Culture and Corporate Social Responsibility
The effect of CEOs’ national culture on CSR Performance and the
moderating role of Recent Financial Performance
MSc International Business and Management, Semester 2A 2018
Master Thesis
Author:
Svenja Belke (S2751305); Email: s.belke@student.rug.nl
Supervisor:
Dr. O. Lindahl; Email: olof.lindahl@fek.uu.se
Co-assessor:
Dr. S. Gubbi; Email: s.r.gubbi@rug.nl
Faculty of Economics and Business
University of Groningen
Duisenberg Building, Nettelbosje 2, 9747 AE Groningen, Netherlands
P.O. Box 800, 9700 AV Groningen, Netherlands
http://www.rug.nl/feb
2
Abstract
Culture as well as Corporate Social Responsibility depicts current and increasingly important topics in today’s global business environment. CSR performances of firms vary and influencing factors are diverse. The upper echelon theory supports the argument that individual values of CEOs influence organizational decision making and practices. Using it as a theoretical basis, this thesis fills the gap by testing the individual-level effect of CEOs’ national culture on CSR performance in public organizations operating in various industries spread over Europe, Asia, the US and Australia in order to test if culture plays a pivotal role in determining CSR practices. In addition, the moderating role of Recent Financial Performance on the aforementioned relationship will be tested to analyze whether CEOs are restricted by and dependent on financial outcomes with regard to their impact on CSR activities. Using multiple regression analysis, CEOs’ cultural effects on CSR performances of 80 companies located in 18 different countries and operating in 27 different industries are investigated, leading to surprising findings. Accordingly, the negative effect of CEOs showing high levels of power distance and masculinity was statistically supported. However, the negative impact of individualistic CEOs and the positive effect of uncertainty avoiding CEOs on CSR performance as regularly described in existing research could not be found. Also, against the notions and predictions of the majority of researchers, recent financial performance does not seem to critically influence the aforementioned relations. These findings have important practical and empirical implications which will be discussed at the end.
Key words: CEOs’ national culture, CSR Performance, cultural dimensions, Hofstede,
3
Table of Content
1. Introduction 5
2. Literature Review 8
2.1 CEO national Culture 8 2.1.1 Power Distance 10
2.1.2 Uncertainty Avoidance 10
2.1.3 Individualism 10
2.1.4 Masculinity 11
2.2 CSR Performance 11
2.3 Recent Financial Performance 13
2.4 CEO national culture and CSR Performance 13
2.4.1 Power Distance and CSR Performance 14
2.4.2 Uncertainty Avoidance and CSR Performance 15
2.4.3 Individualism and CSR Performance 15
2.4.4 Masculinity and CSR Performance 16
2.5 The Moderating Role of Recent Financial Performance 17
2.6 Conceptual Model 20
3. Methodology 21
3.1 Variables and Measurement 21
3.1.1 Dependent Variable 21 3.1.2 Independent Variable 22 3.1.3 Moderator 22 3.1.4 Control Variables 23 3.2 Data Collection 24 3.3 Sample 25 3.4 Analysis 26 4. Empirical Results 28
4.1 Assumptions of Multiple Regression 28
4.1.1 Outliers 28
4.1.2 Linearity 28
4.1.3 Normality 30
4.1.4 Multicollinearity 30
4
4.1.6 Heteroscedasticity 31
4.2 Multiple Regression Analysis 31
4.2.1 Hierarchical Multiple Regression 32
4.2.2 Moderation Analysis 34
5. Conclusion 36
5.1 Discussion of Results 36
5.2 Limitations and Future Research 39
5.3 Relevance and Practical Implications 40
References 42
5
1. Introduction
Corporate Social Responsibility (CSR) is a current topic that arouses public interest on a daily basis worldwide. It presents a global concept which is neither restricted to specific areas nor industries anymore. Researchers define CSR as ‘actions that appear to further some social good, beyond the interest of the firm and that which is required by law’ (McWilliams and Siegel, 2001, p. 117). A growing number of organizations integrate social, ethical and environmental routines into their business operations due to various stakeholder groups that demand them to allocate resources to these activities (McWilliams and Siegel, 2001). Nevertheless, CSR practices as well as CSR performances of companies vary (Chin, Hambrick, and Trevino, 2013). CSR performance cannot be linked to a single description in existing literature. However, as researchers clearly distinguish between firms that ‘talk’ and firms that ‘walk’ in terms of CSR engagements (Wickert, Scherer, and Spence, 2016), this thesis specifically refers to CSR performance as the actual implementation of ethical, social, and environmental actions (‘walk’) as evaluated by external rating organizations. Thereby, it widely disregards the CSR ‘talk’, meaning the pure outbound communication of firms presenting their own CSR initiatives, whilst focusing on the objective assessment of the firms’ ‘walk’.
One crucial factor that might influence and explain the heterogeneity of organizational CSR performances is the cultural background of the organization’s top manager, or more precisely, the CEO’s national culture. Central to this thesis is the upper echelon theory developed by Hambrick and Mason (1984), which states that choices of CEOs are individually made and influenced by personal experiences and values that critically influence multiple levels of organizational performance. Consequently, it can be assumed that the national and cultural background of CEOs plays an important role in terms of CSR practices and resulting performances of firms.
6 Until now, the majority of researchers studied the effect of CEO personality types or values on firms’ financial performance (e. g. Tomczyk, Lee, and Winslow, 2013). While a significant effect on financial performance was found, the specific impact on organizational CSR performance has not been analyzed. Nevertheless, research suggests that CEOs play an important role in order to influence companies’ CSR practices as they can act as the drivers of CSR activities (Godos-Díez, Fernández-Gago, and Martínez-Campillo, 2011), showing a crucial effect of the individual-level impact on CSR practices, but not on the organizational CSR performance in particular. Research additionally infers that firms seem to integrate business ethics that are congruent with their national cultural values (Svensson, Wood, Singh, Carasco, and Callaghan, 2009), but studies lack evidence that these cultural values affect CSR performance. Thanetsunthorn (2015) particularly investigated the impact of culture on CSR performance in Asian and European countries and found a significant effect of national culture on CSR. However, the researcher uses the corporate headquarter as a basis for measuring the national culture of the respective country. Given the aforementioned presumed effect of the individual CEOs’ national culture, Thanetsunthorn’s approach is questionable, as nowadays international managers oftentimes lead firms in a foreign country (Linehan and Scullion, 2002), meaning that the CEOs’ national culture and the national culture prevalent in the country of the corporate headquarter are not necessarily identical. Therefore, this thesis wants to take this approach further by focusing on individual CEOs’ cultural values and in addition, integrates a higher number of regions and countries, aiming to increase the generalizability of the results.
