The Mediating Effect of Reputation and Innovation on the Relationship Between Board Diversity and Firm Performance in Fortune 500 Companies
Business & Economics Master Thesis (15 ECT)
Martina D. van Wieren 10598588
15 July 2018
Word Count: 12,245 Supervisor: Thomas Buser
MSc Business Economics, specialization Managerial Economics & Strategy Faculty of Economics and Business, University of Amsterdam
Statement of Originality
This document is written by Martina D. van Wieren who declares to take full responsibility for the content of this document.
I declare that the text and the work presented in this document are 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
Abstract
The majority of existing literature focuses on analyzing the direct relationship between board diversity (gender & racial) and firm performance. This paper gives insights into a deeper layer of analysis within this topic by analyzing two mediators, innovation and reputation, and their suggested association to board diversity and firm performance on a sample of the Fortune 500 firms. It will expand academic literature in understanding the relationship between board diversity and firm performance at a deeper level. The research question is analyzed using both graphical results and panel data regression analysis that interprets the mediating effect of innovation and reputation on the direct relationship. Not all findings verify the argued hypotheses. Gender diversity in top management is positively and significantly related to firm performance as expected. However similar results are not found for racial diversity. Furthermore, neither innovation and reputation mediate this relationship.
Table of Contents
1. Introduction ... 5
2. Theoretical Framework ... 6
2.1 Board Diversity and Financial Performance ... 6
2.1.1 Upper Echelon Theory ... 7
2.1.2 Agency theory ... 8
2.1.3 Human capital theory ... 9
2.1.4 Resource dependency theory ... 9
2.1.5 Social psychological theory ... 10
2.2 Board diversity and Innovation ... 13
2.3 Board diversity and Reputation ... 15
2.4 Mediation of Innovation and Reputation ... 18
2.5 The Conceptual Model ... 19
3. Methodology ... 19
3.1 Sample... 20
3.2 Independent variables: Gender and Racial Board Diversity ... 20
3.3 Dependent variables: Corporate Financial Performance ... 21
3.4 Mediators: Innovation and Reputation... 22
3.5 Control Variables ... 23
3.6 Statistical analysis ... 24
4. Empirical Results ... 26
4.1 Summary Descriptive Statistics ... 26
4.2 Correlation Matrix ... 27
4.3 Graphical Results ... 30
4.4 Regression Results ... 32
5. Discussion & Conclusion ... 38
“Companies have increasingly begun to regard inclusion and diversity as a source of competitive advantage, and specifically as a key enabler of growth.” (McKinsey, 2018)
1. Introduction
Diversity in firms is a popular topic within and between firms and employees and the media over the last years. Diversity in the workforce has been a discussion topic for over 50 years as research goes back further than this. Yet, it was not until 2014 when the first ever diversity report was published by Google about their employee’s diversity statistics (Gibbs, 2014). Since then, many firms have followed Google’s footsteps and it has now become the norm to publish a yearly diversity report. These reports often contain various statistics and explanations about many different aspects of diversity. The McKinsey diversity report shows that females and racial minorities continue to improve their presence within top management of their firms. The 2018 report has increased their sample size to 1,000 companies, compared to the 400 in the 2015 report, across 12 countries compared to their first diversity report and concluded a positive correlation for both cultural and gender diversity with firm performance (Hunt et al., 2018). This constant attention for diversity in the workplace has sparked much interest in scholars over the decades and has inspired a lot of existing literature. The majority focus on the effect this diversity measurement in top management could have on a firm’s success and what strategical implications it could have.
The majority of the research that has been conducted is mainly focused on the direct relationship between board diversity and firm performance. Some researchers find a positive relationship while other find a negative or a neutral relationship (Carter, et al., 2003;
Dimovski & Brooks, 2006; 2012; Adams & Ferreira, 2009). Within this research topic there is a disperse number of conclusions that have been presented on the relationship between the two. Little research has tried to dig deeper into trying to find out what impacts this
relationship. Therefore, this paper aims to delve one layer deeper and try to understand and research whether there is a mediating effect on the direct relationship, using innovation and reputation as proposed mediators. Why specifically these two variables? It has been shown in prior research that these two are important factors to take into account when explaining the success or failure of firm performance (Roberts & Dowling, 2002; Hall, 1993; Zahra & Garvis, 2000). Reputation will be interpreted as the publics collective opinions and
judgements of firms’ actions over time. While innovation will be interpreted as the interaction between people and groups to come up with new and creative ideas.
Therefore, the main research questions this empirical study will answer is:
To what extent do reputation and innovation mediate the relationship of gender and racial board diversity and financial performance of the Fortune 500 companies?
