheses I will use moderation variables. Similar to the corruption level matching procedure I will connect the right region code to every Global Fortune 500 firm. For statistical purpose I will connect each region to a number. 1 = Asia, 2 = Africa, 3 = Europe, 4 = North-America, 5 = South-America, and 6 = Australia. Every home-country is again replaced by the corresponding region code.
Another variable I will use in this study are the ICB industry codes. ICB stands for Industry Classification Benchmark. ICB is a definitive system categorizing 70,000 companies and 75,000 securities worldwide, enabling the comparison of companies across four levels of classification and national boundaries. The ICB system is supported by the ICB database (http://www.icbenchmark.com/). For this study we will only use the first level of industry classification that classifies the Global Fortune 500 companies in one of the 10 different industries. The different industry classifications are: 0001 = oil and gas industry, 1000 = basic materials industry, 2000 = industrials industry, 3000 = consumer goods industry, 4000 = health care industry, 5000 = consumer services industry, 6000 = telecommunications industry, 7000 = utilities industry, 8000 = financials industry, and 9000 = technology industry.
As described in the conceptual framework, I want to test if different MNE regions and different MNE industries have an effect on the relationship between home-country corruption levels and the MNEs social performance. I will test this by creating dummy variables for the six firm regions, as well as the ten different industry classifications. By adding the dummy variable of a specific region to my
19 model, I will test if this region had a significant effect on the relationship between home-country corruption levels and social performance of MNEs. For example, I will test if this relationship is influenced for firms whom have their home-country located in Europe. The dummy variable will code all MNEs with a home-country coded as region Europe (3) with the number 1 and every firm with a home-country located anywhere else but Europe with the number 0. In this way I can test if the relationship of corruption levels on social performance of MNEs located in Europe is significantly affected by the fact that they are from Europe. I will apply the same procedure for creating dummy variables for the 10 different industries.
Table 1 variable definition, measures, data sources
Measure Measurement Source
ISIN Code International Securities Identification Number per company DataStream
Name Company name DataStream
Nation The company’s home-country World scope
Return on Assets Proxy for company profitability World scope
Total Assets Proxy for company size World scope
ICB Industry Code International Classification Benchmark industry code per company DataStream
20
5 Results
In this section I will explain the results of my study. I will evaluate them both statistically and theoretically by defining what every statistical outcome means for the predicted hypotheses. First, I will discuss the sample and the reliability of the sample, then I will discuss all the hypotheses and if they have a significant effect and what this effect means. In table 2 are the results of how many companies operate in which industry, and in table 3 the results of how many companies have their home-country in which region. Table 4 states variable statistics, Table 5 presents the complete correlation matrix, and table 6 captures the regression results.
5.1 Sample statistics
As mentioned earlier, I used the Global Fortune 500 companies list from 2013 as sample size. While searching for all 500 companies, for some companies there weren’t data available at all. When assembling the variables of total assets, return on assets, ICB industry code, and social performance score from data stream, some of these variables weren’t available for some companies. Eventually, I was left with a complete dataset for 350 companies (N = 350). For these companies all the variables were available. As you can see in table 2, all the industries are fairly represented in the sample. However, this is not the case for different world regions. 95.14% of all the firms have a home-country in Asia, Europe, or North-America. Also 0 MNEs have a home-country in Africa and therefore I will not be able to answer hypothesis 8. Sample wise, the results of this study are generalizable for the 10 different industries but not for the different world regions. The results will consequently be only valid for Europe, Asia and North-America.
Table 2 Sample distribution ICB industries
ICB industry Number of Firms Percentage of total N
Oil and Gas industry 40 11,43%
Basic Materials industry 23 6,57%
Industrials industry 53 15,14%
Consumer goods industry 50 14,29%
Health Care industry 18 5,14%
Consumer Services industry 45 12,86%
Telecommunications industry 16 4,57%
Utilities industry 16 4,57%
Financials industry 69 19,71%
21
Table 3 Sample distribution regions
Region Number of firms Percentage of total N
(1) Asia 116 33,14% (2) Africa 0 0,00% (3) Europe 105 30,00% (4) North America 112 32,00% (5) South America 9 2,57% (6) Australia 8 2,29% 5.2 Data statistics
The different variables used to test the hypotheses are home-country corruption levels, corporate social performance, total assets, and return on assets. In table 4 the statistics for the variables can be found. The variable Corruption Index, ranked home-countries accordingly to their corruption levels. Home-country corruption levels were ranked on a nominal scale of 0 to 100, where 100 represents no corruption at all and 0 represents total corruption. The variable corporate social performance was ranked on a continuous scale of 0 to 100, where 100 represents perfect corporate social performance and 0 represents no social performance at all. The variable of total assets is represented by every MNEs book value of total assets in 2013, which represents the size of every MNE. The variable of return on assets is represented by every MNEs book value of income divided by total assets in 2013, which produces the values for profitability of every MNE.
