Understanding stakeholder activism:
What is the role of CSR,
firm size, and time horizon?
Master’s thesis
Casper van Leeningen10184147
Date: 26 June 2015 Final draft
MSc. in Business Administration – Strategy Track Amsterdam Business School (University of Amsterdam) Supervisor: Dr. K.J.P. Quintelier
2
Statement of originality
This document is written by Student Casper van Leeningen who declares to take full responsibility for the 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 in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
3
Preface
This master’s thesis is the end product of four years of studying at the Amsterdam Business School. After successfully graduating the bachelor Economics and Business, I started with the master Business studies in 2014. I am very proud that I can share my knowledge through means of this thesis, and contribute to the available academic literature. Writing the thesis has been an interesting and instructive process.
I would like to thank three people who helped me during the process. First and foremost, dr. Katinka Quintelier, my supervisor. From day one of the process, she guided me with writing the thesis. She provided useful and interesting ideas about possible topics. During the process of writing, she always gave quick and extensive feedback. Furthermore, if we ran into trouble, she consulted other teachers to find a solution. Without Katinka, I would not have been able to deliver this thesis. Another person I would like to thank is dr. Michelle Westermann-Behaylo. She made me streetwise in the CHRD database. She too was quick with providing answers to my questions. Thanks to her training session to use the CHRD database, I had not much problems coding for and use of the database. Finally, I would like to thank Robert Kleinknecht. During the writing process, I faced the problem that I did not know how to measure the time horizon of a firm. Robert made time available to help me out, and suggested several methods to determine the time horizon.
I hope you will enjoy this thesis, and that you will gain some new knowledge.
4 Table of contents Abstract 5 1. Introduction 6 2. Literature review 8 2.1. CSR and CSP 8 2.2. Stakeholder activism 10
2.3. Relation between CSR and stakeholder activism 13
2.4. Size of firm 14
2.5. Time horizon of firm 16
2.6. Moderating effects of firm size and time horizon 18
3. Methodology 21
3.1. Design and sample 21
3.2. Measurements 24
3.3. Analyses and predictions 29
4. Results 31
4.1. Assumptions and transformations 31
4.2. Correlation analysis 34
4.3. Testing the hypotheses 37
5. Discussion 45
5.1. Summary, hypotheses, and unpredicted results 45
5.2. Alternative explanations 46 5.3. Theoretical contributions 48 5.4. Practical implications 50 5.5. Limitations 51 5.6. Positive points 53 5.7. Future studies 54 6. Conclusion 56 Bibliography 57
5
Abstract
Stakeholders have become more active in punishing and influencing firms that last decades. This stakeholder activism can be harmful to companies, in terms of economic cost and reputation damage. Therefore, firms need to know which firms are the main targets of stakeholder activists. This thesis researches what kind of firms are the targets of activists, by empirically testing quantitative data from three different databases. The paper tests the influence of corporate social performance, firm size, and firm time horizon on the number of lawsuits filed by stakeholders in case of a human rights violation committed by a firm. The results show that large firms are the primary targets of stakeholder activists, but that corporate social performance and time horizon do not have an influence on the number of lawsuits filed against these firms.
6
1. Introduction
Companies engaged in unethical business have faced more and more activism from their stakeholders in recent years. Famous are the cases of Unocal and Royal Dutch Shell. These major oil companies were accused of violating human rights in Myanmar and Nigeria, respectively. Activist organizations like EarthRights International sued Unocal and Shell on behalf of the victims. More recently, the online activist group Anonymous published a list of companies supporting the Islamic State. The negative attention created by these actions from activists obviously harms the companies. Therefore, it is important for business to know more about the relation between social performance and activism. Furthermore, it is helpful to know what kinds of companies are targeted most. Does it make a difference whether a company is large, small, short-term oriented, or long-term oriented?
In this master thesis, the link between corporate social responsibility (CSR) and stakeholder activism will be studied. In addition, this paper will examine the differences between large and small firms, as well as the differences between short-term focused and long-term focused firms, and how these factors make a difference in the relationship between CSR and activism. The goal of the thesis is to answer the following research question: What is the relation between corporate social performance and stakeholder activism, and does firm size and time horizon moderate this relation?
This thesis will contribute to academic theory in three ways. First, Rehbein, Waddock and Graves (2004) demonstrated that shareholder activism mainly focuses on socially irresponsible firms. This relation has only been tested for shareholder activism though. This study will complement the study of Rehbein et al (2004) by including other stakeholders than shareholders. Second, another important finding of Rehbein et al (2004) is that activists target large firms more often than small firms. This study tests the generalizability of this theory in the light of other stakeholders than shareholders. Third, stakeholders are likely to be sceptical
7
about short-term focused firms engaging in CSR activities. Therefore, the moderating effect of short- and long-term focused firms on the relationship between corporate social performance and stakeholder activism will be tested empirically.
The distinctions between different kinds of firms made in this study are interesting to managerial practice. If the hypotheses are confirmed, this will have considerable consequences for managers of professional companies. Managers of large firms, as well as managers of firms with a short-term focus should be aware of the danger of sceptical reactions from stakeholders to their corporate social performance.
In the next section, four hypotheses are developed, based on the academic literature. These hypotheses will be tested using quantitative data from three different databases: KLD Social Ratings Database, Compustat, and the Corporations and Human Rights Database (CHRD).
In the next section, the academic literature on CSR, corporate social performance, and stakeholder activism is reviewed. Based on theoretical arguments, we develop four testable hypotheses. Section 3 describes the methodology. Here, a precise description of the research method, databases, variables, and statistical analyses used is given. Section 4 contains the analysis of the data, including results. In this chapter, the four hypotheses are confirmed or rejected. Subsequently, the results are discussed in section 5. Here, the relevance and added value of this study for academia and managerial practice are presented, as well as limitations and suggestions for future research. Finally, the paper ends with a conclusion in section 6.
8
2. Literature review
This section gives an overview of the progress scholars have made over the past decades in the field of corporate social responsibility and stakeholder activism. During the literature review, several hypotheses are formulated that will be tested in this study.
2.1. CSR and CSP
Scholars have studied corporate social responsibility (CSR) for decades. Interest in the field of CSR has become more widespread over the last decades (Lee, 2008; Aguinis & Glavas, 2012). The theory of Milton Friedman (1970) that the only social responsibility of business is to increase its profits is outdated. Corporations understand nowadays that they have a responsibility to other parties than only their shareholders. In 1977, less than half of the Fortune 500 companies acknowledged CSR in their annual report. Two decades later, 90% of these firms regarded CSR as a core element of business (Boli & Hartsuiker, 2001, as cited in Lee, 2008). Many stakeholder groups have an interest in business, and they will demand the business to take care of their interests (McWilliams & Siegel, 2001). Taking care of these stakeholders is one of the main purposes of CSR, as will be explained in the next paragraph.
