Master Thesis
The relationship between Degree of
Internationalization in low CSR contexts and
Corporate Social (ir)Responsibility
Name: Caspar Tijssen
Student number: 10726403
Supervisor: Dr. Alan Muller
Date: 16/03/2015
Master: Business Administration
Track: International Management
2
Abstract
In this study, the relationship between degree of internationalization, geographical contexts in which multinational enterprises (MNEs) operate and Corporate social (ir)responsibility is researched. Using panel data of 407 firms over 10 firm years, this thesis aims to provide evidence for the statement that the corporate social responsibility of multinationals is affected by their degree of internationalization and the contexts in which they do business. Using generalized estimating equations in combination with binary logistic regression analysis, the results find support for a relationship between degree of internationalization in low CSR contexts and corporate social responsibility. No support was found for a direct relationship between the degree internationalization and an MNEs CS(i)R. Also the CSR context in which a firm operates does not seem to be related to social irresponsibility. The relationship among these variables might be investigated further in future research using different measures and/or analyses. One of the limitations is the difficult distinction between low and high CSR contexts in the data, which may affect validity.
Keywords: Corporate social (ir)responsibility; Multinational Enterprises (MNEs); Internationalization; CSR contexts
3
Contents
Introduction...4
Literature review ...7
Corporate Social Responsibility ...7
Variation in CSR activities ...8
CSR dimensions ...8
CSR strengths and concerns ...9
Geographical dispersion ...9
Conceptual model ... 13
Data and Methods ... 14
Variables ... 15 Methods ... 16 Results ... 17 Descriptives ... 17 Correlations ... 18 Regression ... 19 CSiR ... 21 Discussion ... 24
Limitations and further research ... 25
Conclusion ... 26
4
Introduction
Last year, Facebook introduced a new mobile app to offer basic internet services for free in developing countries. Later that year Google announced a project called ‘Android One’. This line of relatively cheap smart phones targets developing markets like Asia and Africa. It is just a fraction of multinationals that offer their products or spin-offs of that products as a cheap alternative to poor countries. Subsequently, those multinationals very often deploy development programs for the local community. Apart from increasing their sales, a motive may be to make the basic needs from today available to poorer societies and help them to be more self-sufficient.
Corporate Social Responsibility (CSR) has become an increasingly important topic for companies, not in the least for multinationals. Various stakeholders like customers, shareholders, and governments exert pressure on these firms. The fact that companies become increasingly global, having activities in a growing number of countries, make the stakeholder base of a company even more diverse and harder to deal with. Many scholars have emphasized the way that social performance is influenced by a firm’s interdependence with international markets. For example, Johnson & Greening (1999) highlight that the increasing competition in the global market means an increase in accountability that is expected from multinationals. Subsequently, Hillman & Keim (2001) state that firms face pressures to help solve problems worldwide due to globalization. As a result, CSR and its potential links to and relationships with other topics has received growing attention from scholars.
The relation between CSR and this geographical dispersion is more extensively researched recently (e.g. Strike et al, 2006; Brammer et al, 2006). An often used approach is the stakeholder perspective in explaining the link between the firm and the society it operates in. Because an MNEs’ activities are dispersed over the world, its stakeholders base is consequently larger and more diverse than that of a firm that operates solely domestic. This geographical dispersion influences the way firms act. Landier et al (2007) found a negative relation between geographically dispersion and employee friendliness. Distance plays an important role because it is often associated with information asymmetry (Coval & Moskowitz, 1999) causing a deterioration of information quality for a decision maker. Other scholars, like Kostova and Zaheer (1999) and Bartlett and Ghoshal (1989), state that the extent and nature of social activities of multinationals are influenced by the extent to which it is geographically dispersed (or internationalized). The firm is forced to either export their domestic strategy, adopt foreign standards and practices (isomorphism) or even develop an entirely new social strategy (Donaldson & Dunfee, 1999).
The opinions on whether multinationals act mainly responsible or if they are especially irresponsible are diverse. In this globalizing world, there are critics that argue that MNEs take advantage of deprived local conditions, for economic reasons (Bouquet and Deutsch, 2008). They are accused of abusing them to engage in activities like child labour, bribery, corruption etc. (Osland,
5 2003). Low & Yeats (1992) also argue that MNEs are believed to exploit the low social and environmental standards in economies that lack a strong regulatory environment. An example is the pollution haven hypothesis which states that companies relocate their polluting activities into countries with low environmental regulations as a result of increasing regulations in their domestic context. Although this is an often heard argument, evidence is mixed. Low and Yeats (1992) for example provide evidence of the hypothesis, but there are also studies that state that firms become more conscious and invest in innovating ways for cleaner production rather than moving their activities to countries with low standards (e.g. Letchumanan & Kodama, 2000).
However, an increasing number of organizations is engaging in more extensive CSR programs in their decision making all over the world (Dunning, 2003). The worldwide discussions about unethical companies contribute to this trend. Scholars argue that firms transfer their best practices to developing countries which in turn could improve social standards (Bansal & Roth, 2000; Christmann, 2004). Sharfman et al (2004) suggest that MNEs are pushed to higher social performance by pressures from stakeholders like global and domestic institutions. Bouquet and Deutsch (2008) focused on the effects that social activities may have on a firm’s ability to be successful across borders. Conversely they find that high levels of CSP have a positive effect on a firm’s multinationality.
Strike et al (2005) have attempted to bridge the opposing arguments by arguing that MNEs can both be responsible and irresponsible at the same time. Using the resource-based view, they state that firms on the one side create value by engaging in responsible activities and on the other side destroy value by engaging in irresponsible activities. They further argue that when firms act irresponsible this does not necessarily imply the company’s ill will, but that financial pressures can be the sole reason for irresponsible behavior. MNEs often have a great level of dispersed activities and are not always willing to deploy a great amount of resources and capabilities to avoid irresponsible activities. Thus, Financial pressures and refusing to use great amounts of resources may imply that MNEs with activities in low CSR contexts have a lower corporate social performance than companies that focus on high CSR contexts, like the Western economies of North America and Western Europe.
When looking at implications for CSR on overall firm performance, … et al (…) find that both highly responsible firms and firms that ignore CSR to a great extent, can do well in foreign countries. An intermediate level of social activities has no effect on the performance. CSR also has its implications for financial performance. A great number of scholars focused on this relation (e.g. …) and even though there still is no consensus on whether this relationship exists or what this implies, the evidence seem to show a positive relationship between the two variables.
In a working paper, Attig et al (2013) used firm internationalization as a starting point attempting to find evidence between internationalization and the CSR ratings of a firm. Among others, they found that firm internationalization is positively related to the corporate social performance.
6 The concepts of internationalization and CSP thus seem to be highly intertwined. Studies on the influence of CSP on multinationality focus especially on the quality of multinationality due to this performance, whereas research on the effect of geographical dispersion on social performance mostly focuses on the implications that multinationality has on social activities of an MNE.