7 the internal financial opportunities a company provides or not. Taking all this into consideration, one finds a clear research gap regarding the integration of the moderating effect of recent financial performance on the relationship between CEO’s national culture and CSR performance. Although the broad effect of the national culture dominant in a country on CSR performance has been analyzed, empirical investigations remain scarce. More specifically, the individual-level effect of CEOs’ cultural values has not been studied and no proof can be found that this relationship is moderated by a firm’s recent financial performance.
Therefore, the resulting research question of this thesis is:
Does a CEO’s national culture affect organizational CSR performance and does a firm’s recent financial performance moderate this relationship?
Thereby, this thesis can contribute to existing research and aims to deepen the understanding of CSR performance variations between organizations by recognizing the specific effect of a CEO’s national culture. Since it integrates the moderating effect of recent financial performance, it further expands current results and adds another aspect by focusing on an individual level explanation for CSR performance variations.
8
2. Literature Review and Hypotheses Development
This section contains a more detailed presentation of the main literature on which this thesis is based and gives an overview of the most important concepts and definitions, namely culture, CSR performance, and recent financial performance. In addition, hypotheses based on this literature are developed and finally visualized in a conceptual model.
2.1 Culture
Culture plays a major role in the business environment of today as the number of international exchanges and relationships grows and therefore, global communication is vital (Melluish, 2014). Important implications, as, for example, differences regarding values and underlying behavior between national cultures (Hofstede, 2011) and thus, relating leadership practices (Dorfmann & Tate, 1997) need to be considered and reflected by CEOs in order to act or react successfully on an international basis.
The theoretical concept of culture has been approached in numerous ways. Researchers aim to understand and explain differences in peoples’ behavior in international interactions in order to facilitate encounters as well as the initiation and maintenance of business relationships and therefore, have tried to develop ways of identifying and measuring culture and its traits. One of the first and most famous measurements of culture in the management literature was developed by Hofstede (2011, p. 3), who defines culture as ‘the collective programming of the mind that distinguishes the members of one group or category of people from others’. Initially collecting data on another phenomenon by using data of the multinational organization IBM, he discovered systematic differences in answers of study participants coming from a variety of countries. As a result, he created the original four dimensions to describe and categorize the main characteristics of a country’s culture (Hofstede, 2011):
1. Power Distance: refers to the acceptance of inequality in terms of power 2. Uncertainty avoidance: refers to the levels of stress in unstructured situations 3. Individualism vs. Collectivism: refers to the embeddedness into groups 4. Masculinity vs. Femininity: refers to the primary values in a society
9 5. Long Term vs. Short Term Orientation: refers to the selection of peoples’ action,
namely the future or the present and past
6. Indulgence vs. Restraint: refers to the degree of allowance of fun and enjoyment
This approach resulted in several extensions as well as criticism. Examples of extensions are the GLOBE studies (House, 2004), that aimed to extend the initial six dimensions of Hofstede by providing more detailed information. In addition, Schwartz (2006, p. 136), who defined culture as ‘the rich complex of meanings, beliefs, practices, symbols, norms, and values prevalent among people in a society’, developed seven cultural value orientations that build three value dimensions also seeking to create a more detailed and specific categorization (Autonomy -Intellectual and Affective- vs. Embeddedness, Harmony vs. Mastery, and Hierarchy vs. Egalitarianism).
However, these dimensions and value orientations show clear similarities to the six Hofstede dimensions, both in terms of categorization and content. Until now, Hofstede’s approach to measure culture remains the most dominant and most popular categorization among researchers (Beugelsdijk et al., 2017). It is seen as the most comprehensive measurement of culture and is widely used in the field of international business studies (Wilhelm and Gunawong, 2016; Beugelsdijk et al., 2017). Even though research often criticizes Hofstede’s work for not being applicable anymore due to the development of societies, Beugelsdijk, Maseland, and Hoorn (2015) show that countries’ scores relative to other country scores are rather stable over time which shows that Hofstede’s work can still be seen as an appropriate tool to distinguish cultural value systems on a national level.
10 of power distance, uncertainty avoidance, individualism, and masculinity on CSR performance representing the concept of CEO national culture.
2.1.1 Power Distance
Power distance refers to the degree to which people accept and presume unequal power
relationships (Hofstede, 2011). In countries that are characterized by high power distance, inequality in terms of power is not only enacted by the higher ranks of the society, but it is equally expected from the lower ranks of the societal members within a country (Hofstede and McCrae, 2004). There are hierarchical structures where superiors differentiate themselves from subordinates and vice versa (Hofstede, 1983). In these countries, the claim of legitimacy is not questioned and children are taught obedience (Hofstede, 2011). In contrast, countries with low levels of power distance show structures of independence, and their societies aim to reduce inequalities (Hofstede, 1983). Power needs to be legitimized and is not seen as a given precondition. Hofstede (2011) clearly states that inequality prevails in each and every society; however, some countries are dominated by higher levels of inequality than others.
2.1.2 Uncertainty Avoidance
Hofstede (2011) describes uncertainty avoidance as the level of stress people perceive in unstructured situations. In particular, societies that show high levels of uncertainty avoidance seek to prevent unknown and novel situations from occurring. Therefore, risks are aimed to be minimized and routines as well as clear rules and structures are implemented and favored (Hofstede, 2011). Furthermore, individuals from countries scoring high on uncertainty avoidance prefer certainty and strive for consent. Safety and security are central to uncertainty avoiding cultures and there is a strong belief in an absolute truth (Hofstede, 1983), which fosters the reliance on experts (Hofstede and McCrae, 2004). As opposed to this, uncertainty accepting societies are characterized by higher flexibility, lower stress and anxiety as well as by the refusal of rules (Hofstede, 2011). People in these countries are generally described as being more permissive of different opinions (Hofstede, 2011).
2.1.3 Individualism vs. Collectivism
11 McCrae, 2004). Autonomy and self-orientation are central prerequisites in individualistic societies and the individual is expected to care of him- or herself. Contrastingly, collectivistic cultures value harmony and strong group cohesiveness (Hofstede, 1983). Decisions should be made at the group-level as there is a high ‘we-consciousness’ (Hofstede, 2011). Relationships are extremely important for collectivistic societies, which are defined by high stability and longevity (Hofstede, 1983). Hofstede (1983) points out that the term ‘collectivism’ does not have political connotations, but it relates to societal structures.
2.1.4 Masculinity vs. Femininity
Masculinity vs. Femininity refers to the primary values that are prevalent within a society.