Before analyzing with these variables I needed to test them for normality, as normality is a criterion for your outcomes to be reliable. To see if a variable is normally distributed, one must look at the skewness value. This value is preferably between -0.8 and 0.8 (http://fmwww.bc.edu/repec/bocode/t/transint.html). As the table shows, corruption index and social performance are negatively skewed, and total assets is very positively skewed. Therefore the variables corruption index, social performance and total assets had to be transformed into a normally distributed variable (Cox, 2007). After transformation, the value of skewness for the transformed corruption index variable is--0.390, the value of skewness for the transformed social performance variable is -0.167, and the value for the transformed total assets variables is 0.498. These transformed variables and the return on assets variable used together with the dummy variables in the regression analysis.
22
Table 4 Descriptive statistics variables
Variable min max mean std. dev skewness
Corruption Index 28.00 91.00 68.6886 13.60864 -1.424
Social Performance 3.63 97.15 73.4103 23.43096 -1.330
Total Assets 1527077 3120067181 219305314 449575547.6 3.909
Return on Assets -22.64 24.40 4.7073 5.30214 0.446
5.3 Correlation Results
For correlation results for the full sample, for all specifications, we estimated the coefficients using Pearson correlation. Pearson correlation tells us whether two variables have a significant relationship with each other. This relationship can be either positive or negative and the value is in between 1 and -1. The value 1 represents perfect positive correlation which means two variables move in the exact same direction, 0 means there is no relationship between to variables at all, and -1 represents perfect negative correlation which means two variables move in the exact opposite direction.
In table 5 we can find the correlation matrix. We find that CSP and home-country corruption levels have a significant positive relationship with each other, which means that MNEs whom have a home-country with low corruption levels tend to have a better social performance. This result is consistent with hypothesis 1. Also, home-country corruption levels have a significant positive correlation with regions Asia and South-America, and the oil and gas industry. Home-country corruption levels have a significant negative correlation with regions Europe, North-America and Australia. This means that MNEs that have home-countries with high corruption levels are more likely to be found in Asia, South-America and the oil and gas industry, and that MNEs that have home-countries with low corruption levels are more likely to be found in Europe, North-America and Australia. On the other hand, CSP has significant positive relationships with Asia and North- America, and a significant negative relationship with Europe. This tells us that MNEs with little social performance are more likely to be found in Asia and North-America, and MNEs with a good social performance are more likely to be found in Europe.
23
Table 5 Correlation Matrix
24
5.4 Regression Results
In table 6 you can find the regression results. In total seven different regression models were run, of which six models tested the hypotheses. In table 6 the coefficients of all the variables with CSP are displayed together with their significance.
5.4.1 Home-country corruption levels
In model 1 (table 6), a regression was run using only control variables size and profit as independent variables and CSP as dependent variable. This model was not significant as F = 2.573 and p > 0.05.
R2 = 0.015, which means 1.5% of the effect is explained by the control variables. We therefore can conclude that our control variables do not significantly predict the outcome variable Social Performance.
In model 2, a regression was run to predict MNEs’ social performance from home-country corruption levels. Home-country corruption levels statistically significantly predicted CSP as F = 3.787 and p < 0.05. However, R2 = 0.032 which means only 3.2% of the model is explained by this regression effect. Therefore we can conclude that even though home-country corruption levels explains very little of the total effect, it does predict CSP significantly which is consistent with hypothesis 1.
5.4.2 Moderation by industry sector
In model 3 a multiple regression was run to predict CSP from corruption levels, oil and gas industry dummy variable, and the interaction variable of corruption levels and the oil and gas industry dummy. This model was run to test if operating in the oil and gas industry has a moderating effect on the relationship between home-country corruption levels and CSP. Even though the model statistically predicts CSP, F = 2.761 and p < 0.05, and the coefficient is negative, the oil and gas industry has no significant effect on the relationship between corruption levels and CSP. R2 = 0.039, which means the dummy variable only explains 0.7% more of the model than without the dummy variable. These results are inconsistent with hypothesis 2, which is therefore untrue.
In model 4 a multiple regression was run to predict CSP from corruption levels, basic materials industry dummy variable, and the interaction variable of corruption levels and the basic materials industry dummy. This model was run to test if operating in the basic materials industry has a moderating effect on the relationship between home-country corruption levels and CSP. Even though the model statistically predicts CSP, F = 2.318 and p < 0.05, and the coefficient is negative,
25 the basic materials industry has no significant effect on the relationship between corruption levels and CSP. R2 = 0.033, which means the dummy variable only explains 0.1% more of the model than without the dummy variable. These results are inconsistent with hypothesis 3, which is therefore untrue.