Up until now, the research completed has focused on a superficial level. The
concluding results could have great impact on decisions made by the firms in terms of hiring and promotions. They will learn to have a broader view on the people available for the position allowing them to have a higher chance of success. Furthermore, it could create useful and insightful information when making strategical decisions within the corporate processes.
Giving a quick overview of the results. A significant positive relationship was found between gender diversity and firm performance. While a negative and insignificant
relationship was found for racial diversity. The mediators, innovation and reputation, do not present to be significantly statistical mediator for the relationship between board diversity and firm performance.
The paper is structured as follows. In Section 2, theoretical and empirical evidence of previous research is summarized to present a strong foundation for the development of the hypotheses. Focusing firstly on the main relationship between board diversity and firm performance, followed by the explanation of innovation and reputation and their role within the relationship between diversity and firm performance. Section 3 presents the methodology of the empirical research. This includes the explanation of the variables and the statistical analysis that are conducted. Section 4, explains the results obtained from the statistical analysis. Section 5, provides an overview of the main results, limitations and possible future research to consider within this topic.
2. Theoretical Framework
2.1 Board Diversity and Financial Performance
Currently, there already exists a vast amount of research concerning the relationship between board diversity and firm performance. However, even with such a huge amount of literature and research there is still no clear consensus about what can be concluded from this
relationship. There is ambiguity between the results of various papers that have been published over the years. A clear overview of the existing empirical evidence will be discussed after the theoretical overview.
Economic reason is one of the major arguments given for the importance of diversity and firm performance in the McKinsey diversity report (Hunt, et al., 2018). This is further explained with specific theories in the following sections. Besides looking at the direct relationship, other variables will be considered that could impact the relationship and build on current literature by explaining the possible mediators and their effect on the relationship.
Carter et al. (2010) explains, “being part of the board entails having certain duties you need to fulfill: the monitoring and controlling of managers, providing information and
counsel to managers, monitoring the compliance with new laws and linking the corporation to the external environment,” (p. 6). There are certain theories that support the prediction that board composition is related to the success of the firm. No single theory is believed to explain the nature of the relationship between board diversity and firm value but several theories provide insight to certain aspects of the relationship and the linkage to the duties required to be fulfilled by board members. Therefore, five important theories are assessed from various research disciplines: Organizational Theory, Economics and Social Psychology. A well-developed theoretical argumentation together with an overview of existing empirical evidence will is given to provide a strong foundation for the development that are tested.
2.1.1 Upper Echelon Theory
To fully comprehend the decisions that are made by organizations or why organizations perform the way they do, it is important to understand to understand who reaches these decisions. Top executives are responsible for making crucial decisions and these decisions can be influenced by their opinions. The main idea of the theory is that executives act based on their personalized interpretations when making strategic decisions within the board. These personalized interpretations are formed and based on a director’s experience, values, and personalities (Hambrick, 2007).
Leadership is a very complex web in organizations and is not a solo activity but a shared activity within the organizations with executive boards at the highest level. The collective combination of every one of these individuals’ capabilities, knowledge and
board of directors (Hambrick, 2007). The use of demographic characteristics of executives has been validated through previous research. Validating these characteristics allows demographics to be used as a proxy for members cognitive frame. Researchers can thus use this reliable information on executive’s background for predictive value for strategic decision making (Hambrick 2007).
An important note to mention, through the use of only demographic characteristics we lose the aspect of social processes and psychological behavior that influences an executive decision making. However, even though this is the case, substantial evidence has been provided by researchers over the years about the demographics profiles of various executives and their relations to performance outcomes of firms (Hamilton, 2000; Eisenhardt &
Schoonhoven, 1990; Boeker, 1997). These characteristics are therefore important aspects to take into consideration in analysis on board member decisions.
2.1.2 Agency theory
The agency theory is often used by researchers to understand the link between characteristics of a board and the value this can create for a firm within the financial and economic sectors (Carter et al., 2003). Carter et al. (2003) evaluates whether gender and racial diversity could possibly enhance the mechanism of control over managers. Concluding that corporate governance is critically important for the relationship between board diversity and firm value.