More specifically, it indicates the extent to which traditional male or female values dominate a society (Hofstede, 1980). For example, cultures that score high on masculinity place a higher importance on work in contrast to family (Hofstede, 2011). Feminine cultures can be characterized by a high sympathy for weaker members of the society, whereas strong and powerful people are venerated in masculine cultures (Hofstede, 2011). In addition, feminine cultures desire a high quality of life and the environment by being people-oriented, while masculine cultures value achievements and independence by being money-oriented (Hofstede, 1983). Prevailing values foster the endeavor to be the best, as ‘big and fast are beautiful’ (Hofstede, 1983, p. 63). Hofstede (2011) explains that this dimension refers to deep-seated and sometimes unconscious ideals of a culture that create taboos, which cannot easily be explicitly debated.
2.2 CSR Performance
12 Researchers’ definitions of CSR vary. Nevertheless, authors agree on basic elements. One of the first definitions was created by Carroll (1979), who includes certain core aspects of firms, namely economic, legal, ethical, and discretionary components that together form CSR. In contrast, more recent literature generally focuses on companies’ activities that extend legal requirements and define CSR as ‘actions that appear to further some social good, beyond the interest of the firm and that which is required by law’ (McWilliams and Siegel, 2001, p. 117). Consequently, CSR can be described as voluntary actions that exceed firms’ economic and legal concerns. Friedman (1970) denies the corporate role to operate ethically, sustainably, and environmentally friendly beyond legal requirements, as executives should only focus on increasing their shareholder value. In contrast, proponents of the stakeholder theory assign an important role to companies in order to satisfy not only shareholders, but also several groups that have an interest in the firm’s activities, such as employees, customers, or local communities (Freeman, 1984). According to this latter approach, CSR is a highly important concept that companies should integrate into their business operations.
13
2.3 Recent Financial Performance
One can find various definitions for the term financial performance. Therefore, these shall be addressed and additionally, it shall be clarified how this term is used in this thesis. Commonly, definitions of financial performance are closely linked to its corresponding measurements.
A lot of definitions refer to organizational performance as ‘organizational outcomes in terms of increased customer loyalty and consumption’ (Bhattacharya and Sen, 2003). This thesis refers to an organization’s financial performance as ‘a company’s [economic] viability […]’ (Orlitzky et al., 2003, p. 411). Former research makes use of the term financial performance in terms of organizational accounting outcomes, such as profits, sales, return on assets and return on equity (e. g. Griffin and Mahon, 1997). Rather than reflecting market fluctuations, accounting measures indicate the internal efficiency of an organization (Vitezic et al., 2012), which can be critical in terms of CSR investments. Financial performance can be further distinguished by a firm’s profitability, asset utilization, growth, and liquidity (Griffin and Mahon, 1997). Integrating the initial definition of CSR, these actions are voluntary and beyond the interest of the firm (McWilliams and Siegel, 2001), meaning that CSR practices do not only need managerial attention and support, but also additional financial investments. Opposed to Bihari and Prahadan (2011) who emphasize the importance of external financing opportunities, Vitezic et al. (2012) state that without an organization’s internal profitability, there will be no long-term survival of the firm and voluntary CSR investments seem to be unlikely.
Therefore, a company’s profitability seems to be an important aspect with regard to CSR implementations and is consequently used as a proxy for financial performance in this thesis.
2.4 CEO National Culture and CSR Performance
14 and hence, the CSR performance of a firm. Using the dimensions of Hofstede (2011), the following relationships are expected between a CEO’s national culture and CSR Performance.
2.4.1 Power Distance and CSR performance
Referring to section 2.1.1, power distance refers to the degree to which people accept and presume unequal power relationships (Hofstede, 2011). To extend this definition, cultures that show a high power distance can be described as countries where corruption is rather common, where scandals are hidden, and where ‘power is a basic fact for society antedating good or evil: its legitimacy is given’ (Hofstede, 2011, p. 9).
Blodgett, Lu, Rose, and Vitell (2001) conclude that lower power distance indexes correlate with an increase in ethical sensitivity with regard to stakeholders while people scoring higher on power distance follow their own interests. In addition, the latter are more inclined to consider questionable business operations as ethical (Cohen, Pant, and Sharp, 1996). This supports the argument that individuals from countries displaying a high power distance are not likely to engage in social and ethical activities that positively affect CSR performance. In addition, employees represent one of several stakeholder groups that CSR activities address. In cultures showing a high power distance, employees can be expected to be obedient and do not have extensive opportunities to present their opinion to management. Referring to Hofstede’s description, they can rather be expected to accept deficiencies. Vice versa, CEOs can be assumed to not accept or even listen to the employee’s voice. Moreover, public pressure to persuade organizations to integrate social and ethical practices is expected to be lower in high power distance cultures. There are two reasons for this statement: first, people in high power distance cultures seem to accept the actions of CEOs as they usually have less power in the community and inequality is widely accepted. Secondly, scandals in these societies are usually covered up; therefore, if environmental or social accidents or deficiencies occur, they can be assumed to be hidden by CEOs in order to further justify their superior position. Communication and agreement between management and stakeholders do not seem to be very likely, as their roles are taken as given without much interaction. Accordingly, high power distance is expected to have a negative impact on CSR performance as CEOs are assumed to engage less in CSR activities than CEOs from low power distance countries. The first hypothesis is therefore:
15
2.4.2 Uncertainty Avoidance and CSR Performance
As previously mentioned, Hofstede (2011) describes uncertainty avoidance as the level of stress people perceive in unstructured situations. More specifically, cultures that show high levels of uncertainty avoidance seek to prevent unknown and novel situations from occurring and prefer routines as well as clear rules and structures (Hofstede, 2011). Furthermore, individuals prefer certainty and strive for consent.
Research states that people from societies that score very low on uncertainty avoidance are more likely to show risky behavior (Hofstede, 1984). Taking this further, Rallapalli, Vitell, Wiebe, and Barnes (1994) conclude that risky behavior is more likely to result in unethical practices, which could be an indicator that low uncertainty avoidance of CEOs has a negative impact on organizational CSR performances. In addition, it can be assumed that CEOs from uncertainty avoiding cultures favor the creation of codes of conducts with their several stakeholder groups in order to be able to follow detailed rules and regulations and finally, to have clear business structures. Thus, CSR can be expected to present a means to reduce social and environmental uncertainties. Considering the increasing public expectation of communities and stakeholders to implement sustainable, social, ethical, and environmental practices, CEOs from uncertainty avoiding countries are likely to engage in CSR activities in order to prevent their business from risks such as negative reputational consequences or other public turmoil of activist groups or non-governmental organizations. It is assumed that these CEOs try to avert any threats that might involve negative consequences. Accordingly, high uncertainty avoidance is expected to have a positive effect on CSR performance, as CEOs are likely to engage more on CSR initiatives than CEOs from uncertainty taking countries. The second hypothesis is therefore:
H2: The higher the level of uncertainty avoidance of a CEO’s national culture, the higher will be the CSR performance of the company.