Despite the widespread interest in CSR and the large number of studies, there remains disagreement about a universal accepted definition (Carroll, 1999). Mackey, Mackey and Barney (2007, p. 818) define CSR as “voluntary corporate actions designed to improve social conditions”. McWilliams and Siegel (2001, p. 117) confirm the voluntary nature: “CSR are actions that appear to further some social good, beyond the interest of the firm and that which is required by law”. However, CSR is not always beyond the interest of the firm. Many studies point at the fact that engagement in social activities can enhance financial performance (Porter & Van der Linde, 1995; Turban & Greening, 1997; Barnett, 2007; Ambec & Lanoie, 2008; Godfrey, Merrill & Hansen, 2009; Barnett & Salomon, 2012; Margolis, Elfenbein &
9
Walsh, 2015). This study focuses specifically on the relation between CSR and stakeholders. Therefore, the definition of CSR as offered by Aguinis (2011, p. 855) is used in this paper: “context-specific organizational actions and policies that take into account stakeholders’ expectations and the triple bottom line of economic, social, and environmental performance”.
CSR can take many forms. There are two characteristics that distinguish CSR activities from other corporate activities: stakeholder relationship orientation and social welfare orientation, including environmental concern (Barnett, 2007). CSR is a means to maximize the positive corporate impact on stakeholders, while minimizing its negative impact (Maignan & Ralston, 2002). Maignan and Ralston (2002) identified seven CSR processes: philanthropic programs, sponsorships, volunteerism, codes of ethics, quality programs, health and safety programs, and management of environmental impacts. Almost all of the CSR activities can be placed under one of these seven processes.
This study focuses on the relation between corporate social performance (CSP) and stakeholder activism. CSP is related to CSR, but these terms are not synonyms. CSR activities are the investments which result into a certain CSP posture (Barnett, 2007). Again, there is disagreement about a precise definition of CSP (Zhang, Di Fan & Zhu, 2014), and about how to measure it (Wood & Jones, 1995). Here, the definition of Maignan and Ferrell (2000, p. 284) is used: “[CSP is] the extent to which businesses meet the legal, ethical, and discretionary responsibilities imposed on them by their stakeholders”. This definition is suitable because it involves the stakeholders. It is important to notice that this definition lacks any economical factors. Financial gain is an important driver for businesses to engage in social behaviour. But the financial gain associated with CSP is a result of it, not a determinant. Some of the factors that determine CSP are product quality, treatment of women and minorities, good relations with unions, safe and healthy working conditions, emphasis to environmental protection, and donation to charities (Turban & Greening, 1997; Zhang et al,
10
2014). Thus, CSP is the overall social performance of a firm at a particular point in time (Barnett, 2007).
2.2. Stakeholder activism
As mentioned before, every company has a variety of different stakeholders. Managing relations with these stakeholders is one of the main functions of CSR, according to the previously mentioned definition of Aguinis (2011). Stakeholders are “any group or individual that can affect or is affected by the achievement of an organization’s purpose” (Freeman, 2010: p. 53). Shareholders, employees, customers, suppliers, non-governmental organizations (NGOs) and the government are typical stakeholders of businesses. They have an interest in the business. The problem is that businesses and stakeholders often have different interests. Likewise, the interests of different stakeholders also typically diverge from each other (Jones, Felps & Bigley, 2007). In the case that the behaviour and actions of the business are harmful for the stakeholders, they have an incentive to influence and change the focal firm. The way of doing this is often punishing (Boyd, Gintis, Bowles & Richerson, 2003; Marlowe, Berbesque, Barr, Barrett, Bolyanatz, Cardenas & Tracer, 2008; Boyd, Gintis & Bowles, 2010). This punishing of firms which harm stakeholders can be labelled stakeholder activism. Stakeholders often unite in order to have a greater impact on the focal firm (Boyd et al, 2010). There are two reasons for this. First, groups of stakeholders have more power than individuals, and therefore it is more likely that they are able to have an impact on the focal firm (Boyd et al, 2010). Second, it is difficult for individuals to monitor firms. Therefore, there is a need for specialization of monitoring (Marlowe et al, 2008). Specialized third parties like NGOs serve as the punishers of businesses. This is known as third-party punishment (Marlowe et al, 2008). An example of specialized organizations that represent individual stakeholders are unions. These unions defend the labour rights of the employees. It
11
is also possible that a third party stakeholder acts on the behalf of non-human stakeholders. An example of this is an NGO that protects the environment. Because the environment can not influence businesses by itself, other parties take care of the interest of the environment. The same holds for children, poor people and animals.
It is useful to illustrate the concept of stakeholder activism with some examples. This thesis focuses on stakeholder activism as a reaction on human rights violations. All the data for this study is based on violating firms. The following three examples are all cases of firms violating human rights. Although they are violating human rights, there are some differences in the degree of involvedness and in the reaction of stakeholders.
The first example of stakeholder activism is the lawsuit that three Eritrean refugees filed against the Canadian mining firm Nevsun Resources (Anderson, 2014). Nevsun formed a joint venture with Segen Construction, an Eritrean state-owned contractor. Together, Nevsun and Segen Construction operate the Bisha Mine. The three refugees, who are former workers of the mine, claim that several human rights were violated. They had to work many hours per day under very harsh conditions for only 36 euro per month. Their work involved exposure to harmful gases and extreme high temperatures. With the lawsuit, the three workers, who currently live in a refugee camp in Ethiopia, try to change the policy of the Canadian mining firm. Human Rights Watch interfered in the process, by publishing a report in which it states that Nevsun Resources is indeed responsible for overseas violation of human rights.
The second example is the case of the British food firm Cadbury Schweppes (Friends of the Earth, 2004). This firm uses palm oil as an ingredient for many of its products. Most of this palm oil is coming from palm oil plantations in Malaysia and Indonesia. The growing demand for palm oil is one of the main reasons for deforestation of tropical rainforests. But deforestation is only one negative result. Other negative externalities of palm oil plantations
12
and deforestation is the encroachment of local communities, air contamination, and destruction of wildlife. Workers on palm oil plantations are poorly paid and they are often forced to enlist the help of their entire family to meet production targets. The international NGO Friends of the Earth called Cadbury Schweppes to broaden its policy to address forest and community issues. Moreover, they called on the British Government to introduce corporate accountability legislation in the UK to make British companies abroad accountable for their environmental and social impacts. But they did not file lawsuits against Cadbury Schweppes.