Important to note is that CSP has several dimensions, like activities regarding the environment, community, corporate governance and employee relations. Brammer et al (2005) found that dispersed activities of UK firms have different implications for both the several dimensions of CSP and the countries where the activities are located (i.e. there is a difference between activities located in western Europe and eastern Europe). Taking previous research and its implications into account has led to the following research question that will be addressed in this thesis:
What is the effect of the degree of internationalization on corporate social (ir)responsibility and how does presence in low CSR contexts affect this relationship?
As stated, interest in the relation between geographic dispersion and corporate social performance has been growing in the recent years. However, the amount of studies on this field is still small and also provides mixed evidence. Simerly & Li (2000) do not find a statistically significant link between CSR and internationalization whereas Brammer et al (2006) and Kang (2013) did find a positive relation between international diversification and social activities (Attig et al, 2013).
By using a sample of US firms over the years 2004-2013 and an extra focus beyond
internationalization or geographical dispersion, the addition of the CSR context (i.e. high or low), this paper aims to contribute to the existing literature on the relationship between the degree of
7
Literature review
The literature is reviewed to create a better understanding of the concepts of CSR and geographical dispersion. This is needed in order to develop the hypotheses that can answer the research question.
Corporate Social Responsibility
Although a lot of organizational scholars have written and did research on the topic of CSR, there still is little consensus on what the term really means. The result is that there is no, and perhaps never will be, a completely universal definition on what CSR really means. A relatively old definition is from Frooman (1997) which describes CSR as “an action by a firm, which the firms chooses to take, that substantially affects an identifiable social stakeholder’s welfare.” Aguinis (2011) however, argues that CSR consists of organizational actions and policies that depend on the context, which takes stakeholders’ expectations into account and state that it focuses on three aspects; economic, social and environmental performance. The one that is cited often in academic papers on the topic of CSR is from McWilliams and Siegel (2001), who define CSR as “actions that appear to further some social good, beyond the interest of the firm and that which is required by law.” They especially state that CSR is more than just following the law or broadly accepted norms.
As for CSR within firms, and especially multinational enterprises (MNEs), Tsoutsoura (2004) argues that MNEs that engage in CSR should go beyond minimum legal requirements when adopting business practices and policies to contribute to their stakeholders’ welfare. The author looked at CSR like a set of practices, policies and programs that are integrated into a business’ operations and processes like decision-making and their supply chain. They should include business ethics, investment in communities, aim at environmental goals, human rights, etc. These dimensions are all included in the Kinder, Lydenberg and Domini (KLD) database, which will be used in this thesis. In the same paper it is stated that every firm differs in how it engages in CSR activities. This depends on factors like the size of the company, the industry it is involved in, the culture within the organization, the demands of stakeholders and to what degree they are committed to CSR activities. There are firms that engage in specific, isolated CSR activities and there are firms that integrate aspects of CSR in all their operations and throughout all their organizational processes. The MNE’s values and strategic planning seem to be key as well as the commitment of the management and employees of the firm.
Reasons to engage in CSR can be driven by both intrinsic and extrinsic factors. Extrinsic drivers in particular can be pressures like shareholder demands, regulation or media pressure (Muller & Kolk, 2009). Intrinsic drivers, however, mean that CSR is a goal on its own and is driven by morality (Carroll, 2000; Lindenberg, 2001) and focus mainly on managerial motivations (Heugens et al, 2008).
In their study in which they examine these roles of both extrinsic and intrinsic factors as a determinant for corporate social performance (CSP), Muller and Kolk (2009) use Barnett’s definition
8 (2007): “A snapshot of a firm’s overall social performance at a particular point in time”. A more extensive definition of CSP is found in Ioannou and Serafeim (2012). They wrote a paper about what drives corporate social performance and especially focused on the role that nation-level institutions have in the matter. In their study they used a definition of CSP (e.g. Hillman & Keim, 2001; Waddock & Graves, 1997) that sees the concept as “a business organization’s configuration of principles of social
responsibility, processes of social responsiveness, and policies, programs, and observable outcomes as they relate to the firm’s social relationships” (Wood, 1991: 693). CSP therefore can be seen as how well
(or how bad) companies act regarding corporate social responsibility.
Variation in CSR activities
MNEs can vary when it comes to their CSR activities, and also practices across countries can differ significantly. Maignan and Ralston (2002) studied firms in France, the UK, the US and the Netherlands and their extent to which they exposed their CSR activities on their corporate website. Their findings were that there was a systematic and significant difference across countries on topics like stakeholder pressures and the incentives that managers had to engage in social and environmental programs. Another study, focused more on Asian countries like Malaysia, South Korea and Singapore, Chapple and Moon (2005) for example found that multinational enterprises adjust their CSR activities mainly to the context(s) in which they operate. Furthermore they found that the differences among countries could not be explained by the countries’ stage of development, but suggest national factors could explain cross-country variation. Aguilera et al (2007: 836) contribute to these nation-specific pressures by comparing the Anglo-Saxon and the European model of corporate governance. They state that organizations will experience different degrees of pressures in engaging in socially responsible initiatives, both internal and external, because these firms are embedded in different national systems.
Another, more recent contribution to literature on this topic is made by Jackson and Apostolakou (2010) who researched the influence that the institutional environment has on CSR activities of European MNEs. They found that firms that engage more in Anglo-Saxon countries (thus liberal market economies) achieve higher levels of performance in CSR, than countries that engage in markets that are more coordinated by the market; continental Europe. Their study provides evidence that institutions influence CSP in national business systems.
CSR dimensions
A reason for ambiguous results on the relation between social and firm performance lies in the multi-dimensional construct that CSR is. CSR policies include social, environmental and business behavior and according to Cavaco and Crifo (2014) there is a possible quantity-quality tradeoff between the different dimensions. Quantity reflects the effect of isolated dimensions and dimensions together,
9 while quality is associated with the interactions between the different dimensions. Using a single CSR dimension to measure the impact of the CSR as a whole on firm performance therefore may lead to other results than treating the dimensions separately (Surroca et al, 2010). Therefore, Barcos et al (2013) suggest that CSR needs to be treated as multi-dimensional in order to study its possible diverging impact on performance. Brammer, Pavelin and Porter (2006) provided evidence for this suggestion by finding that geographical diversification is associated with both environmental and community performance to a certain extent, while that same diversification does not impact employee-related aspects of social performance. Therefore, one may conclude that the relationship between CSP and geographical diversification varies across the several dimensions of the CSR construct.
CSR strengths and concerns
Apart from the several CSR dimensions, CSR measures often report strengths and concerns. The KLD database reports CSR along seven dimensions each divided by strengths and concerns. A firm can be rated as having strengths in environmental activities but at the same time have concerns regarding that same topic. Distinguishing between strengths and concerns can help to understand the impact of social performance on firm performance (Strike, Gao and Bansal, 2006). Furthermore, several studies showed that both responsible and irresponsible variables (i.e. strengths VS concerns) are conceptually different, which evidently means that the opposites should be treated accordingly. Having more strengths is not the same as having less concerns and vice versa.