2.4.3 Individualism and CSR Performance
16 contrast, individualists seek to pursue their own goals and value independence and self-reliance (Markus and Kitayama, 1991). Hence, individualistic cultures focus on their ‘I-consciousness’, whereas collectivistic cultures have a ‘we-consciousness’ (Hofstede, 2011). CSR activities emphasize the benefits of the community and contain voluntary actions that create social and environmental good for society beyond the financial success of the individual company. Given the description and corresponding characteristics of individualistic cultures, such CEOs are likely to engage less in CSR, as they primarily seek to increase the company’s success and hence, pursuing their own goals. They will only invest in CSR if it will help to achieve their personal goals. Opposed to this, collectivistic managers can be assumed to invest more in CSR activities in order to increase group-welfare and to create benefits for the community they are part of due to their strong in-group feelings. As a consequence, CSR performance of organizations led by individualistic CEOs is assumed to be minimized compared to collectivistic CEOs. The third hypothesis is therefore:
H3: The higher the level of individualism of a CEO’s national culture, the lower will be CSR performance of the organization.
2.4.4 Masculinity and CSR Performance
As described in section 2.1.4, masculinity refers to the primary values that are prevalent within a society. More specifically, in countries that score high on masculinity work is more dominant than family (Hofstede, 2011). Masculine societies tend to be dominated by values such as success, strength, and personal achievements, whereas feminine cultures display sympathy for weaker people and generally show caring traits.
17 expected to negatively impact CSR performance as CEOs of these cultures are less likely to invest and engage in CSR initiatives than CEOs from feminine cultures. The fourth hypothesis is therefore:
H4: The higher the level of masculinity of a CEO’s national culture, the lower will be the CSR performance of the organization.
2.5 The Moderating Role of Recent Financial Performance
The relationship between financial performance and CSR performance has been at the center of numerous discussions. However, positions were intertwined and contradictory (Ullmann, 1985). Alexander and Buchholz (1978), Ullmann (1985), and Shane and Spicer (1983) show ambiguous results, while more recent literature concludes that there is a positive relationship between an organization’s recent financial performance and corresponding CSR activities. For instance, Vitezic et al. (2012) and Orlitzky et al. (2003) state that firms with better financial outcomes will also show better CSR performances as they are able to invest more in CSR activities. The abovementioned sections of this thesis refer to CSR as voluntary activities that need additional financial investments. As a consequence, it is assumed that CEOs can invest and engage in organizational CSR initiatives if the organization provides internal financial resources in order to do so. In contrast, a small portion of researchers notes that organizations might also use external funding such as bank loans for CSR initiatives (Bihari and Prahadan, 2011) which underlines that financial institutions play an increasingly important role in today’s sustainable business practices (Bihari and Prahadan, 2011). Thereby, it questions the widely agreed upon notion that internal profitability is central in order to undertake CSR investment while providing an alternative to firms’ investment opportunities.
18 performance on the one hand, and uncertainty avoidance and CSR performance on the other hand.
Combining this with the hypotheses derived from Hofstede’s dimensions, it can be assumed that the relationship between the individual decisions of CEOs from individualistic, masculine and high power distance countries and CSR performance is strongly influenced by a company’s recent financial performance. When profits are low or even negative, CEOs of described countries are less likely to voluntarily invest in CSR practices and instead, will follow their usual business structures. Their general values do not initiate processes of CSR instantly. However, they can be assumed to engage in CSR activities if they are obviously able to afford and benefit from it, as for example through an improvement of the firm’s reputation. If the organization generates financial surpluses, these CEOs can undertake CSR initiatives without being forced to sacrifice aspects of their preferred business orientations and values. Therefore, one can say, the higher the financial performance, the higher the CSR engagements of the CEOs. As a result, the recent financial performance of a company is expected to affect the relationship between a CEO’s national culture regarding the dimensions power distance, individualism, and masculinity and CSR performance. More specifically, the negative relationships are assumed to be weakened if recent financial performance is high.
The resulting hypotheses are the following:
H5a: The negative relationship between a CEO’s national culture in terms of power distance and CSR performance is negatively moderated by the organization’s recent financial performance.
H5b: The negative relationship between a CEO’s national culture in terms of individualism and CSR performance is negatively moderated by the organization’s recent financial performance.
H5c: The negative relationship between a CEO’s national culture in terms of masculinity and CSR performance is negatively moderated by the organization’s recent financial performance.
19 incorporation of sustainable, social, ethical, and environmental practices regulated by various codes of conduct presents itself as a standard business pattern. Therefore, such CEOs are expected to invest in a rather stable pattern even if profits are low or negative. Instead of cutting CSR expenditures, they are more likely to consider taking out a loan in order to finance sustainable activities that present a core part of their business. Thus, uncertainty avoiding CEOs seem to be less influenced by a company’s recent financial performance than CEOs from masculine, individualistic and high power distance countries as CSR investments are an integral part of their operations. Assuming this, the recently common finding that financial performance has a positive impact on CSR performance is put into question. Therefore, the moderating role, in this case, is assumed to be non-significant.
H5d: The positive relationship between a CEO’s national culture in terms of uncertainty avoidance and CSR performance is not moderated by the organization’s recent financial performance.
This approach is relatively uncommon as usually, hypotheses are stated in an alternative manner. Hypothesis 5d presents a Null-Hypothesis predicting no impact of the moderator on the outcome variable. Traditionally, researchers make use of the alternative approach due to the belief that testing an expected non-existing effect would not enrich existing findings (Sommer, 2006). Nevertheless, refuting findings or theories can be extremely important in order to test them precisely which makes the incorporation of Null-Hypotheses necessary (McBride, 2010). Gallistel (2009) adds that Null-Hypotheses are as important as hypotheses that predict an effect and emphasizes their theoretical importance. On the contrary, Greenwald (1975) argues that accepting a Null-Hypothesis is related to a bias resulting from the fact that the majority of researchers reject the Null-Hypotheses and hence, does not support the approach of testing them. Similarly, Anderson, Burnham and Thompson (2000) suggest reducing the use of testing Null-Hypothesis. However, given the theoretical background leading to Hypothesis 5d, this thesis will test the null-effect as any other approach would falsely predict the impact of recent financial performance on the main relationship between uncertainty avoidance and CSR performance. It thereby challenges the recently common belief that recent financial performance has a positive influence on CSR performance.
20 financial performance is high; however, this moderating effect is assumed to be non-significant if the CEOs’ national culture is uncertainty avoiding.