The third example is that of a bunch more than 50 western companies that supported the apartheid regime in South Africa (Raphael, 2002). These companies include, amongst others, IBM, Citigroup, Exxon Mobil, and General Motors. These multinationals sold products to the apartheid regime, despite an international embargo. In the apartheid era, lasting from 1960 until 1994, the South African government killed, tortured, and discriminated the black citizens. The government was violating human rights on a large scale. Without the support of the multinational corporations, the apartheid regime could not have been maintained in the same manner for such a long time. The Khulumani Support Group filed a lawsuit against all the supporting western multinationals. The Khulumani Support Group is a South African NGO that represents 32,000 victims of the apartheid regime. This group seeks for monetary reparations from multinationals for the victims. The process started in November 2002. Finally, all the accusations were dismissed. This example illustrates that a company can face stakeholder activism while it does not violate human rights by itself. Supporting a violating party can also lead to activism.
Although all these three examples are essentially about firms violating human rights, there are some important differences and similarities. Cadbury Schweppes and Nevsun are quite similar in their human rights abuses. Both firms are directly involved in violating human
13
rights. However, only Nevsun faced stakeholder activism in the form of a lawsuit. There were no lawsuits filed against Cadbury Schweppes. This means that not all the violations of human rights result in the same sort stakeholder activism. The apartheid case differs from the other two examples in the sense that the involvement of the companies in the human rights violations is not directly. IBM did not kill and torture opponents of the apartheid regime. The South African government did this by themselves. The western companies are only indirectly involved, by supporting the violating party. But as the case shows, this can lead to lawsuits as well. In short, these examples illustrate that firms can be involved in human rights violation in different manners. Firms can directly commit the violations, but they can also support another party that abuses human rights. Furthermore, stakeholders do not always react in the same way. In some cases, they will file lawsuits, and in other cases, they will do nothing against the violating firm.
2.3. Relation between CSR and stakeholder activism
Businesses use stakeholder management to take care of the interests of stakeholder groups, and to decrease stakeholder activism (Jones, 1995). CSR can be seen as a tool of stakeholder management. CSR aims to improve the firm’s relations with stakeholders, and thus to decrease stakeholder activism (Barnett, 2007). Firms with extensive CSR activities are likely to have high CSP at a particular moment (Barnett, 2007). Therefore it seems plausible to argue that high CSP at any point in time is negatively related to stakeholder activism. But it is not that simple. Stakeholders draw from prior knowledge of a focal firm when they evaluate new knowledge that results from CSR activities (Barnett, 2007). In other words, the evaluation of CSR activities by stakeholders is history-dependent. The implication of this is that businesses should have high CSP over a longer time period in order to decrease
14
stakeholder activism. A time period of five years should be long enough to have a significant effect on stakeholder activism (Barnett, 2007).
Rehbein et al (2004) confirm the relation between low CSP and shareholder activism. Their study focuses exclusively on shareholders, which are only one party of all the stakeholders. It remains unclear whether low CSP of firms results in increased activism of other stakeholders. CSP includes many different elements, like corporate governance, environmental performance, social performance, and engagement in controversial businesses like tobacco and fire arms. This means that CSP does not only have an impact on shareholders, but also on other stakeholders. In order to measure enduring CSP, we can take the average of a firm’s five year CSP. Lawsuits filed by stakeholders can be used as a measure for stakeholder activism. The Cadbury Schweppes, apartheid, and Nevsun cases are typical examples of activism on which this study relies. In some sense, this study can be seen as an extension of the Rehbein et al (2004) study. The claims above will be tested by the first hypothesis.
Hypothesis 1: Enduring high corporate social performance is negatively related to stakeholder lawsuits.
2.4. Size of firm
Hypothesis 1 is applicable to all kind of firms. It does not make a distinction between different kind of firms. However, it is likely that not every kind of firm has to deal with stakeholder activism in the same way. Based on the literature, large firms have a different relation with stakeholder activism than small firms. It seems that large firms are more frequently confronted with stakeholder activism than small firms. There are several reasons
15
for this. This section will outline the reasons why large firms face more stakeholder activism than small firms.
First, large firms are more visible to society (Bushee & Miller, 2012). Large firms are the firms that are on the front pages of the news papers. These large visible firms are put under the microscope. This means that every action of the large firm catches the attention of many eyes, including stakeholder activists (Eesley & Lenox, 2006). For example, if Royal Dutch Shell is involved in human rights violation, this is more likely to catch the attention of stakeholders than when a small and unknown firm does the same thing.
Related to the visibility of large firms is the attention stakeholder activists can get when they target irresponsible large firms (Rehbein et al, 2004). Again, if stakeholder activists such as Greenpeace targets Walmart, this is likely to attract more media attention than the case where Greenpeace targets a small local supermarket in Belgium. This is also attractive for activists, since extensive media coverage is likely to attract more brand awareness for the NGO, volunteers, and donors.
Third, confidence in big companies is low. Only 24 percent of respondents said they had at least quite a lot of confidence in these big companies (Ellen, Webb & Mohr, 2006). This means that people are often sceptical about large firms. Scepticism refers to a person’s tendency to doubt, disbelieve, and question (Forehand & Grier, 2003). The reason for this scepticism is the perceived motive of companies for their behaviour. Bigger companies are perceived as mainly egoistic- or stakeholder-driven (Skarmeas & Leonidou, 2013). In other words, they are using CSR as a tool to keep the large numbers of powerful shareholders and other stakeholders satisfied, instead of promoting the real cause of the CSR activity. For example, a large public corporation investing in rainforest conservation does this to keep environmental NGOs satisfied. The decreased activism that should result from these actions should ultimately increase profits. This is in the interest of the shareholders. Stakeholders will
16
be sceptical about these motives, since these motives are not related to the core business and the mission of the firm (Ellen et al, 2006). On the other hand, motives for smaller companies to engage in CSR are perceived as values-driven (Skarmeas & Leonidou, 2013). This implies that there is a fit between the firm’s core business and the cause’s mission (Ellen et al, 2006).
Fourth, the shareholder value-loss in the context of a negative event is greater for larger firms than for smaller firms (Godfrey et al, 2009). This implies that large corporations have a greater incentive to engage in CSR activities than smaller companies. This incentive for CSR may be a cause for the scepticism of consumers towards large corporations, because it is driven by self-interest instead of the interest of society at large (Godfrey et al, 2009).
Finally, Activists can have more impact when they target large firms. Large firms are those that can have a significant negative or positive effect on society and on the environment. Stakeholder activists that target these large firms can have a much bigger impact on society and the environment, than when they target small firms with a small impact.
These five arguments clearly indicate that large firms are more likely to face stakeholder activism than small firms. This reasoning leads to the second hypothesis.
Hypothesis 2: Large firms will attract more lawsuits than small firms.
2.5. Time horizon of firm
Not only CSP and firm size are likely to have an effect on stakeholder lawsuits. Another factor to hold into account is the time horizon of the firm. A firm with a short time focus is a firm that mainly cares about its current shareholder value. Its main goal is to maximize shareholder value. In this sense, Friedman (1970) can be seen as a scholar championing short time focus. He argues that total welfare to society will be maximized if firms only care about their value to shareholders. Freeman (2010) is a business scholar at the other side of the
17
spectrum. He states that when firms take care of all the different stakeholder interests, long term value to the firm and to society will be maximized. This study hypothesizes that firms with a short time horizon will be confronted with more stakeholder lawsuits than firms with a long time horizon. There are several arguments in favour of this proposition.