Geographical dispersion
There are several definitions and formulations of geographical dispersion. It can be referred to as internationalization, multinationality or geographical dispersion. In this study geographical dispersion is defined as “the extent to which a company achieves profitable sales outside its domestic market” (Hitt et al, 1997; Kotabe et al, 2002).
Academics have focused on the relationship between CSP and geographical dispersion in different ways. Bouquet and Deutsch (2008) found a U-shaped relationship between CSP and multinationality, which implies that, in order to achieve satisfying sales and profits in the foreign market, companies need to achieve high levels or no very low levels of CSR. In addition, MNEs that engaged at an intermediate level of CSR, achieve lower levels of performance in multinationality than firms that either have very low or very high levels of CSR engagement. Another, reversed study, focused on the implications that internationalization have for a firm’s CSR ratings. In this paper, Attig et al (2013) found a positive relation between a firm’s level of internationalization and CSR ratings.
10 Rugman (1997) found that firms engage in international dispersion in order to exploit opportunities and market imperfections that characterize these countries. According to Shakrokh (2002) firms can create or increase value by expanding to foreign countries when the pros outweigh the cons (i.e. gains are expected to be higher than the costs). The research that has been done in this area often suggests that MNEs are pushed to a higher social performance by pressures from stakeholders like global and domestic institutions (Sharfman et al, 2004). Note that it are the MNEs that dominate the industries associated mostly with pollution and the benefits of the markets are often cost-driven (Rugman & Verbeke, 1998). As a result MNEs are often considered irresponsible because they search for lower costs. However, these findings are often anecdotal and not so much statistically supported (Rugman & Verbeke, 1998).
A different reason why companies are driven to higher social performance by stakeholders is caused by expanding to a higher number of foreign markets. This subsequently tends to increase in the diversity of the stakeholders base and variations in social systems, legal environment etc. Every country has its own norms and values and ways of doing business, which pressures the firm to conform in some sort of way.
For MNEs to enhance CSR, the ‘normative appeal’ (Zyglidopoulos, 2002), is clear. However, Pava and Krausz (1997) argue that to enhance a firms’ CSP also involves a set of costs for attending, which can be very high for multinationals, like the adoption of global ethical codes of conduct (Kaptein, 2004). Complying to these codes would require specials efforts for corporations. They do not only have to mind their own CSR practices, but also take those of their business partners into account (Kolk & van Tulder, 2002), like suppliers and foreign subsidiaries or joint venture allies. The Economist (2005) even states that MNEs should at least have a senior executive with the sole purpose of developing the CSR function and coordinate that function in all their locations across the world.
Previous studies found pros and cons for using CSR to achieve high performance in multinationality. Lewin et al (1997) found strong evidence that many foreign stakeholders value certain attributes of CSR. Furthermore, Kostova and Zaheer (1999) show that countries that have a far going global CSR program face less legitimacy issues in foreign markets. Bouquet and Deutsch (2007) even state that investing in CSR can lead to “moral capital” and that it improves the image of an MNE in comparison to competitors. According to these authors, this can be seen as an ownership firm-specific advantage (FSA) (Dunning, 1981) because it gives opportunities for doing business in a foreign country. Nachum & Zaheer (2005) say that internationalization can be a strategy to maintain or even increase a firm’s competitive advantage. Reasons lie in economies of scale and scope (Kogut, 1985), benefits of diversification (Geringer, Beamish & DaCosta, 1989), etc. The downside however lies in the fact that internationalization also increases complexity and uncertainty, especially the pressure from new and more diverse stakeholders with other values and cultures.
11 So, although multinationality or internationalization has a lot of positive implications for a firm, it also has its downsides. Because it can be seen as a location-bound FSA, an MNE can find it hard to transfer these to foreign subsidiaries, like a firm’s reputation (Bromley, 1993). It might even be harder to implement all these CSR practices in all countries that an MNE operates in. This also depends on the type of foreign subsidiary. The transferability of FSAs may be harder for Greenfields than they are for joint ventures in the host country. However, this topic is outside the scope of this study.
It takes a lot of engagement from a firm’s management to implement all these practices everywhere. Especially because one CSR activity may not work as good in one country as it does in the other, which influences the overall firm performance and obviously the financial performance. The more countries an MNE operates in, the more difficult it may be to keep CSR practices manageable. In other words, the degree of internationalization may influence the performance of CSR activities across countries. This leads to the first hypothesis:
H1a: The higher the degree of internationalization of an MNE, the lower the corporate social responsibility.
In addition to that, one might expect that because it is harder to implement and manage these practices everywhere, this will not only force pressure on the positive sides of CSR, but may also influence and strengthen the negative sides, the social irresponsibility:
H1b: The higher the degree of internationalization of an MNE, the higher the corporate social irresponsibility.
Brammer, Pavelin and Porter (2006) found that there are differences in the relationship between geographical dispersion on the one side and environmental, community and employee-related performance on the other side. Geographical dispersion is for example associated with positive community performance, however, this only holds as long as the firms operations did not extend to Eastern Europe. That conclusion might be drawn because Eastern Europe can be seen as a low CSR context, or at least lower than the western context. The same conclusion could be drawn for environmental performance. This implies that firms apparently do better in terms of community and environmental aspects as long as they do not engage in eastern Europe, or: MNEs especially do good in social performance as long as the activities focus on western markets.
Because of this findings it might be interesting to look if the same conclusions could be drawn when the MNE’s activities do reach beyond Western contexts. As shown before, there is a discussion about whether firms expand to developing countries (i.e. low csr contexts) for the sole reason of cost reductions and lack of regulatory forces and not care for the social impact they leave behind, or on the
12 other hand that the presence of an MNE in a low CSR context actually means that they are doing good in terms of social performance.
Earlier in this review it is described how several authors think about what the impact of MNEs in developing countries is. Here was also showed that big firms might be pushed to do good in developing markets by forces in their domestic context. Think of the amount of criticism there has been on H&M and other clothing retailers for the deprived labor circumstances in their factories in countries like Bangladesh. Or Apple that faces accusations of violating human rights because of their partnership with Foxconn in China. Although they do not own this company, they do a lot of business with them (production of the iPhones and iPads) and their reputation suffers from it. Therefore, one might expect that presence in low CSR contexts (i.e. developing countries) causes pressures from regulatory and other forces in their home country and therefore may lead to more responsibility. This leads to the second hypothesis:
H2a: The impact that the degree of internationalization has on CSR is moderated by whether or not
an MNE has operations in low CSR contexts
That same moderation is hypothesized for the relation between degree of internationalization and corporate social irresponsibility, because being active in low CSR contexts can also affect the probability of being socially irresponsible.