2.6 Conceptual Model
The corresponding conceptual model presenting the abovementioned hypotheses is the following: H5a (-) H5d (0) H1 (-) H5b (-) H2 (+) H3 (-) H5c (-) H4 (-)
CEO national culture
21
3. Methodology
This section provides a detailed overview of the research methodology of this thesis. In particular, the variables and their corresponding measurement, the data collection strategy and the sampling strategy, as well as the methods regarding the statistical analysis are explained.
3.1 Variables and Measurement 3.1.1 Dependent Variable
Firstly, the dependent variable presents the CSR performance of the company. As mentioned above, CSR performance refers to the actual implementation of CSR activities as evaluated by external organizations.
There are multiple opportunities to measure CSR performance of companies. Several indexes exist such as the KLD index, the GRI index, the Fortune rating, as well as the CSRHub index; however, many of these measurement tools have particular disadvantages. The KLD index is one of the most popular CSR indexes and evaluates a company’s CSR performance by assessing activities on eight equally weighted dimensions (e. g. community relations, employee relations, and environment) (Griffin and Mahon, 1997). However, a major weakness refers to the availability of data, as information is mainly limited to American companies. Secondly, the Global Reporting Initiative (GRI) index provides an assessment that is based on sustainability reports of the organizations themselves. Therefore, it is based on the subjective ‘talk’ of companies, which is not an appropriate measure for this thesis. Thirdly, there are several Fortune Ratings. Although they are regularly used to evaluate firms’ CSR performances, these ratings (e. g. the Fortune Reputation Rating) especially refer to a company’s image rather than to its real CSR activities (Griffin and Mahon, 1997). Nevertheless, besides the difficulties mentioned before, each measurement tool also provides its specific advantages. Therefore, in order to make an appropriate evaluation, there is a strong need for including various sources of CSR performances (Carroll, 1994).
22 critical position. The overall company ratings are based on four different categories, namely on the firm’s performance regarding:
- Community (supply chain, human rights, etc.) - Employees (labor rights, health, safety etc.)
- Environment (environmental policy, waste management, energy use etc.) - Governance (leadership ethics, stakeholder treatment, transparency, etc.)
The overall scores is measured from 1 to 100, where 100 refers to the highest CSR performance and one to the lowest/worst performance.
3.1.2 Independent Variable
To measure the independent variable, CEO’s national culture, four variables are created each representing one of Hofstede’s initial four dimensions. Until now, Hofstede’s approach to measure culture remains the most dominant and most popular categorization among researchers (Beugelsdijk, Kostova, Roth, 2017). Literature in the field of business and management extensively integrate his dimensions to explain leadership behavior (Schwartz, 2006). More specifically, this variable is measured by using four of Hofstede’s six dimensions, namely power distance, uncertainty avoidance, individualism, and masculinity. These dimensions present the original four categories Hofstede found in his IBM study (Hofstede, 2011). Although especially the dimension of individualism is of utmost interest in management literature, this thesis analyzes the impact of all four dimensions. As already explained in previous sections, this thesis complies with existing literature analyzing the effect of culture (e .g. Thanetsunthorn, 2015; Wang, Rieger, and Hens, 2017) to obtain more complete results. The extended dimensions, long-term orientation and indulgence vs. restraint are neglected as researchers do not seem to extensively integrate them in their studies and they are based on a different dataset (see section 2.1). Each of these cultural scores is measured per country and also presents a range from 1 to 100, where 100 presents the highest level of power distance, individualism, uncertainty avoidance, and masculinity in the respective country.
3.1.3 Moderator
23 most widely used measures for corporate financial performance are return on assets (ROA) and return on equity (ROE) (e. g. Griffin and Mahon, 1997; Vitezik et al., 2012). Whereas ROA reflects an organization’s operating performance by presenting the firm’s asset utilization, ROE specifically refers to a firm’s financial performance and is seen as a very popular measure (Lindenberg and Ross, 1981; Chakravarthy, 1986). More specifically, it measures a company’s profitability by dividing the net income generated by the amount of shareholders equity. As this thesis analyzes the CSR performance of public companies, the use of ROE provides an appropriate measurement of financial performance. Using this ratio contributes to the understanding whether managers are able to satisfy the often contrasting interests of shareholders and stakeholders as elaborated by the stakeholder theory in section 2 and whether they regard CSR expenditures as additional costs that reduce shareholder profits (section 2.5) . Therefore, this thesis measures recent financial performance by a company’s ROE. To encounter the influence at different levels of ROE, the effect of low, medium, and high ROE is analyzed in order to investigate this ratio’s impact on the main relationships in more detail.
It is important to consider that the CSRHub scores are published 2017 and companies that invested in CSR operations are likely to have used profits of the previous year(s). According to CSRHub (2016), it takes approximately two years to reflect a change in a company’s CSR performance in its scores. Therefore, the time lag needs to be incorporated and consequently, the ROE ratios of 2015 are taken in order to be able to obtain adequate results.
3.1.4 Control Variables
Several factors are mentioned in previous studies that are expected to influence CSR performance of organizations. In order to ensure a high accuracy of this thesis, the impacts of these factors need to be excluded by controlling for their effect.
24 chosen by the Forbes 2000 list, presenting the largest firms in terms of sales, assets, market value, and profits. Nevertheless, differences in their size are still prevalent. Camacho (1991, p. 137) defines organizational size as ‘the number of individuals participating in the activity of the firm’. Researchers that focus on the topic of CSR define size in terms of employees (Baumann-Pauly et al., 2013; Wickert et al., 2016). Additionally, in order to not control for impacts that present determinants for the selection of the sample firms, this thesis complies with former literature in this field and uses the number of employees as a proxy for organizational size.
Secondly, the industry in which a company operates can have an impact on its CSR performance. Several industries needed to engage in CSR implementations relatively early, such as the chemical industry (Reniers, Sörensen, Vrancken, 2013), or face higher pressures, for instance, by NGOs to act in an environmentally friendly and socially acceptable behavior. To exclude this impact, this study controls for the influence of the industry in which the sample firms operate. In particular, the industry effect is excluded by creating a categorical variable, which indicates the industry affiliation of each observation. Thereby, each industry is automatically transformed into a specific number and it can be statistically analyzed whether industry affiliation has a significant impact, which will be controlled for.
3.2 Data Collection
First of all, the CSRHub scores for the sample companies can be found on the corresponding website. They can be sorted by countries or industry. For instance, by being able to choose a specific industry, the scores of the companies they rated can be obtained.
The scores of Hofstede’s dimensions can also be found in online databases, such as on the website ‘Hofstede Insights’ (2017). It is created by a consulting company that specifically focuses on the cultural interaction that builds on the work of Hofstede (Hofstede Insights, 2017) and additionally offers the opportunity to compare national culture scores that were created by the researcher. By selecting a specific country, each dimensional score can be obtained.