First, firms can be divided in firms with a short-term focus and firms with a long-term focus. CSR is believed to be an investment in the future performance of a firm (Bhattacharyya, 2010; Quaak, Aalbers & Goedee, 2007; Smith & Sharicz, 2011). For example, a firm that invests in clean technology first makes a large investment. Only later, the firm can profit from a better reputation and lower taxes. Because of this long-term nature of CSR, it makes no sense for short-term focused firms to engage in CSR activities. Following this logic, only long-term focused firms should invest in CSR.
Second, if firms breach this logic, stakeholder scepticism arises. There are four main reasons for stakeholder scepticism towards CSR. The first reason is based on Ellen et al’s (2006) motives for CSR. These motives can be egoistic, strategic, values, or stakeholder-driven. Egoistic-driven motives are personal motives from the top managers. An important reason for managers to engage in CSR is an improved personal social status (Hemingway & Maclagan, 2004). Friedman (1962) has critique on this motive, since it is not in the interest of the shareholders, while they are the owners of the firm. If CSR serves to create a competitive advantage, the motive is strategic (Ellen et al, 2006). Porter and Kramer (2002, 2011) show how firms can improve their competitive position in the market, while at the same time improving the welfare of local communities and the environment. Values-driven motives for CSR are a result from the corporate values of the firm (Ellen et al, 2006). For example, one of Ben & Jerry’s core values is fair treatment of suppliers. Therefore, they will only buy products with a fair-trade label, in order to pay a fair price to their suppliers. If the only purpose of CSR practices is to keep stakeholders satisfied, the motive for CSR is
stakeholder-18
driven (Ellen et al, 2006). All the motives for CSR, except the values-based motive, can be seen as motivated by for-profit reasons. Indeed, different studies confirm these for-profit reasons (Fry, Keim & Meiners, 1982; Weber, 2008; Carroll & Shabana, 2010). Fry et al (1982) proved that corporate giving correlates with advertising costs. This means that companies try to make consumers aware of their social behaviour, in order to attract these consumers.
A second cause of stakeholder scepticism is the legitimizing function of CSR (O’Dwyer, 2002). Corporations use disclosure of information regarding CSR to control the public perception of the organization. This is a response to legitimize the company while facing social pressure (O’Donovan, 1999; Bansal & Roth, 2000). In other words, CSR is regarded as a tool to promote the corporate (private) interest instead of the stakeholder (public) interest. This is especially true for environmental disclosure (O’Dwyer, 2002).
The third reason why stakeholders are often sceptical about CSR is caused by conflicting information they receive about CSR. Therefore, stakeholders have trouble to distinguish socially responsible from socially irresponsible companies (Skarmeas & Leonidou, 2013). This inability to evaluate CSR well leads to scepticism (Vanhamme & Grobben, 2009).
Overall, if stakeholders are sceptical about CSR activities, it is more likely that they will file lawsuits against these firms. This means that short term focused firms should face more stakeholder activism in the form of lawsuits than long term focused firms.
Together, the arguments in this paragraph lead to the third hypothesis.
Hypothesis 3: Short-term focused firms will attract more lawsuits than long-term focused firms.
19
2.6. Moderating effects of firm size and time horizon
Hypothesis 1-3 are all about main effects. Beside these main effects, there are also possible moderating effects of firm size and firm time horizon on the hypothesized negative relation between CSP and stakeholder lawsuits (hypothesis 1).
Scepticism is the central theme in the explanation of this moderating effect. As explained above, stakeholders are more sceptical about large firms and about firms with a long time horizon. The intentions for large firms to engage in CSR are perceived as mainly economic and not as sincere (Yoon, Gürhan-Canli & Schwarz, 2006). The interest of the shareholders is more important than the interest of other stakeholders. CSR activities are viewed as publicity stunts for these large corporations (Kim, 2014). In contrast, small firms generally are less dependent on shareholders. Their motives to engage in CSR are closely related to the personal motives of the owner or manager(s) of the firm. These motives for CSR are perceived as sincere (Yoon et al, 2006). The same reasoning holds for short-term focused firms. Stakeholders are likely to be suspicious about motives for CSR of these firms, since CSR can be regarded as an investment in the long-term. Together, these theoretical arguments underpin that the negative relation between CSP and stakeholder lawsuits should be weaker for large firms and for short-term focused firms and vice versa. However, due to a lack of data, it is not possible to test the moderating effect of the time horizon. Therefore, this study will only test the following hypothesis.
Hypothesis 4: The negative relation between CSP and stakeholder lawsuits will be weaker for large firms and stronger for small firms.
20 Figure 1. Conceptual model
21
3. Methodology
This section gives a detailed description of the methods used to answer the research question and to test the hypotheses. First, the design and sample are discussed. Second, the variables are introduced, including the measure, number of items, database, scale, and other characteristics. This section ends with a description of the statistical analyses that will be used to analyze the data.
3.1. Design and Sample
Data is collected through means of database research. Data from a total of three databases has been collected. The dependent variable, stakeholder lawsuits, is based on data from the Corporations & Human Rights Database (CHRD). This database contains information about firms which were accused from abusing human rights. For every case, the database contains data about judicial action and non-judicial action taken against the accused firm. In May 2015, a total of 975 unique allegations formed the database. The database has not been completed yet; at the end, there will be 6,000 allegations in it. For this study, only the data from firms that were also in the Compustat and/or KLD database was used. After screening the CHRD database on this criterion, 131 firms were left. These 131 firms were in total 499 times accused. In other words, 499 of the 975 allegations in CHRD were used for this thesis.
The data to measure CSP has been found in the Kinder, Lydenberg & Domini (KLD) database. The Wharton Research Data Services (WRDS) provided the data from KLD. This database contains data about firm’s strengths and weaknesses on aspects related to CSR. Together, these strengths and weaknesses give a social rating to the firm. KLD contains data from approximately 8,700 companies listed at a North American stock exchange. Only 53 companies in KLD matched the companies in CHRD. The CSP scores for all the other firms are useless for this study, since this study examines the relation between CSP and stakeholder
22
lawsuits. The small number of KLD firms useful for this study can be explained by the fact that KLD only contains relatively large North American listed companies. In contrast, CHRD contains data about companies worldwide, including small companies that are not listed on a stock exchange. Furthermore, the CHRD database is far from complete. Therefore it seems plausible to expect that the number of firms that are in KLD as well as CHRD will increase to a large extent in the coming months and years.