H2b: The impact that the degree of internationalization has on CSiR is moderated by whether or not
13
Conceptual model
The hypotheses that are described above are below presented in a conceptual model, which gives an overview of the hypothesized relationships between the variables in this study. The figure visualizes the impact of degree of internationalization on corporate social (ir)responsibility and the moderating role of having operations in low CSR contexts. The hypotheses are interpreted under control of the variables R&D intensity, advertising intensity, firm size, financial performance and industries in which firms are active.
Degree of internationalization Corporate social responsibility Corporate social irresponsibility Operations in low CSR contexts H1a (-) H1b (+) H2a H2b
14
Data and Methods
This study uses a sample of 407 multinationals that all reported foreign sales in non-western countries between 2004 and 2013 and were also rated by KLD regarding their social performance. The majority of the firms is US based. The raw data was collected from two different sources. CompuStat was used to collect data on all financial data and the KLD database was used to collect data on concerns and strengths of CSR.
In the process of data cleaning, a large number of observations is excluded from the database because of missing data on either KLD data or CompuStat financial data, needed for the control variables in this study. The biggest part of cleaning the data was done in Excel. The CompuStat data and KLD data were merged mostly by the VLOOKUP function in Excel, which can be used to match various data with the same identifier (i.e. TIC codes of companies in the sample). An additional cleaning was also largely done in Excel; missing values were identified by making Pivot Tables and filter out the empty cells, which would be missing values in SPSS. Using this method for cleaning left the data without any missing values. The subsequent steps like descriptive data and developing a correlation matrix and the regressions were done in SPSS. In order to make the data available for SPSS this needed to be cleaned very thorough.
The distinction of high and low CSR contexts was done manually because there is no easy way to do that. CompuStat reports segment names, but these classifications are not uniform. For example, some segments are classified as EMEA (Europe, Middle-East and Asia), and others were reported as All
non-US countries. This made it hard to make a distinction. Therefore, when there was no clear
classification of the segment name, the observations was thrown out of the data, retaining the observations for which a reasonable distinction between high and low CSR contexts could be made. However not that this is just a quick and dirty solution of sorting the data, which has implications for the validity of this study.
Data collection on social performance was derived from the KLD database. This database rates US firms on several social aspects in their business practices and consists of firms from the S&P 500, Domini 400 Social SM Index, and from 2008 the 3,000 largest U.S. publicly traded companies by market capitalization. KLD evaluates firms according to thirteen categories of CSR strengths and weaknesses (Strike, Gao & Bansal, 2006) and is used in several studies like Griffin and Mahon (1997) and Hillman and Keim (2001). KLD assigns every company a ‘1’ or a ‘0’, which indicates whether a firm meets certain criteria or not. The dimensions that are included in the KLD database are community, corporate governance, employee-related, human rights, environmental, diversity and product aspects. Concerns like alcohol, nuclear weapons and others are left out of the data.
Data on total sales, foreign sales, profit and other financial data was collected from COMPUSTAT, both from the historical segments and annual data. The data of KLD and CompuStat
15 needed to be merged into one database, which resulted in deleting a substantial part of the data because it did not met all the criteria, namely: foreign sales, total sales, net profit, CSR rating, R&D expenses and advertising expenses, all between the years 2004 and 2013. 407 firms with a total of 4070 observations were found down the line, consisting of one observation per firm (407), per year (10).
Variables
There are nine variables included in this study, consisting of sales, financial data and CSR ratings. Two dependent, two independent and five control variables are used to test the hypotheses needed to answer the research question.
Dependent variables
The thesis has two dependent variables, namely CSR and CSiR. These variables are derived of the total CSR strengths and total concerns of the KLD data. These variables are recoded into dummy variables, resulting in two variables that report zeros for no CSR/CSiR and “1” for observations with CSR/CSiR.
Independent variables
The first independent variable is the degree of internationalization of a firm. These are not necessarily non-western markets, but also include western markets. The variable is measured as the ratio of the foreign sales divided by the total sales of a firm, following Eden et al (2002) and Strike et al (2006), who call this variable foreign market penetration. The higher the ratio (ranging from 0-1), the more internationalized a company is.
The second independent variable ‘CSR context’ is a dummy variable, where “1” reflect non-western sales (treated as low CSR contexts) and “0” all other, non-western, sales (i.e. high CSR contexts). CompuStat reports regions where companies report sales. These can either be domestic or foreign. Also the segment names are reported on which basis the distinction between western and non-western sales could be made. The dummy reports non-non-western sales as a “1” and all other (non-western) sales as a “0”, so Western sales is the reference category. So having non-western sales is treated like having operations in low CSR contexts (=1) and having only western sales is treated like having operations in high CSR contexts (=0)
Control variables
The study includes five control variables which previous studies have proven to take into account as controlling a relation: industry, firm size, firm performance, R&D intensity and advertising intensity.
Industry is included as a control variable because the effect of the IV on the DVs might be
16 reporting that specific industry or not. The industries are derived from the standard industry classification codes from CompuStat. Several researchers take this into account as a control variable (e.g. Strike et al, 2006; Bouquet & Deutsch, 2007;
According to previous studies firm size is strongly correlating with corporate social
performance measures. The variable is measured as the natural log of total assets, following Johnson & Greening (1999). This natural log also deals with the skewness (Skewness = ,619; SE = ,038), which is now found to be acceptable resulting in normality.
Financial performance is measured as the profits divided by the total assets, resulting in the
return on assets (ROA). The measure of financial performance differs a lot per study, some researchers take return on sales (Strike et al, 2006), others use return on equity and still others use the measure of ROA as in this study (e.g. Bouquet & Deutsch, 2007; Johnson & Greening, 1999). This measure is suffering from large skewness (Skewness = -,4610; SE = ,038) and kurtosis (Kurtosis = 53,52; SE = ,077). Therefore usually the variable needs to be log transformed, but as financial performance can take negative values this will not be possible.
R&D intensity and advertising intensity affect CSP according to McWilliams and Siegel (2000).
These are measured by dividing respectively R&D expenses and advertising expenses by total sales of the firm, which is an often used measure in previous studies. Both R&D and advertising intensity have a strong positive relation with corporate social performance, so will very likely be a predictor of CS(i)R. Because of a large number of missing values on especially advertising expenses and in a lesser way R&D expenses, dummies are created reporting whether firms have (=1) or do not have (=0) R&D and advertising expenses.
Methods
The hypotheses are tested via binary logistic regression in generalized estimating equations (GEE). This is the most appropriate method because of the nature of the variables in the study and the GEE’s possibility of analyzing panel data over multiple years. Three different models will be tested per dependent variable, adding variables per step, starting with the control variables. This results in two model with each three steps.