One of the roles of board members is to ensure they resolve issues between managers and shareholders. Their level of independence is of great value for this to work at the most optimal level (Fama & Jensen, 1983). The theory suggests the importance of independent oversight needed from board members. The diversity of the board could serve to be one of several tools to minimize potential agency issues (Erhardt, et al., 2003). The level of independence is key, with the goal of avoiding collusion between outside and inside directors (Carter, et al., 2003). Diversity allows for a higher probability of independence because of the different backgrounds of board members, including age, ethnicity, gender (Campbell & Minguez-Vera, 2007). Directors with various backgrounds might ask different questions or resolve issues differently compared to a homogeneous board composition. Diverse boards may have a different view on how to handle situations due to their difference in experiences. However, a different perspective does not automatically indicate that it will also be a more
effective way to handle situations. Most of the research states that the more diversity the greater the chance of independent thinking when it comes to such relational issues (Campbell & Minguez-Vera, 2007; Fama & Jensen, 1983).
2.1.3 Human capital theory
Human capital theory is derived from the research by Becker (1964) which researches the role of a person’s demographic characteristics that can be beneficial for an organization. Diversity creates a pool of various unique human capitals combined in one board. Terjersen et al. (2009) researched this based on gender diversity. They concluded that it can improve corporate governance and increase company value by pooling together a group of diverse people to use it to the firm’s maximum ability. It is logical that ethnic diversity measure could have similar effect of creating unique human capital as gender diversity has shown to create as it creates for greater variety in human capital.
Human capital can be a key resource, that can create a competitive advantage for a firm (Terjesen, et al., 2009). Creating a diverse board can lead to a diverse base of
perspective that could improve problem solving skills. Creating the ability to manage diversity within a firm can stimulate firms to incorporate diverse perspectives into
organizational decision-making processes (Shrader, et al., 1997). Having a diverse employee base increases the probability to exploit the use of external resources and sources
(Ostergaard, et al., 2011). Intangible resources can prove to be just as valuable as tangible resources. In this case, specifically the human capital resources on which firms heavily rely on, can be an important factor for firm performance. A firm’s human capital is affected by diversity, as it determines the composition of groups and how employees interact with one another (Lauren, et al., 2005). Therefore, diversity is key to understand the human capital base of firms. Diversity of human capital is often measured through the use of individuals demographics as a proxy to further our understanding of different attitudes and cognitive decisions (Harrison and Klein, 2007).
2.1.4 Resource dependency theory
Another important role of boards is to serve as a link between the firm and other external parties to ensure all environmental dependencies are met (Pfeffer and Salancik, 1978). In other words, the board is required to ensure they can obtain all necessary resources
needed to be fully functioning. The board is in charge of ensuring the provision and access to these resources (Hillman and Dalziel, 2003).
Pfeffer and Salancik (1978) explain in more detail the four primary resource benefits which should be provided by the boards. (1) Advice and council, (2) legitimacy, (3) channels of communication between the firm and external parties and (4) preferential access to
commitments. These responsibilities have a possible direct effect on the performance of the firm. The less dependent the firm is on others to provide resources, the more it will diminish uncertainty, lower transaction costs and in the long run with long-term survival of the firm (Carter et al., 2010). Uncertainty can cloud organizational control and choices of strategies which could obstruct the daily functioning of a firm. In other words, if a board can build a relationship with more ease with external parties they could reduce the uncertainty and thus help increase firm value through ensuring regular functioning teams (Hillman et al., 2000).
As mentioned, an important role of board members is to connect and build
relationship with external parties to support the needs of the firm. The more diverse the board composition the more likely they will be successful in developing relationship between external parties and the firm. Developing these relationships will reduce the uncertainty and lower transaction costs for the firm in general. Leading to belief that a diverse board will lead to more successful firm performance.
2.1.5 Social psychological theory
The social psychological theory predict that diverse board members have a minimum amount of influence on board decisions. Westphal and Milton (2000) address a contradicting view that in the opinion of corporate stakeholders, diverse minorities (racial and gender related) demographics within a board have a positive impact on firms. However, when looking at the topic from an academic literature point of view it covers a more negative view on how minorities can influence decision making within groups. Arguing that socially diverse group of people can decrease the cohesion between groups. The authors argue that this
conclusion is obtained from the social impact theory. It predicts that, “all else equal,
individuals who have a majority status on a salient attitude, belief, or social feature have the potential to exert a disproportionate amount of influence in decision making,” (Westphal & Milton, 2000, p. 368). They further conclude that those people who are in the majority are more likely to resist the influence on any type of minority when it comes to decision making.
This underlying resistance can negatively influence the intergroup relations of a board which is necessary for a cohesive group towards managers in lower levels of firms.