In model 5 a multiple regression was run to predict CSP from corruption levels, utilities industry dummy variable, and the interaction variable of corruption levels and the utilities industry dummy. This model was run to test if operating in the utilities industry has a moderating effect on the relationship between home-country corruption levels and CSP. Even though the model statistically predicts CSP, F = 3.273 and p < 0.01, and the coefficient is negative, the utilities industry has no significant effect on the relationship between corruption levels and CSP. R2 = 0.045, which means the dummy variable only explains 1.3 % more of the model than without the dummy variable. These results are inconsistent with hypothesis 4.
5.4.3 Moderation by world region
In model 6 a multiple regression was run to predict CSP from corruption levels, region Europe dummy variable, and the interaction variable of corruption levels and the region Europe dummy. This model was run to test if MNE home-countries located in world region Europe would have a moderating effect on the relationship between home country corruption levels and CSP. The model statistically predicts CSP, F = 18.488 and p < 0.000, and the coefficient is negative. CSP is significantly affected by the moderation effect of region Europe. R2 = 0.212, which means the dummy variable explains 18% more of the model. This means that the relationship between home-country corruption levels and CSP is weaker for MNEs who have their home-country in region Europe, which is consistent with hypothesis 5.
In model 7 a multiple regression was run to predict CSP from corruption levels, region North-America dummy variable, and the interaction variable of corruption levels and the North-North-America dummy. This model was run to test if MNE home-countries located in world region North-America would have a moderating effect on the relationship between home country corruption levels and CSP. The model statistically predicts CSP, F = 9.762 and p < 0.000, and the coefficient is negative. CSP is significantly affected by the moderation effect of region North-America. R2 = 0.102, which means the dummy variable explains 7% more of the model. This means that the relationship between home-country corruption levels and CSP is weaker for MNEs who have the region North-America as their home-country, which is consistent with hypothesis 6.
26 In model 8 a multiple regression was run to predict CSP from corruption levels, region Asia dummy variable, and the interaction variable of corruption levels and the region Asia dummy. This model was run to test if MNE home-countries located in Asia would have a moderating effect on the relationship between home country corruption levels and CSP. The model statistically predicts CSP, F = 4.414 and p < 0.001, and the coefficient is positive. However, CSP is not significantly affected by the moderation effect of region Asia. R2 = 0.060, which means the dummy variable explains 2.8% more of the model. This means that the relationship between home-country corruption levels and CSP is not significantly stronger for MNEs who have region Asia as their home-country, which is inconsistent with hypothesis.
The multiple regressions for model 9 couldn’t be run because no MNEs from the sample have their home-country in region Africa. Therefore hypothesis 8 couldn’t be tested.
27
Table 6 Regression Analysis
28
6 Discussion
6.1 The link between corruption and CSR
Prior research has stated that different factors affect the social performance of MNEs (e.g., Campbell, 2007). However, besides Ioannou and Serafeim (2012), no empirical studies have explored how CSP is influenced by external factors across more than 25 countries worldwide. Also few studies exist where social performance of MNEs is the dependent or outcome variable, as many scholars focused on testing CSRs’ impact on other factors using it as the independent variable (Egri and Ralston, 2008). In this study I provide empirical evidence for the profound role home-country corruption levels play in the CSP of MNEs. I achieved this evidence utilizing the ASSET4 dataset from Thomson Reuters for social performance scores of the Global Fortunes’ largest MNEs, and Transparency Internationals’ corruption index for corruption scores of the MNEs home-countries covering 350 firms and 34 countries. Drawing from the three lenses theory, and previous literature on corruption and CSP, my overarching hypothesis is that high corruption levels in an MNE’s home-country negatively affects and MNE’s social performance. My results provided evidence for the negative influence of high corruption levels in an MNE’s home-country on an MNE’s CSP. These findings are consistent with prior literature. Ioannou and Serafeim (2012) also found evidence for the negative relationship between corruption and CSP, and these results are also consistent with the fact that MNEs social performance is influenced by national external factors as stated by Campbell (2007). As earlier mentioned, corruption is very closely intertwined with the government of a certain country. Therefore, this study also suggests that MNEs and their social performance is indeed influenced by institutions such as the government in their home-country (Jackson and Apostolakou, 2010)
One of the moderation variables used in this study was industry classification. For the industry classification of the MNEs I used ICB codes. All the different industries were represented by 16 MNEs or more in the sample. Using prior literature on corruption and CSP within different industries, I drafted three hypotheses concerning the possible moderating effects of industry on the relationship between home-country corruption levels and an MNE’s social performance. The second hypothesis stated that the oil and gas industry strengthens the negative relationship between home-country corruption levels and an MNE’s social performance. Including the oil and gas industry as a moderator didn’t have a significant impact on the significant negative relationship between corruption and CSP. However, the oil and gas industry did significantly and positively correlate with corruption, which is consistent with prior literature on the relationship between corruption and the oil and gas industry (Du, 2012; Mauro, 1995). This positive correlation suggests that maybe the sample of MNEs from the oil and gas industry wasn’t large enough to significantly influence the relationship between corruption and CSP.