This method for regression is used to test the relationship between independent variables and individual dependent variables, at the same time including covariates that need to be controlled for. The regression coefficients are estimated through the process of maximum likelihood (MLE) and leaves me with log-odds ratios or incident rate ratios, the exponentiated Beta (Exp(B)). An Exp(B) value of 1,34 for example means than with a one unit increase in the dependent variable, the likelihood of being a “1” in the dependent dummy variable increase with 34 %, or is 1,34 times more likely, controlling for the other predictive variables in the model.
17
Results
The results contain descriptives of the included variables and the correlations among them. Further the relationship will be assessed through binary logistic regression, resulting in statistically testing the stated hypotheses.
Descriptives
Table 1 shows descriptive information of the variables in the total sample included in the study. There is a total of 4070 observations, divided over 407 firms in 10 years which both have KLD and CompuStat data. The CSR dimensions all have a minimum score of “0”, which indicate zero strengths of concerns on the specific dimension. The max scores are obviously “1” because they are dummies, where the score of 1 indicates that the firms has one or more strength(s) or concern(s). Employee-related concerns shows the highest mean (M = 0,386), so this firms show the most concerns on this dimensions relative to the other concerns.
R&D and advertising intensity are also dummies. The means of these variables show that the firms in the sample report R&D expenses more often than expenses on advertising.
The descriptives for financial performance is the only variable that also report negative minimal scores, which is caused by the way it is measured. A firm for example can have negative financial performance in a given year. The mean of 0.056 indicates that the firms in the total sample have a positive financial performance on average. This can differ per firm.
Variables N Min Max Mean SD
1 Corporate social responsibility 4070 0 1 0,087 0,282
2 Corporate social irresponsibility 4070 0 1 0,165 0,371
3 R&D Intensity 4070 0 1 0,687 0,464 4 Advertising Intensity 4070 0 1 0,326 0,469 5 Firm Size 4070 3,97 13,93 7,951 1,579 6 Financial Performance 4070 -1,54 0,5 0,056 0,097 7 Degree of internationalization 4070 0,004 1 0,468 0,405 8 CSR context (1=”Low”) 4070 0 1 0,713 0,453
Table 1 Descriptives of included variables in the overall sample
On the bottom of the table the observations for the independent variables are shown. The dummy CSR context obviously has a min and max of 0-1. The CSR context in which a firm operates has a mean of ,731. This implies that companies in the sample are operating more often in low CSR contexts than only in high CSR contexts. Note that operating in low CSR contexts does not mean that a firm does not operate in high CSR contexts. The degree of internationalization shows that firms range from 0,004 to 1 with a mean of 0.468, indicating that there are both firms with almost now foreign sales and firms with only foreign sales. The mean shows that on average, firms sell their products or services in foreign
18 countries. The industry dummies are not included in these descriptives because they are discussed further on.
The industry dummies are distributed per year as follows:
Industries before Number of firms Industries after Number of firms
Agriculture 2 Mining, Construction and Agriculture 25
Construction 2 Services, Finance and Public Adm. 72
Finance 14 Manufacturing 273
Manufacturing 273 Trade and Transportation 37
Mining 21 Public administration 2 Retail trade 10 Services 56 Transportation 13 Wholesale Trade 14 Total 407 Total 407
Table 2 Firms per industry before and after manipulation
As shown in table 2, Manufacturing is by far the best represented of all the industries and therefore will be used as the reference category. Agriculture, Construction and Public administration are the smallest with only two firms per year each. To get more representation of all industries the data will be somewhat manipulated. Agriculture and construction are combined with Mining (N=25), Finance, and Public administration will be combined with Services (N=72) and Wholesale Trade, Retail Trade and Transportation are combined into the industry Trade and Transportation (N=37). Manufacturing remains intact and is still the reference category as it is still the most represented industry.
Correlations
In table 3 the correlation matrix is given for the entire sample, so for all 407 firms in 10 years, resulting in 4070 observations.
The correlation matrix for the entire sample shows the correlations among the dependent, independent and control variables. The first coefficient that is striking is the not significant correlation between the two dependent variables CSR and CSiR (R = ,002; n.s.). The constructs are apparently not related in anyway or at least they are not in this sample. However, the dependent variables and degree of internationalization (DOI) seem to be related. DOI shows a positive correlation with both CSR (R = ,045; P<.001) and CSiR (R = ,024; n.s.) but is only significant for CSR. Both CSR and CSiR are significantly positively correlated to the dummy CSR context (respectively R = ,081; ,038; P<.001), which could indicate that both CSR and CSiR go up when the dummy of CSR context reports a “1”, which is the low CSR context. The two independent variables DOI and CSR context show a moderate positively correlation, which could imply multicollinearity, which will be addressed later on.
19 Some other correlations that are worth mentioning is the correlation between Firm size and both CSR (R = ,385; P<.001) and CSiR (R = ,114; P<.001). Thus when firm size goes up, so does the CSR and CSiR, although the correlation is stronger for CSR. Further, the dependent variables correlations with some of the industry variables, although they are not very strong. Also, some industries seem to correlate with each other, for example Trade and Manufacturing (R = ,451; P<.001) and Manufacturing and Services, Finance & Public Administration (R = ,662; P<.001).
Table 3 Correlation matrix overall sample
Regression
There are several models that will be tested to test the stated hypotheses. The two dependent variables (CSR and CSiR) will be tested in two separate models, with both the same predictors and control variables. Every model has several steps as shown in table 4.
Model 1: CSR Model 2: CSiR
Step 1: - Firm size
- Financial performance - R&D intensity - Advertising intensity - Industry dummies - Firm size - Financial performance - R&D intensity - Advertising intensity - Industry dummies
Step 2: - Degree of internationalization - Degree of internationalization
Step 3: Interaction term:
- DOI * CSR context
Interaction term:
- DOI * CSR context Table 4 Regression models
CSR
First, model 1, with CSR as the dependent variable, will be analyzed and some descriptives and the regression coefficients for binary logistic regression can be found in table 5. Longitudinal data can be
Variables 1 2 3 4 5 6 7 8 9 10 11 1 CSR Y/N 1 2 CSiR Y/N ,002 1 3 Degree of Internationalization ,045** ,024 1 4 CSR context ,081** ,038* ,335** 1 5 Advertising Intensity ,050** -,010 -,029 ,037* 1 6 R&D intensity ,004 ,023 ,310** ,218** ,126** 1 7 Firm Size ,385** ,114** -,040* ,026 -,077** -,104** 1 8 Financial Performance ,055** -,044** -,013 ,015 ,098** -,037* ,070** 1 9 Manufacturing -,017 ,033* ,270** ,115** ,008 ,655** -,105** -,032* 1 10 Trade -,005 -,007 -,217** -,122** ,001 -,446** ,064** ,025 -,451** 1
11 Mining, Construction, Agriculture ,042**
,085** ,020 ,081** -,110** -,271** ,134** ,034* -,365** -,081** 1 12 Services, Finance & Public Administration -,001 -,088**
-,182** -,100** ,058** -,299** -,003 -,001 -,662** -,147** -,119** **. Correl a tion i s s i gni fi ca nt a t the 0.01 l evel (2-tai l ed); *. Correl a tion i s s i gni fi ca nt a t the 0.05 l evel (2-tai l ed).