The data to determine the industry, country, firm size, and time horizon has been collected using the online database Compustat. Information from companies listed on U.S. or Canadian stock exchanges has been collected. Compustat provides a wide range of financial and descriptive information from thousands of firms and tens of different industries. Again, only data from firms that were in CHRD as well as Compustat is useful. This resulted in a number of 131 different firms. For all of these firms, the industry could be found. The number of employees was available for 125 of the 131 firms. The revenues were found for 118 different firms. For all 131 firms was the SIC code of the industry given in Compustat. This resulted in a list of 37 different industries. However, a lot of different industries were closely related. For example, the SIC industries Services-Computer Programming, Data Processing, Etc. and Services-Prepackaged Software can be combined under the label ICT industry. The 37 SIC industries from the sample were combined and resulted in ten different industries (See table 3). The firm time horizon was measured by using the R&D expenditures of the different firms. These R&D expenditures were available for 51 companies that were also in the CHRD database. The country of the headquarters could be determined for every firm in Compustat (See table 1).
Prior to finishing the data collection process, unlikely data points were removed. For example, firms with too high or too low R&D expenditures in comparison with other firms in that industry were removed from the dataset. This resulted in a final sample of 131 firms that
23
were accused of violating human rights. The firm size could be determined for 125 of these companies, either by revenues, number of employees, or both. There are 19 (37%) companies with a short time horizon and 32 (63%) companies with a long time horizon. The next paragraph describes how time horizon was calculated. Average CSP could have been calculated for 53 companies. The 131 firms have headquarters located in 20 different countries. 47% of these companies have their headquarters in the United States. 24% of the companies were located in Canada. These high percentages make sense, since KLD and Compustat only contain information from companies that are listed on a North American stock exchange. Table 1 show the list of different countries, including the number of companies located there and the corresponding percentage. For the analysis, a final division into two categories is made: US and non-US. 62 firms were in the US category. The other firms were non-US. While testing the hypotheses, country will be a control variable.
24
Country of headquarters Number of firms Percentage Argentina 2 1,5 Australia 3 2,3 Bermuda 1 0,8 Brazil 2 1,5 Canada 31 23,7 China 3 2,3 Colombia 1 0,8 France 1 0,8 Germany 2 1,5 India 2 1,5 Italy 3 2,3 Japan 4 3,1 Jersey 1 0,8 Netherlands 2 1,5 South Africa 2 1,5 South Korea 1 0,8 Spain 1 0,8 Switzerland 2 1,5 United Kingdom 5 3,8 United States 62 47,3
Table 1. Firm’s countries.
3.2. Measurements
Dependent variable
The dependent variable is stakeholder lawsuits. This variable was measured by using data from the CHRD database. CHRD presents data about alleged corporate human rights abuses. The database includes data about judicial and non-judicial actions. A legal court plays an important role in judicial actions, also called legal actions. Non-judicial actions do not include a legal court, but it can involve human rights commissions or likewise institutions. All the cases in the CHRD database were divided in three categories. The first category includes the cases where judicial and/or non-judicial action was taken by stakeholders. The second
25
category exists of violating firms that did not face judicial or non-judicial action. The remaining cases form the third category. These are the cases where the coder was unable to determine whether judicial or non-judicial action was taken. After coding all the cases in one of the three categories, the cases were sorted by company. After that, two calculations were made. First the number of cases of every firm in category one were counted. The sum represents the first measure of stakeholder lawsuits. This is stakeholder activism (absolute). The scores range from 0 to infinite, with steps of one. A score of 10 represents a firm that was confronted with a total of ten judicial or non-judicial actions as a reaction on human rights abuses. The second calculation is the number of cases in category one for a certain company, divided by the total of allegations of human rights abuses by that company. This number is stakeholder activism (relative). The score is somewhere between 0 and 1. A high score represents high stakeholder activism. An example of Occidental Petroleum Corporation will clarify the differences and meanings of stakeholder activism (absolute) and stakeholder activism (relative). The CHRD contains 11 cases where Occidental was accused of violating human rights. In 9 of these cases, judicial and/or non-judicial action was taken by stakeholders. For the remaining 2 cases, there was no judicial or non-judicial action taken. For Occidental, stakeholder activism (absolute) is 9, and stakeholder activism (relative) is 9/11 = 0.82. This last number means that Occidental faces a relatively high degree of stakeholder activism in the form of lawsuits.
Independent variables
The first independent variable is corporate social performance (CSP). This variable was measured by using data from the KLD database. This database displays social ratings on a wide variety of issues for more than 8,700 firms, for the period 1991-2013. The data is coded on a binary scale. If a company has a score of 0 on an item, than it means that this item
26
does not apply to that company. A 1 means that the company is involved in the item. An example item is Waste Management. If a company had policies in use to actively manage the waste it produces in order to reduce it, then this company scores a 1 on this item. Another, yet related, item is hazardous waste. However, a 1 on this item would be negative. All the items indicating a corporate social strength are positive. The items that represent a social concern are negative and had to be reverse coded. In total, 124 items were included. These are all the strengths and concerns in KLD. Of these items, 51 were strengths and the other 73 items were concerns. The items were all a part of one of 13 categories. Table 2 presents these categories and the number of items it includes, differentiating between strengths and weaknesses.
Strengths categories (number of items) Concerns categories (number of items)
Community (7) Community (5)
Corporate governance (7) Corporate governance (11)
Diversity (8) Diversity (6)
Employee relations (12) Employee relations (7)
Environment (9) Environment (13)
Human rights (9) Human rights (10)
Product (4) Product (6) Alcohol (2) Firearms (1) Gambling (2) Military (4) Nuclear power (4) Tobacco (2)
Table 2. KLD items for measuring CSP.
All the strengths and concerns were collected for 53 different companies. These companies are in the CHRD as well as the KLD database. Because this study aims to measure CSP as an average over a longer time period, data was collected from five years. This time period is 2005-2009. These years are the most recent years with all the data available. For the years 2010 and later, data for some items is missing. If a company has a score of 0 on a KLD item,
27
this is neither positive nor negative. All scores of 1 for concerns items are negative and scores of 1 for strength items have a positive effect on CSP. Therefore, annual CSP was calculated by adding all the strength items and deducting all the concerns items. The variable CSP was ultimately calculated by taking the average of the five annual scores. A high CSP represents a company doing well at CSR.
The second independent variable is firm size. This variable was measured by using data from the Compustat database. Two measures were calculated for firm size. First, the number of employees is a measure for firm size (Calof, 1994; Wagner, 1995). The number of employees was available in Compustat for 125 CHRD companies. The smallest company has two employees, while the largest company has 2.2 million employees. The data was collected for the most recent year available. For most companies, this is 2014. The second measure for firm size is the annual revenues (Pearcy & Giunipero, 2008). The annual revenues were available for 118 of the CHRD companies. Again, the most recent data was used. All the revenues are in US dollar. If Compustat presented the revenues in another currency, it was converted to US dollars, using the exchange rate of May 2015. The annual revenues range from $2.96 million to $484 billion.