25 Regarding the control variables, the firm size in terms of employees as well as the industry of the sample organizations can be found in the Forbes Global 2000 list of 2017, which is also used as a basis for the sample selection.
3.3 Sample
The sample used in this thesis consists of public companies operating in a variety of industries and countries. Analyzing public companies has the advantage that these organizations are required by law to publish their financial results which is essential for the data collection in this study. Including a variety of firms with regard to their headquarter location and main industry enhances the generalizability of results. Such companies are chosen by using the Forbes Global 2000 list of 2017 which contains the 2,000 largest public companies worldwide. Firms presented in this list are selected by their size which reflects a combination of sales, assets, market value, and profits (Forbes, 2017). As already explained in previous sections, especially large organizations incorporate CSR activities as they arouse high public attention and face increased public pressure (Cowen, Ferreri, and Parker, 1987), making this list especially suitable.
With respect to sample size, multiple studies propose to use ten observations per independent variable (e. g. Altman, Vergouwe, Royston, and Moons, 2009; Laupacis and Sekar, 1997; Concato, Peduzzi, Holford, and Feinstein, 1995). Accordingly, as the analysis incorporates four predictor variables (Power Distance, Uncertainty Avoidance, Individualism, and
Masculinity), two control variables (Industry and Size), and one moderator (Recent Financial Performance), ten observations per variable would result in a sample of 70 observations.
26
3.4 Analysis
By using the software program SPSS, the statistical analysis can be conducted. More specifically, a hierarchical multiple regression analysis is conducted. Due to the fact that more than one predictor variable and also two control variables are part of this study, hierarchical regression presents an appropriate tool to test the relationships between CEOs’ national culture and CSR Performance. In order to control for industry and size effects, a statistical two-step procedure is needed which firstly generates a model presenting the effects of the control variables and in a second step shows the model including the predictors.
Several statistical assumptions for this analysis need to be tested and met in advance, as violations might affect and bias the results. First of all, significant outliers that could bias the model need to be detected to be sure that the outcome of the regression is valid. Secondly, the normality assumption is tested, as deviations from normality detrimentally weaken the validity of statistical outcomes (Hair et al., 1998). In addition, a linearity test between the independent variables and the dependent variable is conducted to ensure that the outcome variable is linearly related to any independent variables (Field, 2009). Fourthly, it is important that there is no multicollinearity between the predictors in order to observe the unique effects of the individual predictors. Furthermore, the values of the residuals need to be independent and the variance of the residuals constant because unequal variances, or heteroscedasticity, results in inconsistent estimates (Field, 2009).
28
4. Empirical Results
This section contains the empirical results of the statistical analysis. More specifically, it provides the outcomes of the assumption tests, the results of the hierarchical multiple regression analysis and the moderation analysis.
4.1 Assumptions of Multiple Regression Analysis 4.1.1 Outliers
First of all, the data is tested for potential outliers. As these unusual combinations of scores can severely bias the results, they need to be detected and adjusted if necessary. In order to do so, the Cook’s distance is used to determine these points. Thereby, a new variable is shown in the dataset. If the corresponding values of this variable are greater than 1, the presence of an outlier can be assumed. However, all values score clearly below 1.00, meaning that the dataset does not contain any outliers that might bias the results. This outcome is additionally supported by calculating the Mahalanobis Distance. This measure similarly detects any multivariate outliers by including a new variable in the dataset. Thresholds are different per number of predictors and sample size (Field, 2009). Therefore, I additionally created a variable showing the probability of each observation to be an outlier based on the Mahalanobis variable. In particular, observations that show a value smaller than 0.001 can be suggested to be unusual. Supporting the outcome of the Cook’s distance, no outliers were found by using this method.
4.1.2 Linearity
Despite visual options, such as a scatter plot, SPSS offers a statistical test for linearity. In particular, the linearity between each predictor variable and the dependent variable is tested and an ANOVA table including the significance for linearity and for the deviation from linearity is produced.
29 ANOVA Table Sum of Squares df Mean Square F Sig. CSR Performance * Power Distance Between Groups (Combined) 500,969 14 35,783 1,804 ,057 Linearity 150,868 1 150,868 7,605 ,008 Deviation from Linearity 350,101 13 26,931 1,357 ,205 Within Groups 1289,519 65 19,839 Total 1790,487 79
Table 1: ANOVA Table CSR Performance – Power Distance
As it can be seen in Table 1, the significance of linearity between Power Distance and CSR
Performance is < 0.05. Consequently, there is a linear relationship between the independent
and dependent variable. In addition, the significance of the deviation from linearity between these variables is > 0.05, supporting the assumption that a linear relationship exists. The same test is conducted for the remaining three independent variables.
Again, the statistical outcome presented in Table 2 (Appendix 1) supports a linear relationship. The significance of linearity between CSR Performance and Individualism is < 0.05, whereas the deviation from linearity is > 0.05. Therefore, a linear relationship between
CSR Performance and Individualism exists. Regarding CSR Performance and Masculinity
(Table 3, Appendix 1), the results also support a linear relationship, as the significance of linearity is < 0.05 and the deviation from linearity is > 0.05.
Table 4 (Appendix 1) shows different outcomes. The relationship between CSR Performance
and Uncertainty Avoidance does not show linearity, as the significance value is > 0.05. Although there is a linear component given by the significance value of the deviation from linearity (> 0.05), a general linear relationship cannot be observed. Therefore, the variable
Uncertainty Avoidance is transformed by taking the logarithm. Table 5 (Appendix 1) presents
the new values after transforming the independent variable. Although the values show a positive development in terms of the significance value, the value of linearity is still > 0.05, indicating a non-linear relationship and thereby violating the assumption of linearity regarding
Uncertainty Avoidance. Consequently, interpretations of the main results of the multiple
30 are violated (Field, 2009). Therefore, more valid results can be obtained with regard to this specific variable, even though one should still be careful with interpretations concerning this specific variable. Despite these problems and to be as precise as possible, two regressions will be conducted: one including Uncertainty Avoidance and one disregarding this variable. As the transformed variable show better values although violations are still obvious, the transformed variable UA_log will be used for the following tests.
4.1.3 Normality
In order to check for normality, a histogram is created. To fulfill the assumption of normally distributed values, the curve presented in the histogram needs to be symmetrically skewed and should not be extremely peaked or flat.