20 analyzed in SPSS through the so-called Generalized Estimating Equations. Subject variables therefore are the Tickers (identifiers for specific firms) with a total number of 407. The within-subject variable that is used is year of observation with a total number of 10 (2004-2013). As there are 10 industry dummies, one needs to serve as the reference category. Therefore the dummy for Manufacturing is chosen. This was represented the most of all industries.
Model 1a has only included the control variables. Advertising intensity (Exp(B) = 1,55; P<.001) and Firm size (Exp(B) = 1,90; P<.001) have a significant effect on the dependent variable which is the dummy variable for whether a firm has CSR strengths or not. For advertising intensity this means than when a firm reports advertising expenses, the likelihood of that same firm reporting CSR strengths is 1,55 times higher, controlling for the other variables in the model. The same goes for the Firm size, only the effect here is even stronger; with every one unit increase in firm size, the probability that the firm reports CSR strengths is 1,9 times higher, controlling for the other variables. The coefficients of the industry dummies show insignificant findings: Trade & Transportation (Exp(B) = ,93; P = ,82), Mining, Construction & Agriculture (Exp(B) = 1,15; P = ,60) and Services, Finance & Public Administration (Exp(B) = 1.18; P = ,41). Also R&D intensity and financial performance are not significant.
Model 1b is the model where the independent variable ‘degree of internationalization’ is added to the equation. In this model Advertising intensity is still significant and the likelihood even increases a little (Exp(B) = 1,57, P<.001). Firm size is also still significant and the likelihood stayed the same (Exp(B) = 1,9, P<.001), so with every one unit increase in firm size, the probability of firm having CSR strengths is 1,9 times higher. The industry dummies are still not significant, so nothing changed there. The coefficients for the degree of internationalization show that it is not a significant predictor for CSR (Exp(B) = 1,50; P = ,15). It does show a positive likelihood of 1,5 times higher with a one unit increase, but because it is not significant it cannot be interpreted as such. Therefore, hypothesis 1a stating that a higher degree of internationalization leads to less corporate social responsibility is not supported.
When adding the interaction-term of degree of internationalization and CSR context, not much is changing in comparison with model 1b. The same control variables remain significant and also the likelihood ratios do not differ much. Also the control variables that were not significant before are still not significant. The degree of internationalization shows another likelihood of CSR (Exp(B) = ,96; P = ,90) but this is still not significant so not interpretable. Adding the interaction made it a lot more insignificant than before. For the interaction-term between the two independent variables, degree of internationalization was mean-centered to avoid multicollinearity, but when this term was added, the model could not process which in turn indicated multicollinearity so the normal measure for DOI is used. The interaction is found to be marginally significant (Exp(B) = 1,59; P = ,05), resulting in the
21 conclusion that an interaction between the degree of internationalization and the CSR context seems to be 59% more likely when this is a low CSR context (i.e. a one unit change in CSR context means 1 = low CSR context). Therefore, marginal support is found for the hypothesis that the degree of internationalization is positively related to CSR as long as the CSR context is low.
Model 1: CSR Model 1a Model 1b Model 1c
B P Exp(B) B P Exp(B) B P Exp(B)
(Intercept) -4,58 ,00 ,01 -4,77 ,00 ,01 -4,73 ,00 ,01
Advertising intensity ,44 ,00 1,55 ,45 ,00 1,57 ,44 ,00 1,56 R&D intensity ,18 ,30 1,20 ,14 ,42 1,15 ,12 ,51 1,12
Firm size ,64 ,00 1,90 ,64 ,00 1,90 ,64 ,00 1,91
Financial performance -,41 ,19 ,66 -,41 ,20 ,66 -,41 ,19 ,66 Trade & Transportation -,07 ,82 ,93 -,02 ,94 ,98 -,04 ,88 ,96 Mining, Construction & Agriculture ,14 ,60 1,15 ,12 ,65 1,12 ,07 ,80 1,07 Services, Finance & Public Adm. ,17 ,41 1,18 ,20 ,32 1,22 ,19 ,35 1,21 Degree of internationalization ,40 ,15 1,50 -,04 ,90 ,96
DOI * CSR context (1=”Low”) ,47 ,05 1,59
QICC 4554,54 4544,74 4535,58
Table 5 Regression coefficients for CSR
At the bottom of the table the Corrected Quasi Likelihood under Independence Model Criterion (QICC) is found. This is a measure for goodness-of-fit of the models. The rule is that the model with the lowest value is the best fit for the data. In this case this is model 1c, the one with the interaction-term.
CSiR
Table 6 shows the second model, with corporate social irresponsibility as the dependent variable. The same variables are entered in three steps as with the model for CSR.
Model 2a shows that advertising (Exp(B) = 1,07; P = ,64) and R&D intensity (Exp(B) = 1,18; P = ,37) are no significant predictors of CSiR, under control of the other variables in the equation. They both show a positive likelihood, but these are not significant. Firm size and financial performance do seem to be predictors of CSiR, where firm size has a likelihood of 1,23 times of CSiR when it increases with one unit, under control of the other variables. Financial performance also shows a significant relationship (Exp(B) = ,32; P<.05), and so firms with a higher financial performance are expected to have a decreased likelihood of CSiR with 68% for every one unit increase.
Two of the three industry dummies show a significant coefficient. Companies in the Mining, Construction & Agriculture are expected to be 5,94 times more irresponsible. This might be expected because of the nature of the industry. For example, Mining and construction is often associated with pollution and is also often more associated with low CSR contexts because of the low labor costs than
22 the Services industry for example. Firms in the Services, Finance & Public Administration industry are less likely to be irresponsible (Exp(B) = ,62; P<,01) with a 38% decrease, under control of the other variables in the model.
Model 2: CSiR Model 2a Model 2b Model 2c
B P Exp(B) B P Exp(B) B P Exp(B)
(Intercept) ,12 ,76 1,13 ,13 ,75 1,14 ,16 ,70 1,18
Advertising intensity ,07 ,64 1,07 ,07 ,64 1,07 ,06 ,66 1,07 R&D intensity ,16 ,37 1,18 ,17 ,36 1,18 ,14 ,43 1,16
Firm size ,21 ,00 1,23 ,21 ,00 1,23 ,21 ,00 1,24
Financial performance -1,13 ,02 ,32 -1,13 ,02 ,32 -1,13 ,02 ,32 Trade & Transportation -,07 ,81 ,94 -,07 ,80 ,93 -,09 ,76 ,92 Mining, Construction & Agriculture 1,78 ,00 5,94 1,78 ,00 5,95 1,75 ,00 5,74 Services, Finance & Public Adm. -,47 ,01 ,62 -,48 ,01 ,62 -,49 ,01 ,62 Degree of internationalization -,03 ,92 ,97 -,34 ,46 ,71
DOI * CSR context ,33 ,32 1,39
QICC 3164,57 3166,62 3166,92
Table 6 Regression coefficients for CSiR
The model where the degree of internationalization is added, does not really affect the previous parameters of the control variables. The significance levels remain intact and also the likelihood ratios do not strongly change.