The last independent variable is firm time horizon. Research and development (R&D) expenditures serve as a good measure to determine the time horizon of a firm (Griffin & Hauser, 1996). It can be argued that firms with a long term focus invest more in R&D than firms with a short term focus. The logic behind this statement is that R&D involves an investment in the short term, while the results only appear in the long term (Griffin & Hauser, 1996). R&D expenditures as a measure of the time horizon only make sense if they are corrected for firm size and industry. Larger firms have in general higher R&D expenditures. Therefore, annual R&D expenditures as a percentage of the annual revenues are calculated. But this does not mean that these scores can be compared for two firms in different industries.
28
Some industries are more knowledge intensive than other industries. All the 52 companies, for which the R&D expenditures and revenues could be collected, belong to an industry. The companies were divided in 39 industries based on the Standard Industrial Classification (SIC). Based on data from Compustat, the revenues and R&D expenditures of all the companies belonging to these 39 SIC industries were collected. For every industry, the mean value of R&D expenditures as a percentage of revenues was calculated. The scores for the 52 companies of this study were compared with this industry central value. A score lower than the mean value indicates a short term focus, while a score higher than the mean value represents a long term focus. In this study’s sample, 33 companies have a long term time horizon, and the other 19 companies have a short term focus.
Control variables
There are two variables that only serve as a control variable in the different statistical analyses testing the conceptual model. The first control variable is country (Table 1). As mentioned before, a distinction is made between US and non-US firms. Because there are only two variables, it is a dichotomous variable. This makes this control variable suitable for the regression analyses that will be done to test the hypotheses.
The second control variable is industry. As mentioned before, the SIC classification is one way to organize firms in industries. However, this classification leads to a fragmentation of dozens of small industries. A lot of these industries are highly related, such as the SIC industries 1311 (Crude Petroleum & Natural Gas) and 2911 (Petroleum Refining). Therefore, the 39 SIC categories are in this study reduced to ten broader industries. Common sense is used to subdivide the industries. After ten industries were determined, a final classification was made into two industry categories: consumer industry and business industry. Firms in consumer industries were dependent on sales to consumers, while firms in business industries
29
mostly sell to other businesses. This final manipulation of the data was done to create a dichotomous variable, which is more suitable for regression analyses than a categorical variable with ten categories. Table 3 includes the ten industries and the classification in consumer or business industry.
Industry name Business or consumer industry
Car industry Consumer
Clothing and footwear Consumer
Household appliances and electronics Consumer
Petroleum industry Consumer
Retail industry Consumer
Chemical industry Business
Financial industry Business
ICT industry Business
Metallurgic industry Business
Wholesale industry Business
Table 3. Industry classification
3.3. Analyses and predictions
The first test in the data analysis is testing for normality of the variables. A lot of statistical tests assume that the data has a normal distribution. Therefore, the data first has to be tested for its normality. Normality will be tested using the statistical values skewness and kurtosis. Furthermore, the Kolmogorov-Smirnov test will be performed, and Q-Q plots will be produced. Based on this information, the normality of the distribution for the quantitative variables can be tested. If a variable significantly deviates from a normal distribution, it can be transformed using logarithms or other computational transformations. It is expected that all the continuous variables, if necessary after a transformation, are normally distributed.
The next step in the analysis process is calculating the correlations between the variables. For all the relations where there is no controlling variable, correlation is based on
30
the Pearson correlation coefficient. If there is a controlling variable, the analysis is done using partial correlation. The correlation involving the variable firm time horizon is a special case. This is a continuous dichotomy variable. This means that there are two possible values (short term or long term). But for these values, an underlying continuum exists. A firm is either short term focused, or long term focused. But within one category, not all the firms are to the same degree long term or short term focused. Some firms are extremely long term focused. Others are to a lesser degree long term focused. The biserial correlation is the right correlation coefficient for these kinds of variables. It is expected that the correlations between the main variables are all according to the conceptual model. This means that the firm size (number of employees and revenues) is positively related to stakeholder lawsuits (relative and absolute). The expectation is that firm time horizon is negatively related to stakeholder lawsuits. The same applies to the relation between CSP and stakeholder lawsuits. The better the CSP is for a firm, the less stakeholder lawsuits are filed against this firm.
The final stage in the analysis process is the regression analysis. Simple regression is the method used to test the hypotheses. After controlling for the control variables industry and firm size (employees and revenues), it is expected that all the four hypotheses are confirmed.
All the steps in the data analysis will be performed using the IBM’s statistical software program SPSS, version 23.
31
4. Results
Based on theoretical arguments, four hypotheses were developed in the literature review. In the methodology section, the data collection process and the variables were examined. Section 3 ended with an overview of the different analyses necessary to test the hypotheses, and predictions of the outcomes of these analyses.
The results section starts with assumptions of regression analysis, and transformations of the variables that are necessary to cope with the assumptions. After that, a correlation analysis is performed in order to create some first insights in the relationships between the different variables. Ultimately, the results of the regression analyses which test the hypotheses are presented.
4.1. Assumptions and transformations
There are two kinds of variables in this study: dichotomous variables and continuous variables. The hypotheses will be tested by simple regression analysis. One assumption for regression analysis is that the data of the continuous variables is normally distributed (Field, 2009). Data can be checked for normality using various methods. One of these methods is analyzing the descriptive data items skewness and kurtosis. Other methods for testing normality are the Kolmogorov-Smirnov test (K-S test) and normal Q-Q plots. Normal Q-Q plots is the most suitable method for testing relatively large samples. Skewness, kurtosis, and the K-S test are applicable to relatively small samples with less than 100 observations.
Average CSP ranges from -9.60 to 8.60 (M = -1.09, SD = 3.90). The means and standard deviations are summarized in table 4. CSP is normally distributed, with skewness of 0.15 (SE = .33) and kurtosis of 0.21 (SE = .64). The two variables measuring firm size are the number of employees and the annual revenues. The number of employees ranges from 2 to 2.2 million (M = 61,100, SD = 207,906). Number of employees is not normally distributed,
32
with skewness of 9.07 (SE = .22) and kurtosis of 92.1 (SE = .43). In order to correct for this positive skew, a new variable is created: LnEmpl. This is the natural logarithm of the old variable number of employees. This variable was tested for normality using the K-S test. LnEmpl, D(47) = 0.085, ns, is not significantly non-normal. With other words, it can be assumed that LnEmpl has a normal distribution. The second variable measuring firm size is annual revenues. Annual revenues range from $2.96 million to $484 billion (M = 42,628 mln, SD = 89,314 mln). Annual revenues are not normally distributed, with skewness of 3.16 (SE = .22) and kurtosis of 10.2 (SE = .44). Again, a new variable is created: LnRev. This is the natural logarithm of the annual revenue variable. This new variable was tested for normality using the K-S test. LnRev, D(47) = 0.073, ns, is normally distributed. The two variables measuring stakeholder activism are relative stakeholder activism and absolute stakeholder activism. The number of observations for these two variables are relatively high (N = 131). Therefore, Q-Q plots is a better method to test for normality than using skewness and kurtosis. Relative stakeholder activism ranges from 0 to 1 (M = .52, SD = .41). Based on the Q-Q plot in figure 2, it can be argued that relative stakeholder activism is normally distributed. Absolute stakeholder activism ranges from 0 to 22 (M = 2.15, SD = 3.58). This variable is strongly positively skewed. To correct for this positive skewness, the new variable RecAbsStkAct is created. This variable is the reciprocal of the original variable absolute stakeholder activism. Based on the Q-Q plot in figure 3, it can be argued that RecAbsStkAct is normally distributed.