29 The third and fourth hypotheses were also drawn using industry as a possible moderation variable. MNEs from undifferentiated industries would weaken the relationship between corruption and CSP, however this was for both the basic materials industry and utilities industries found to be untrue. Since both industries also had almost no correlation with CSP at all, the evidence provided in this study is inconsistent with literature on undifferentiated industries and CSP (Hull and Rothenberg, 2008). An explanation might be that the proposed incentives are simply weaker than the influence of corruption on CSP. However, since no literature is available on this particular subject we can’t be sure if this explanation for the inconsistent results is valid.
The other moderating factor that was included in this study is moderation by world region. Europe and North-America were found to significantly weaken the relationship between corruption levels and CSP. This is consistent with prior literature that both Europe and North-America have well developed regulations and standards concerning CSP of MNEs, and both regions contain only developed countries (Kemp, 2001). Both literature and the results from this study suggest that MNEs from these regions are stimulated to achieve a certain social performance level, irrespective of their home-country corruption levels. On the other hand, I hypothesized that having a home-country in the regions Asia or Africa would strengthen the relationship between corruption and CSP. Even though region Asia did significantly affect the social performance score negatively, the moderation effect of region Asia and home-country corruption levels was not significant. This goes against stated literature, that corruption in Asia is high compared to the rest of the world and CSP of MNEs with a home-country in this region should be affected stronger (Srirak Plilat, 2014; New York Times, 2014). So the results suggest that MNEs from Asia generally have a lower social performance score, but that this is not explained by the home-country corruption levels of this MNE. As mentioned earlier, the last hypothesis couldn’t be tested as the sample didn’t include any MNEs with a home-country located in region Africa.
6.2 Theoretical implications
This study contributes to the stream of literature that suggests the social performance of MNEs is affected by external factors (Jackson and Apostolakou, 2010; Boddewyn, 1988). The results suggest that a MNEs social performance is affected by home-country corruption levels, as well as the possible influence of world region on the social performance level of MNEs. The research also contributes to the emerging stream of literature concerning the three lenses that influence MNEs, corruption, CSR and politics. Both Luo (2006) and Rodriguez et al. (2006) emphasized the importance of research on the interaction of the three lenses. Their concern was the development of three separate streams of literature on the three different lenses that lacked research on the possible effects of one of the other
30 lenses. This study contributes to research done on the interconnectedness of the corruption lens, and the CSR lens. Obviously this study also contributes to research on the topic CSP of MNEs (Jamaili, 2007; Tan and Wang, 2011; Yang and Rives, 2009; Egri and Ralston, 2008). As stated before, the study provides evidence that the CSP of MNEs is influenced by home-country corruption levels. This result suggests support for negative duty motivation approach, explained in the literature review, as stated by Swanson (1995). Finally, the evidence provided by this research supports the statement of Giddens (1984) that MNEs will refrain from social performance activities when operating in a corrupt environment.
6.3 Managerial implications
The intention of this study was to provide a deeper understanding of the relationship between corruption and CSP. In particular this study was intended to clarify the influence of a MNEs home-country corruption levels on their social performance. MNEs can use the results of this study to re-evaluate their social performance activities and conclude if they indeed are influenced and prevented from engaging in CSR because of their corrupt home-countries. On the other hand, governments can use the results of this study to take a closer look at their own regulation regarding CSR of MNEs. One implication of this study is that if governments reduce their levels of corruption, the MNEs from their country might improve their social performance. Another implication that is subject to the results of this study is that the oil and gas industry and the undifferentiated industries of basic materials and utilities don’t moderate the effect of corruption levels on CSP. Rather, managers should look at the world region their MNE is operating in. This study suggests that MNEs from Europe and the USA are not significantly influenced in their social performance by their home-country corruption levels. Again, governments in other regions could look at, and learn from the regulation both Europe and the USA have regarding the CSR of MNEs. But the most important implication managers should subtract from the results of this study is that a MNEs’ CSP is influenced by external factors. In this particular study I provide evidence for the influence of corruption levels on social performance, but this study also suggests that world region and in particular the regulation concerning social performance in this region is of influence on the social performance of MNEs.
6.4 Limitations and suggestions for future research
The first and most important limitation of this study is the research sample. Though the sample did include proportionally enough MNEs from every ICB industry category, the sample didn’t include MNEs from every world region. South-America and Australia were very under represented, and for