Degree of internationalization does not affect the likelihood of being socially irresponsible. The exponentiated B-value indicates that there is a slight decrease in being irresponsible (Exp(B) = ,97; P= ,92) but this is not significant and therefore cannot be interpreted as such. Therefore, the hypothesis stating that the higher the degree of internationalization, the higher the social irresponsibility is not supported.
Also adding the interaction-term of the degree of internationalization and CSR context did not change much in comparison to the previous models. Found relations remain intact under control of the other variables. The interaction-term itself is found to be not significant (Exp(B) = 1,39; P = ,32), which means that degree of internationalization in combination with presence in a low CSR context does not seem to influence whether firms are socially irresponsible or not.
Table 7 gives a quick overview of the hypotheses that have been tested and the corresponding results.
23
Hypothesis 1a Degree of
internationalization
(-) Corporate social responsibility 1,50 ,15 No
Hypothesis 1b Degree of
internationalization
(+) Corporate social irresponsibility ,97 ,92 No
Hypothesis 2a Degree of
internationalization * CSR context
(+) Corporate social responsibility 1,59 ,05 Yes
Hypothesis 2b Degree of
internationalization * CSR context
(+) Corporate social irresponsibility 1,39 ,32 No
Table 7 Overview of the hypotheses
In the following section a recap of the findings will be given and the discussion in which the methods, literature and findings are critically reviewed.
24
Discussion
This report has tried to shed light on the complex relationship between internationalization, corporate social responsibility and the CSR context in which multinationals operate. The latter is an under investigated topic of research in combination with the first two areas of research. There is too little empirical knowledge on what impact the nature of the CSR context has on the relationship between internationalization and corporate social (ir)responsibility. There is too little known about what influence low CSR contexts have on the social responsible and irresponsible behavior of firms. This study aimed to contribute in bridging this gap to a certain extent or at least tried to influence the discussion.
CSR in this study was divided into socially responsibility on the one side and irresponsibility on the other. Following findings of Strike et al (2006) CSR is divided because they found that the constructs are significantly different from each other (i.e. firms can be both responsible and irresponsible at the same time). When the construct of CSR has two opposing aspects, influencing factors might also have different implications across those two parts. They are not just two sides of the same coin.
An important finding of this thesis is that the degree of internationalization in combination with operations in low CSR contexts seems to be an indicator for social responsible behavior. That same evidence could not be found for social irresponsible behavior. Also, no evidence could be found that degree of internationalization is a predictor for lower CSR or higher CSiR, or reversed findings for both of them. The results turned out to be not significant. The relationship was hypothesized following the thought that firms might have a hard time managing their firm activities (including CSR) when they are dispersed across many countries. More countries means more competition and more (diverse) stakeholders, which in turn might result in being less able or willing to allocate resources for CSR. Landier et al (2007) found geographical dispersion and employee friendliness to be negatively related. Note that employee-related aspects is one of the seven dimensions of the KLD database in which both strengths and concerns are rated. Also Bartlett and Ghoshal (1989) state that the nature of social activities are influenced by how geographically dispersed a firm is. Further evidence that the hypothesis might be supported can be derived from the fact that multinationals are often accused of taking advantage of low social and environmental standards in developing countries, as stated by Low and Yeats (1992). Also the Pollution Haven Hypothesis is an often heard criticism in which multinationals are expected to relocate their polluting activities to countries with low environmental regulations.
The fact that this study found a positive relation between degree of internationalization in low CSR contexts and CSR follows previous research. Bansal and Roth (2000) for example stated that multinationals transfer their best practices to developing countries which in turn could improve the standards in those countries and therefore might have a positive effect on the social responsibility
25 ratings of that firm. Also Brammer et al (2006) found that an MNEs dispersion of activities is associated with improvements in social performance, especially for community and employee-related aspects, as long as the operations did not extend to eastern Europe. This thesis tried to contribute to this findings by investigating the relation not only in Europe but across the entire world and with a sample of mostly U.S. firms, as opposed to UK firms in the Brammer et al (2006) study.
Limitations and further research
The study as it is conducted has several limitations that influenced the findings. One of the limitations of this study is the measure of the CSR contexts. This was done based on the assumption that non-western countries have lower regulatory and environmental standards. Although this likely will be the case, it is not empirically researched and also existing literature is not clear about when a country or area should be classified as being a high or low context for CSR. In addition to that, this context might also vary not only between but also within countries.
Another limitation is the measurement of the dependent variables. These were dummy variables which reported ‘zero’ or ‘one or more’ strength/concern. Therefore this study could not say anything about the degree of corporate social (ir)responsibility and could only make implications about whether or not CS(i)R was predicted. Future research might focus on the extent of CS(i)R in relation to the degree of internationalization and CSR contexts.
Furthermore, the data that was used to make the distinction between CSR contexts is not categorized in a uniform way. Where one MNE reports every specific country in which it had sales, others clusters these sales in classifications like EMEA (Europe, Middle-East and Asia). These observations had to be left out of the study because they are too general for the distinction between western and non-western sales (Europe might be seen as Western, where the Middle-East and Asia are seen as non-western). This caused a lot of observations to be excluded from the study. Also, foreign sales might not be the best measure for geographical dispersion. For example, it does not account for the amount of effort a multinational has to put in to create those sales. A firm that only engages in exporting their products and therefore reports sales in foreign countries might not be affected by that sales in terms of CSR. Therefore an implication for future research is that it might be interesting to investigate the relation between the type of foreign direct investment (e.g. Greenfield, joint-ventures, acquisitions etc.) and CSR or to add this to the relationship between degree of internationalization and CS(i)R.
Future research might also focus on the relationships investigated in this thesis but take also the different dimensions of CSR into account. Previous research (e.g. Brammer et al, 2006; Landier et al, 2007) already showed that implications of internationalization for CSR varies across those dimensions (e.g. community, employee-related, environmental, etc.)
26
Conclusion
A lot has been written about the relationship between internationalization and corporate social responsibility. Both in earlier times and in more recent years the findings are still divided into a camp that states that multinationals should be associated with irresponsible behavior and a camp that states that multinational do good over foreign borders. This study tried to contribute to this never ending discussion by stating that a higher degree of internationalization does not necessarily imply a lower corporate social responsibility or a higher irresponsibility. Although it might be harder to manage different practices across countries, the data does not support that this comes at the cost of social irresponsibility. Also no support was found that firms in low CSR contexts are acting more irresponsible than firms that are not active in those contexts. Suggestions like the pollution haven hypothesis are therefore not supported by these findings. The distinction between CSR contexts might be a meaningful contingency to take into account when doing research in the field of multinationals and corporate social responsibility. The findings might actually suggest that multinational enterprises that are present in low CSR contexts behave responsible in a certain way.