33 Figure 2. Q-Q plot of Relative stakeholder activism.
34 Figure 3. Q-Q plot of RecAbsStkAct.
In short, the variables average CSP, LnEmpl, LnRev, relative stakeholder activism, and RecAbsStkAct are all normally distributed. These variables will be used as continuous variables in the regression analyses.
4.2. Correlation analysis
To get more insights in the relationships between the different variables, a correlation analysis has been done. The results from this analysis are displayed in table 5 on the next page. The right upper halve of the table gives the Pearson correlation coefficient between two variables. The left lower halve gives N, the number of corresponding cases for the two variables. Cases have been removed pair wise.
35 Table 5. Table of correlations. * p<.05, ** p<.01
36 Table 6. Number of items.
What immediately catches the attention when looking at the table of correlations, is that not so many variables are significantly correlated. For example, average CSP does not correlate with any other variable. This is somewhat surprising, since it is expected that high CSP leads to more stakeholder activism. However, this could still be true, since correlations do not say anything about causation. Time horizon is the other variable that does not correlate with other variables.
But there are indeed a few significant correlations. Logically, the two measures for firm size, number of employees and annual revenues, have a strong positive correlation, r = .85, p <.01. This makes sense, because the largest firms in terms of revenues are likely to have many employees. Another logical outcome is the negative correlation coefficient between relative and absolute stakeholder activism, r = -.74, p < .01. The correlation coefficient is
37
negative, because absolute stakeholder activism has undergone a reciprocal transformation, which reversed the scores. This means that in reality, absolute and relative stakeholder activism are positively related. Firm size in terms of annual revenues is significantly correlated to relative stakeholder activism, r = .24, p < .05, and absolute stakeholder activism, r = -.37, p < .01. Both correlations are positive, because the scores of absolute stakeholder activism are reversed. These correlations cope with the expectations. Larger firms are more likely to face more stakeholder activism in the form of lawsuits. The other measure for firm size, the number of employees, is also significantly correlated with absolute stakeholder activism, r = -.25, p < .01. However, the number of employees is not significantly related to relative stakeholder activism.
Finally, the two control variables country and industry correlate with some other variables. The variable country has a weak negative correlation with the number of employees, r = -.18, p < .05. This means that the US firms are on average larger than non-US firms. Industry is negatively correlated with the number of employees, r = -.32, p < .01, and annual revenues, r = -.27, p <.01. In other words, firms in consumer industries are on average larger than firms in business industries.
4.3 Testing the hypotheses
A multiple linear regression was calculated to predict relative stakeholder activism based on CSP, while controlling for firm size (employees and revenues), country, and industry. There was no significant regression equation found (F(5, 46) = 1.188, ns), with an R2 of .120 (Table 5). The control variables country, industry, number of employees, and annual revenues did not have a significant effect on the relation between CSP and stakeholder activism. The same analysis was done to predict absolute stakeholder activism based on CSP, controlling for the same variables. There was no significant regression equation found (F(5, 46) = 1.502, ns),
38
with an R2 of .155. Again, the control variables country, industry, number of employees, and annual revenues did not have a significant effect on the relation. These results do not support hypothesis 1. CSP does not predict stakeholder activism.
B SE B Β Step 1 Constant 0.38 0.268 Employees -0.08 0.045 -.469 Revenues 0.10 0.047 .598 Industry -0.10 0.132 -.115 Country -0.04 0.127 -.050 Step 2 Constant 0.38 0.27 Employees -0.09 0.05 -.54 Revenues 0.12 0.05 .68* Industry -0.10 0.13 -.12 Country -0.05 0.13 -.06 CSP 0.01 0.02 .09
Note: R2 = .120 for Step 1, ΔR2 = .007 for Step 2 (ns). * p < .05.
Table 7. Regression testing relative stakeholder activism based on CSP.
B SE B Β Step 1 Constant 0.95 0.206 Employees 0.03 0.035 .232 Revenues -0.08 0.036 -.563* Industry 0.02 0.101 .028 Country 0.001 0.097 .002 Step 2 Constant 0.95 0.21 Employees 0.03 0.04 .24 Revenues -0.08 0.04 -.58 Industry 0.02 0.10 .03 Country 0.002 0.10 .003 CSP -0.001 0.01 -.01
Note: R2 = .155 for Step 1, ΔR2 = .000 for Step 2 (ns). * p < .05.
39
To test hypothesis 2, four multiple regressions were calculated to predict stakeholder activism based on firm size. All the regressions were controlled for the effects of the control variables country and industry. First, a regression was calculated to predict relative stakeholder activism based on the number of employees. This regression is not significant (F(3, 124) = 0.886, ns), with an R2 of .147 (Table 7). The control variables do not have a significant effect on the relation either. The second regression was done to predict relative stakeholder activism based on annual revenues. A significant regression equation was found (F(3, 117) = 6.795, p < .05), with an R2 of .062 (Table 8). The probability to face stakeholder activism in case of human rights abuses is 0.24 + 0.04 higher when the logarithm of revenues increases with 1. Country and industry do not have a significant effect on this relation. The third regression was calculated to predict absolute stakeholder activism based on the number of employees. A significant regression equation was found (F(3, 124) = 2.912, p < .05), with an R2 of .067 (Table 9). The reciprocal of the number of actions by stakeholders against human rights violating firms is 0.85 – 0.03 lower when the logarithm of employees increases with 1. Country and industry do not have a significant effect on the relationship between the number of employees and absolute stakeholder activism. The fourth regression was calculated to predict absolute stakeholder activism based on the annual revenues of a firm. A significant equation was found (F(3, 117) = 6.208, p < .01), with an R2 of .118 (Table 10). The reciprocal of the number of actions by stakeholders against human rights violating firms is 1.00 – 0.05 lower when the logarithm of revenues increases with 1. Country and industry did not have a significant effect on the relation between annual revenues and absolute stakeholder activism. These results provide partial support for hypothesis 2. Larger firms in terms of revenues have a positive effect on stakeholder activism. Larger firms in terms of employees only have a positive effect on the absolute number of stakeholder actions in the firm is violating human rights.