Taking this into account, looking at other research and theoretical argumentation I come to another conclusion about how diversity could affect the group composition. Diverse groups of people will generate more heterogeneous pool of opinions and more critical thinking when making decisions (Carter, et al., 2010). Creativity can be a result when considering more than one opinion about a decision-making topic. Leading to more creative thinking and resulting in a more effective decision-making process with a diverse group of people together in a team. Moreover, experimental evidence indicates that diversity in teams improves firm performance. Important to look at both field and lab experimental studies as experimental studies help improve endogeneity problem. However, there are limited in their real-world resemblance. Hoogendoorn et al. (2013) conducted a field experiment in which they manipulated the gender composition of teams. They concluded that teams with equal gender diversity have higher sales and profits compared to male dominated teams.
Furthermore, Woolley, et al. (2010) concluded from their experiment that teams composed of both genders performed better because they their average level of social sensitivity to other group members is higher than male teams. Sommers, et al. (2008) conducted an experiment on white and heterogeneous groups and found that in the heterogeneous groups there are more novel contributions. These novel contributions could contribute to greater firm value.
2.1.6 Empirical Evidence
This section provides an overview of empirical evidence on gender and racial diversity on firm performance. Many researchers have looked at the relationship between diversity and firm performance. Most researchers focus specifically on a single country and either gender diversity or racial diversity. Some researchers find a positive effect between diversity and firm performance. As demonstrated by Joecks, et al. (2012), who focused on German companies and their top management firms. Initially they conclude that there is a negative correlation between gender diversity and firm performance. However, as Joecks, et al. (2012) calls it, once a ‘critical mass’ of 30% has been reached, gender diversity has a positive relationship on German firm performance. These results are replicated in the research conducted by Mahadeo, et al. (2012), who conducted a similar analysis. The only difference is, they looked at companies in an emerging country. They find similar evidence as Joecks et al. (2012), with a positive correlation between the diversity and firm performance.
Erhardt et al. (2003) and Carter (2003) focus on American companies when conducted their empirical analysis. Similar to the other authors, they find a positive association between gender diversity and firm performance. Carter (2003) demonstrates a positive relationship between diversity and firm performance however in his 2010 paper he is not able to replicate these results and found a negative relationship between gender diversity and firm
performance.
However, some researchers find a negative relationship. Such as Adams and Ferreira (2009), they study US firms and find that board with female directors have an average negative effect on firm value. They believe this is driven by fewer takeover defenses. These results were replicated by Bøhren and Staubo (2014) and Ahem and Dittmar (2012), who focus on a Norwegian sample. Norway was one of the first countries to implement a minimum female ratio on executive boards. However, this led to more inefficiencies and higher costs. Arguing the quota pushed firms to hire women that are less experienced or and too young resulting in the deterioration of operational performance.
Furthermore, researchers Dimovski and Brooks (2006) conclude that there is no relationship to be found between diversity and firm performance. Similarly, Matsa and Miller (2010) cannot conclude the direction of the relationship. They are unable to distinguish between a positive or negative effect on firm performance.
These mixed results demonstrate that there is still a lot of uncertainty within this topic. The majority of the research that has been completed over the last years has focused on the direct relationship between board diversity and firm performance, taking into account demographic variations such as age, tenure and educational background. With time, the focus has turned to gender and racial diversity with the increased awareness of diversity with the development of a global environment.
Some theories and empirical evidence hint at negative relations yet sufficient amount of theoretical evidence and empirical research shows that diversity could positively impact firm performance. Therefore, the following hypotheses follow:
Hypothesis 1a: There is a positive relationship between gender board diversity and firm performance.
Hypothesis 1b: There is a positive relationship between racial board diversity and firm performance.
However, there is more to this relationship than solely looking at the direct
relationship between diversity and performance. Little to no attention has been given to the topic of mediators that could affect the relationship between board diversity (gender and racial) and financial firm performance. What are possible factors that could affect this relationship? Following, a closer look is taken at innovation and reputation as factors that could affect the relationship between board diversity and firm performance.
2.2 Board diversity and Innovation
Innovation is often a result of the interaction between individuals within an
organization. In this context, different attitudes and cognitive decision can generate new ideas when pooling multiple different sources of individuals together. Innovation is simplistically said the profitable implementation of creative ideas (Hassan, 2016). Creativity does not mean innovation yet innovation cannot exist without creativity. They are mutually dependent. Innovation strategies are opportunities that lead to new or developed product lines that could be profitable for firms. Besides the important creativity aspect of innovation, the successful implementation of the ideas is just as important. One of the most important reasons is that innovation and technological advances are an important aspect that firm have to consider as part of their competitive strategy (Rajapathirana, et al., 2017). Due to the rapid and
unpredictable changes in the market most firms often face very competitive challenges. Innovation is a key asset for firm to have in order to create new opportunities within a market.