27
References
Aguilera, R.V., Ganapathi, J., Rupp, D.E., and Williams, C. A. (2007). Putting the S back in corporate social responsibility: A multilevel theory of social change in organizations. Academy of Management Review, 32(3): 836–863.
Aguinis, H. (2011). Organizational responsibility: Doing good and doing well. In S. Zedeck (Ed.), APA handbook of industrial and organizational psychology, 3: 855-879. Washington, DC: American Psychological Association.
Attig, N., Boubakri, N., El Ghoul, S. and Guedham, O. (2013). International Diversification and Corporate Social Responsibility. Working Paper Series, School of Business Administration, American University of Sharjah
Bansal, P. and Roth, K. (2000). Why Companies Go Green: A Model of Ecological Responsiveness. The Academy of Management Journal 43(4): 717-736
Barcos, L., Barroso, A., Surroca, J. and Tribo,J. (2013). Corporate social responsibility and Inventory policy. International Journal of Product Economics 143:580–588.
Barnett, M.L. (2007). Stakeholder influence capacity and the variability of financial returns to corporate social responsibility. Academy of Management Review 32(3): 794-816
Bartlett, C.A. and Ghoshal, S. (1989). Managing across borders: the transnational solution. Boston, MA: Harvard Business School Press.
Bouquet, C. & Deutsch, Y. (2008). The Impact of Corporate Social Performance on a Firm’s Multinationality. Journal of Business Ethics 80: 755-769
Brammer, S., Pavelin, J. and Porter, L.A. (2006). Corporate social performance and geographical diversification. Journal of Business Research 59: 1025-1034
Bromley, D.B. (1993). Reputation, Image and Impression Management. John Wiley & Sons, Chichester.
Carroll, A. (2000). Ethical challenges for business in the new millennium, Corporate Social Responsibility and models of management morality. Business Ethics Quarterly 10(1): 33-42.
Cavaco, S. and Crifo, P. (2014). CSR and financial performance: complementarity between environmental, social and business behaviours. Applied Economics 46 (27)
Chapple, W. and Moon, J. (2005). Corporate social responsibility (CSR) in Asia: A seven-country study of CSR web site reporting. Business and Society 44(4): 415–441.
Christmann, P. (2004). Multinational Companies and the Natural Environment: Determinants of Global Environmental Policy Standardization. The Academy of Management Journal 47(5): 747-760 Coval, J. and Moskoqitz, T. (1999). Home bias at home: Local equity preference in domestic portfolios. Journal of Finance 20(6): 45-74
Crifo, P., Diaye, M.A. and Pekovic, S. (2014). CSR related management practices and firm performance: An empirical analysis of the quantity–quality trade-off on French data. International Journal of Production Economics. Forthcoming
Donaldson, T. and Dunfee, T. (1999). Ties that bind: a social contracts approach to business ethics. Cambridge, MA: Harvard Business School Press.
Dunning, J.H. (1981). International Production and the Multinational Enterprise. Allen & Unwin, London.
Dunning, J.H. (2003). Some Antecedents of internalization theory. Journal of International Business Studies 34(2): 108-115
28 Eden, L., Thomas, D. and Olibe, K. (2002). Why multinationality matters. In P. Gray ‘Extending the Eclectic Paradigm in International Business: Essays in Honor of John Dunning. Edward Edgar Publishing: Cheltenham, UK: 31-54
Frooman, J. (1997). Socially irresponsible and illegal behavior and shareholder wealth: A meta-analysis of event studies. Business and Society 36: 221-249.
Geringer, J.M., Beamish, P.W. and DaCosta, R.C. (1989). Diversification strategy and internationalization: implications for MNE performance. Strategic Management Journal, 10(2): 109-119.
Heugens, P.M.A.R., Kaptein, M. and Van Oosterhout, J. (2008). Contracts to communities: A processual model of organizational virtue. Journal of Management Studies, 45(1): 100-121.
Hillman, A.J. and Keim, G.D. (2001). Shareholder value, stakeholder management and social issues: what's the bottom line? Strategic Management Journal 22(2): 125-139
Hitt, M.A., Hoskisson R.E. and Kim, H. (1997). International Diversification: Effects on Innovation and Firm Performance in Product-Diversified Firms. Academy of Management Journal 40(4): 767–798.
Ioannou and Serafeim. (2012). What drives corporate social performance? The role of nation-level institutions. Journal of International Business Studies 43: 834-864.
Jackson G. and Apostolakou, A. (2010). Corporate social responsibility in Western Europe: An institutional mirror or substitute? Journal of Business Ethics 94(3): 371-394
Johnson, R.D. and Greening, D.W. (1999). The effect of corporate governance and institutional ownership types on corporate social performance. Academic Management Journal 42: 564-576
Kang, J. (2013). The relationship between corporate diversification and corporate social performance. Strategic Management Journal 34: 94-109
Kaptein, M. (2004). Business Codes of Multinational Firms: What do they Say? Journal of Business Ethics 50(1): 13–31.
Kogut, B. (1985). Designing global strategies: profiting from operational flexibility. Sloan Management Review 21: 27-38
Kolk, A. and Tulder, R. van. (2002). Child Labor and Multinational Conduct: A Comparison of International Business and Stakeholder Codes. Journal of Business Ethics 36(3): 291–301.
Kostova, T. and Zaheer, S. (1999). Organizational legitimacy under conditions of complexity: the case of the multinational enterprise. Academic Management Review 24(1): 64-81
Kotabe, M., Srinivasan, S.S. and P. S. Aulakh. (2002). Multinationality and Firm Performance: The Moderating Role of R&D and Marketing Capabilities. Journal of International Business Studies 33(1): 79–97.
Landier, A., Nair, V.B. and Wulf, J. (2007). Trade-offs in staying close: corporate decision making and geographic dispersion. The Review of Financial Studies 22: 1119-1148
Lewin, A.Y., Sakano, T., Stephens C.U. and Victor, B. (1995). Corporate Citizenship in Japan: Survey Results from Japanese Firms. Journal of Business Ethics 14(2): 83–101.
Letchumanan, R. and Kodama, F. (2000). Reconciling the conflict between the Pollution-Haven hypothesis and an emerging trajectory of international technology transfer. Research Policy 29: 59-79.
Lindenberg, S. (2001). Intrinsic motivation in a new light. Kyklos 54(2-3): 317-342.
Low, P. and Yeats, A. (1992). Do dirty industries migrate. IN: Low, P. International Trade and the Environment, World Bank Discussion Paper 159. World Bank, Washington D.C.
Maignan, I., and Ralston, D.A. (2002). Corporate social responsibility in Europe and the US: Insights from businesses’ self-presentations. Journal of International Business Studies 33(3): 497–515.