With regard to the skewness, the histogram (Figure 1, Appendix 2) presents a relatively normal distribution. However, regarding the kurtosis of the distribution, the curve seems to be peaked. Transforming the variables by using the logarithm and square root as suggested by Hair et al. (1998) and Field (2009) does not lead to the favored effect (Figure 2 & Figure 3, Appendix 3). As a consequence and as also suggested in the previous section, bootstrapping will be used in the main analysis to reduce and avoid this bias, as this method does not require normality (Field, 2009). Although one might additionally infer any outliers due to the gaps presented in the histogram, this is not the case as already tested by using the Cooks’ and Mahalanobis Distance (Section 4.1.1). Therefore, the values of the residuals seem to be normally distributed.
4.1.4 Multicollinearity
Multicollinearity refers to the correlation of the predictors. The independent variables should be observed to be independent in order to fulfill this assumption. To evaluate this condition, one can either use the correlation matrix of the predictors or use the Variance Inflation Factor (VIF). This thesis makes use of the latter, as it is said to be more precise and informative (Field, 2009). In order to infer non-collinearity, the corresponding value should not exceed 10 (Bowerman and O’Connell, 1990).
31
4.1.5 Independence of values of Residuals
The values of the residuals, or the individual data points, need to be uncorrelated to obtain valid results. The Durbin-Watson statistic serves as an appropriate tool in SPSS to test this assumption. The corresponding value of this statistic ranges from 0-4, whereas a value of 2.00 indicates that the residuals are perfectly uncorrelated (Field, 2009).
The Durbin-Watson statistic in Table 8 (Appendix 5) presents a value of 2.029. Therefore, the assumption of no correlation between residual values is perfectly met.
4.1.6 Heteroscedasticity
The final assumption that needs to be met is the homoscedasticity of the residuals, which describes the equal variation of them. The normal probability plot is an effective tool to test for heteroscedasticity (Hair et al., 1998). It plots the standardized residuals that the model predicts against the observed standardized residuals. For this assumption to be met, the variance of the values needs to be roughly similar (Field, 2009).
As it can be seen in Figure 4 (Appendix 6), the distribution in the scatter plot seems to be random rather than funneled. Therefore, it can be inferred that there is no heteroscedasticity and the final assumption is met.
4.2 Multiple Regression analysis
32
4.2.1 Hierarchical Multiple Regression of the predictor and outcome variables
CSR Performance (1) CSR Performance (2) CSR Performance (3) CSR Performance (4) Industry_Rec -0.047 -0.014 -0.047 0.014
Size -3.889E-6* -3.631E-6* -3,889E-6* -3,729E-6*
Power Distance (PD) -0.152** -0.157** Individualism (ID) -0.052 -0.055 Masculinity (MAS) -0.119** -0.120** Uncertainty Avoidance (US_log) 2.885 R² 0.054 0.255 0.054 0.247 R² change 0.054 0.201 0.054 0.193 F-value 2.182 4.164** 2,182 4,849** N 80 80 80 80
Table 9: Results of Multiple Regression analyses
*p<0.05 **p<0.01
Table 9 presents all main results for the multiple regressions. The first two columns show the
outcomes of the first multiple regression analysis which includes Uncertainty Avoidance, whereas the last two columns refer to the second analysis which excludes this predictor. There are two models per regression: one referring only to the control variables Industry and Size (CSR Performance (1) & CSR Performance (3)) and one model adding the predictor variables (CSR Performance (2) & CSR Performance (4)).
33 one. By adding the four predictor variables in model two, R² increases to 25.50%, meaning that these independent variables add a significant effect to the explanation in the variation. Higher values of R² generally better explain the variance of the dependent variable. The value of ‘R² Change’ indicates how much of this variation is explained by the predictors
Uncertainty Avoidance, Masculinity, Individualism, and Power Distance only. Correspondingly, the isolated effect of them in explaining the variance of CSR Performance in model two is 20.10%. Comparing the adjusted R² with R² in models one and two, there is a difference of 0,061 between these values which indicates how well this model generalizes (Field, 2009). In this case, if the model would have been taken from the population rather than a sample, it would account for 6.10% less variance in the outcome.
In order to test a potentially changing effect if the independent variable Uncertainty
Avoidance is excluded due to the violations mentioned above, a second Multiple Regression
Analysis is conducted. Referring to model four, by excluding this predictor, R² declines which negatively affects the overall model explanation. In addition, the main results of the two multiple regression analyses that will further be examined in the following paragraphs remain the same. As a consequence, Uncertainty Avoidance will further be incorporated in the analysis in order to be able to test all hypothesis this thesis proposes. Thus, this thesis will focus on the outcomes of the first multiple regression analysis.
Table 9 also shows whether the model is significantly better at predicting the result rather than
simply using the mean (Field, 2009), or more specifically, it explains whether the model is a significant fit of the overall data. By having a look at the F-value, one can infer that model one does not significantly improve the ability to predict CSR Performance compared to not fitting the model (p>0.05). In contrast, model two significantly enhanced this ability, as the corresponding F-value > 1 and also highly significant (p=0,001). Therefore, the second model presents a statistically significant predictor of CSR Performance.
34 affects CSR Performance (b=2,885); however, as with Individualism, this effect is not statistically significant (p>0.05).
4.2.2 Moderation Analysis
In this section moderation effect of Recent Financial Performance on the relationship between the four independent variables and CSR Performance is examined. The following table provides an overview of the individual interaction effects of three different regressions, namely when Recent Financial Performance is low (-18.5091), when it is at the mean value, and when the value of Recent Financial Performance is high (18.5091).
Recent Financial Performance x PD
Low levels of RFP -,0677
Medium levels of RFP -,0807*
High levels of RFP -,0937
Recent Financial Performance x ID
Low levels of RFP ,1008*
Medium levels of RFP ,0263
High levels of RFP -,0483
Recent Financial Performance x MAS
Low levels of RFP -,0997
Medium levels of RFP -,0731
High levels of RFP -,0464
Recent Financial Performance x UA_log
Low levels of RFP -5,8669
Medium levels of RFP 8,7643*
High levels of RFP 23,3955
Table 10: Multiple Regression and Moderation Analysis outcomes
35 To start with the predictor Power Distance, the outcome shows that when Recent Financial
Performance is high, there is a non-significant negative relationship (p>0.05) between Power Distance and CSR Performance (b=-0.0677). However, at the mean value of the moderator, a
significant negative relationship can be observed (b=-0.0807; p<0,05), while a non-significant negative relationship can be inferred at low values of Financial Performance (b=-0,0937; p>0,05). This outcome shows that the relationship between Power Distance and CSR
Performance only becomes obvious when organizations show an average Financial
Performance. In addition, the effect predicted in H5a is not supported. The negative relationship between the predictor and outcome variable is not weakened when financial performance outcomes are high which can be also seen in the conditional effects at different values of Recent Financial Performance (Appendix 7).