40 B SE B Β Step 1 Constant 0.58 0.06 Industry -0.12 0.08 -.14 Country -0.003 0.08 -.004 Step 2 Constant 0.51 0.16 Industry -0.11 0.08 -.13 Country 0.000 0.08 .000 Employees 0.01 0.02 .04
Note: R2 = .141 for Step 1, ΔR2 = .002 for Step 2 (ns).
Table 9. Regression testing relative stakeholder activism based on number of employees.
B SE B Β Step 1 Constant 0.58 0.06 Industry -0.12 0.08 -.14 Country -0.003 0.08 -.004 Step 2 Constant 0.24 0.16 Industry -0.07 0.08 -.08 Country -0.004 0.08 -.01 Revenues 0.04 0.02 .21*
Note: R2 = .020 for Step 1, ΔR2 = .042 for Step 2 (p < .05). * p < .05.
Table 10. Regression testing relative stakeholder activism based on annual revenues.
B SE B Β Step 1 Constant 0.54 0.05 Industry 0.07 0.06 .11 Country -0.02 0.06 -.02 Step 2 Constant 0.85 0.12 Industry 0.03 0.06 .04 Country -0.03 0.06 -.05 Employees -0.03 0.01 -.25*
Note: R2 = .012 for Step 1, ΔR2 = .055 for Step 2 (p < .01). * p < .05.
41 B SE B Β Step 1 Constant 0.54 0.05 Industry 0.07 0.06 .11 Country -0.02 0.06 -.02 Step 2 Constant 1.00 0.12 Industry 0.01 0.06 .01 Country -0.01 0.06 -.02 Revenues -0.05 0.01 -.37*
Note: R2 = .012 for Step 1, ΔR2 = .128 for Step 2 (p < .001). * p < .001.
Table 12. Regression testing absolute stakeholder activism based on annual revenues.
Two multiple regression analyses are done to test hypothesis 3. First, a regression was calculated to predict relative stakeholder activism based on the time horizon of the firm. This relation is controlled for the variables firm size (number of employees and annual revenues), country, and employees. No significant regression equation was found (F(5, 51) = 1.307, ns), with an R2 of .124. The control variables do not have a significant effect. The second regression was calculated to predict absolute stakeholder activism based on the time horizon of the firm. Again, this relation was controlled for the four variables number of employees, annual revenues, country, and industry. No significant regression equation was found (F(5, 51) = 1.686, ns), with an R2 of .155 (Table 12). The control variables do not have a significant effect here as well. These results do not support hypothesis 3.Time horizon is a variable that does not predict the extent of stakeholder activism if a firm violates human rights.
42 B SE B Β Step 1 Constant 0.38 0.25 Employees -0.08 0.04 -.47 Revenues 0.10 0.04 .60* Industry -0.10 0.12 -.12 Country -0.04 0.12 -.05 Step 2 Constant 0.43 0.27 Employees -0.07 0.04 -.45 Revenues 0.10 0.05 .58* Industry -0.10 0.13 -.11 Country -0.05 0.12 -.06 Time horizon -0.06 0.12 -.07
Note: R2 = .120 for Step 1, ΔR2 = .004 for Step 2 (ns). * p < .05.
Table 13. Regression testing relative stakeholder activism based on firm time horizon.
B SE B Β Step 1 Constant 0.95 0.19 Employees 0.03 0.03 .23 Revenues -0.08 0.03 -.56* Industry 0.02 0.10 .03 Country 0.001 0.09 .002 Step 2 Constant 0.95 0.21 Employees 0.03 0.03 .24 Revenues -0.08 0.04 -.57* Industry 0.02 0.10 .03 Country 0.000 0.09 .000 Time horizon -0.01 0.09 -.01
Note: R2 = .155 for Step 1, ΔR2 = .000 for Step 2 (ns). * p < .05.
Table 14. Regression testing absolute stakeholder activism based on firm time horizon.
Hypothesis 4 predicts that the negative relation between CSP and stakeholder activism is stronger for large firms than for small firms. This hypothesis will be tested by a multiple regression model comparing the independent effects of firm size and CSP with the interaction
43
effect of these two variables. To avoid problems with multicollinearity, three new variables are created: the mean centered values for CSP, number of employees, and revenues. After that, two new variables are created: the interaction variable of CSP with the number of employees, and the interaction variable of CSP with annual revenues. The number of employees does not have a moderating effect on the non-significant relation between CSP and relative stakeholder activism (F(3, 46) = 0.206, ns), with R2 of .014 (Table 13). The annual revenues also have no significant moderating effect on the relation between CSP and relative stakeholder activism (F(3, 46) = 1.675, ns), with R2 of .105 (Table 14). The number of employees do not have a significant moderating effect on the relation between CSP and absolute stakeholder activism as well (F(3, 46) = 1.298, ns), with R2 of .083 (Table 15). The final regression analysis shows a significant moderation effect of the annual revenues on the relation between CSP and absolute stakeholder activism (F(3, 46) = 3.976, p < .05), with R2 of .217 (Table 16). The negative relation between CSP and absolute stakeholder activism is stronger for large firms than for small firms (β = .363, p < .05). For all the moderation analyses, there are no control variables included. Therefore, the tables below only include step 1, where the interaction effect is compared with the effects of the two independent variables.
B SE B β Step 1 Constant 0.52 0.06 CSP -0.003 0.03 -0.02 Employees 0.01 0.03 0.07 Interaction CSP and Employees -0.003 0.01 -0.07 Note: R2 = .014.
Table 15. Moderation effect of number of employees on relation between CSP and relative stakeholder activism.
44 B SE B Β Step 1 Constant 0.50 0.06 CSP 0.02 0.02 0.15 Revenues 0.03 0.03 -0.17 Interaction CSP and Revenues -0.01 0.01 -0.29 Note: R2 = .105.
Table 16. Moderation effect of annual revenues on relation between CSP and relative stakeholder activism. B SE B β Step 1 Constant 0.56 0.05 CSP 0.01 0.02 0.06 Employees -0.03 0.02 -0.24 Interaction CSP and Employees 0.003 0.01 0.09 Note: R2 = .083.
Table 17. Moderation effect of number of employees on relation between CSP and absolute stakeholder activism. B SE B β Step 1 Constant 0.59 0.05 CSP -0.01 0.01 -0.17 Employees -0.04 0.02 -0.29 Interaction CSP and Employees 0.01 0.01 0.36* Note: R2 = .217. * p < .05.
Table 18. Moderation effect of number of employees on relation between CSP and absolute stakeholder activism.