Secondly, the moderation effect of Recent Financial Performance on Individualism and CSR
Performance is analyzed. Given the outcome above, the relationship between Individualism
and CSR Performance only becomes obvious at low values of the moderator (b=0.1008; p<0.05). One can argue that when Recent Financial Performance is low, the negative relationship between Individualism and CSR Performance is significantly strengthened. At high levels of financial performance outcome, the negative effect of the predictor on the outcome variable is weakened; however, this statement is not statistically significant (p>0.05). This development can additionally be observed in Appendix 8.
Next, the same test is conducted for the relationship between Masculinity and CSR Performance. By interpreting the values above, the moderation effect of financial outcomes in terms of ROE in % is not significant at all three levels (p>0.05). Therefore, there is no significant interaction of Recent Financial Performance on the relationship between
Masculinity and CSR Performance, even though the weakened effect at high levels of
financial performance as predicted in H5c is given. Again, a more precise overview is attached in Appendix 9.
Lastly, the moderation effect on the relationship between Uncertainty Avoidance and CSR
Performance is examined. As with the predictor Masculinity, no significant interaction can be
concluded. For all three regressions, the significant threshold of 0.05 is exceeded, supporting hypothesis H5d. More specifically, the relationship between Uncertainty Avoidance and CSR
Performance is not moderated by Recent Financial Performance. There is no additional
36
5. Conclusion
The primary objective of this study was to examine whether there is a relationship between the national culture of CEOs and CSR Performance. In particular, the effect of four of Hofstede’s cultural dimensions was tested. In addition, the moderating effect of Recent Financial Performance on the main relationship was analyzed in this thesis. This chapter aims to discuss the empirical results of the previous section. It additionally provides theoretical and as well as managerial implications and reveals limitations of this thesis in connection with potential directions for further research.
5.1 Discussion of Results
Firstly, Hypothesis 1 was found to be supported by showing a significant negative relationship between Power Distance and CSR Performance according to the predictions. As presumed, CEOs whose national culture is dominated by high levels of Power Distance do not seem to intensively engage in CSR activities that positively affect the CSR Performance. This outcome underpins the notion of Blodgett, Lu, Rose, and Vitell (2001) who conclude that higher levels of Power Distance are related to a higher self-concern, rather than being related to ethical sensitivity. Additionally, the findings contradict the argument made by Waldman et al. (2006) that managers might feel obliged to focus on various stakeholders’ needs and hence on CSR activities due to their high and powerful position within the society. In contrast, results show and underline the conclusion of Usunier, Furrer & Furrer-Perrinjaquet, (2011) that CSR and cultures characterized by high power distance are less compatible as the managers’ main focus is to maximize shareholder value. Therefore, they do not feel obliged to benefit various stakeholders but aim to increase the profits for shareholders.
37 approach which might inhibit the real implementation of such actions due to expectedly effortful and time-consuming decision-making processes. Accordingly, individualistic CEOs see social elements as an instrument to reach higher economic goals (Usunier et al., 2011), which might motivate them to purposefully implement such actions. A second explanation might be that individualistic CEOs engage in more explicit voluntary actions, such as donations, while collectivistic managers tend to engage in a more implicit way (Matten & Moon, 2008). Correspondingly, more explicit CSR activities are more likely to be included in CSR performance ratings by external organizations as they are easier to identify.
Thirdly, the negative relationship between Masculinity and CSR Performance is tested to be significant which supports Hypothesis 3. It shows that strong masculine traits of CEOs’ national culture negatively affect the CSR Performance of organizations. This outcome is in line with the statement of Blodgett, Lu, Rose, and Vitell (2001) that CEOs from highly masculine cultures focus mainly on themselves rather than on the interests of the organization’s stakeholders. Additionally, it supports multiple arguments by other researchers such as Tice and Baumeister (1985) who state that masculine values restrain helping behavior. Although researchers do not fully agree on the effect of Hofstede’s cultural dimensions on social and ethical behavior, they seem to do so with regard to masculinity. Masculine values are commonly negatively associated with social and ethical actions (e. g. Scholtens & Dam, 2007; Steensma, Marino & Weaver, 2000; Blodgett et al., 2001; Thanetsunthorn, 2015; Adnan, van Staden & Hay, 2009).
38 Lastly, this thesis tested the moderation effect of Recent Financial Performance on the relationships mentioned above. To start with, outcomes show that for the relationship between Power Distance and CSR Performance, the moderating role is only significant for a very small portion of the moderator. More specifically, a negative impact which weakens the main relationship can only be observed for low to medium values of Recent Financial Performance, meaning that high financial performance outcomes do not influence the main relationship and consequently, the prediction made in Hypothesis 5a is not proved to be valid.
Additionally, a moderating role of Recent Financial Performance was partly determined with regard to Individualism and CSR Performance. In particular, a significantly positive effect was found at low levels of Recent Financial Performance, meaning that the negative impact of Individualism on CSR Performance is strengthened when Recent Financial Performance is low. This partly supports Hypothesis 5b by implying that individualistic CEOs would not invest if recent financial performance outcomes are low.
Finally, the moderating role on the relationships between the predictors Masculinity and Uncertainty Avoidance and the outcome CSR Performance is tested to be non-significant at all levels of Recent Financial Performance. Accordingly, recent financial performance outcomes do not impact the relationship between these two predictors and CSR Performance. As a consequence, Hypothesis 5c is not supported, whereas the prediction made in Hypothesis 5d is supported.
39 To summarize, culture does seem to influence CSR Performance; however, especially values of high power distance and masculinity seem to have a critical impact. Opposed to this, the impact of financial performance on the former relationship can be interpreted to be rather small. At this point, it is important to mention that although a robustness method in form of bootstrapping was used in order to reduce bias, the results need to be interpreted carefully given the violation of linearity between Uncertainty Avoidance and CSR Performance. Results might be different if the assumption of linearity would have been fully met.
5.2 Theoretical and Managerial Implications
This thesis provides several theoretical implications that support the significance of this research. First of all, it presents a valuable extension of prior literature by taking a first step to fill the gap of individual-level effects on CSR performance. Most researchers focused on national- or organizational-level impacts on CSR performance (e. g. Thanetunthorn, 2015), whereas this thesis shows that also individual CEOs have an impact on the firms’ practices and on CSR Performance in particular. Moreover and as explained in section 3.3, by creating a sample of organizations and CEOs whose national culture is identical to the culture prevalent in the corporate headquarters’ nation and excluding those where this condition is not met, the results are more specific and reliable. Thirdly, this thesis provides support for the upper echelon theory by Hambrick and Mason (1984) as it found significant relationships between individual values of CEOs and corresponding business practices. In addition, it does not only widen the overall understanding of why CSR practices and performances differ between organizations, but also between countries.