Diversity plays an important role in predicting the innovative behavior of teams (Van der Vegt & Janssen, 2003). There are several reasons why demographics of a board of directors has influence on the levels of innovation of a firm. Firstly, innovation is often seen as an interactive process within a group of individuals to discuss, develop, modify and realize new ideas. Diverse teams have proven to increase creativity, a factor, as mentioned above, that is crucial for innovation to take place. As Bantel and Jackson (1989) conclude in their empirical study on a sample of 199 banks and their innovative capabilities with respect to demographic characteristics of the board of directors. They conclude that banks whose board is composed of more diverse backgrounds are more innovative in their strategies because the diverse board composition allowed for more creative thinking. As decisions, top management has to make, is influenced by past experiences and background/demographic characteristics (Hambrick & Jackson, 1984). Robinson and Dechant (1997), build on this research and
conclude that values and beliefs that people develop are affected by demographic
characteristics such as education, tenure, experience and age. These beliefs and values are not randomly distributed over the whole population but are actually developed through various demographic characteristics. Concluding these findings, the knowledge incorporated in the human capital of a firm can be considered a competitive advantage and allow for innovative opportunities to lead to competitive advantages for firms is due to demographic backgrounds, which includes the variety attitudes developments due to racial or gender diversity.
Secondly, when observing how a homogenous group acts, the similarities can lead to inferior decision making as it can bring on groupthink. Leading to restricted thinking in terms of alternatives. Filley, House, and Kerr (1976) conclude that routine problem solving can easily be done by a homogeneous group but that new and ambiguous problem solving is best handled within a heterogeneous group. This heterogeneous group composition allows for unrestricted alternative seeking due to various extends of knowledge, backgrounds & cultures as also mentioned above by Hambrick (2007). Adding onto this research, diversity in
experiences has also shown that it can improve innovation though generation of alternative solutions and innovations (Bear et al., 2010). It can produce high-quality innovative ideas as individuals use “critical and investigative interaction processes in which individuals can identify, extract and synthesize their different perspective” (Amason, 1996, p. 124).
The more diverse a boards’ composition the greater the potential of them
understanding complex problems allow for greater probability of successful problem solving that can address the business environment and be successful in furthering a firm’s objectives. Even though this had been concluded by various researchers, having a broad spectrum of opinions could also hinder the process of problem solving as there are too many aspects that have to be taken into account leading to higher probability of inefficacies occurring (Rajan, et al., 2000). On the other hand, diverse opinions of individuals allow for the pooling of
information and combine separate ideas to form brilliant innovative solutions to work-related problems and thus providing innovative performance benefits that outweigh the possible inefficacies that could arise.
Finally, looking at the social relationship of a diverse group of people is also an important argument for innovation intensity levels. The complex web of relational connection will increase with a more diverse group compared to a homogeneous group. The greater the span of contacts, external of the firm, can be very resourceful when it comes to acquiring knowledge and information (Davidsson & Honig, 2003; Burt, 1997). Burt (1997) goes on to argue that when having more ‘volume’ in your contacts you can reach much more
information indirectly than when compared to a homogeneous group with similar contacts. He uses a metaphor of two employees, James and Robert. James has a weak set of connection and Robert, who has a strong set of connections, takes over James’s job. In the process of doing so, he takes over James’s web of networks while adding his own into it, ending up with a bigger social network than before. Bringing this back to the topic of gender and racial diversity, the more variety there is in the relations of board members the more complex the relational web there is to be created to gain information, knowledge and referrals. Having more diverse group leads to more chances of findings new opportunities within the market. It allows the board to bring together a vast amount of information that would beforehand never have linked. These network connections will allow for the provision of advice and expertise that the firm might not have on its own. With the complex web of relational ties the board can now connect and spar with people in the community that will indirectly support a firm’s objectives (Beckman and Haunschild, 2002). Making this argument even more plausible Ibarra (1992 & 1993) concluded that in fact minorities and females have a more extensive network than those of the traditional white males in society. Allowing this web to be build much more complex and extensively compared to a homogenous, white male board. Besides keeping in contact with strong ties, they are also more likely to keep in touch with weak ties, which could be of equal value or even more value than strong ties within a business
environment. These can also build into the necessary knowledge needed to find these new innovative opportunities in the market.
Innovation is a creative and interactive process which needs diversity to promote the interactive creativeness of people’s knowledge pooled together with the complex social relational web of contacts. Diversity is a necessary ingredient for innovative success as it impacts the innovation process. Hence, diversity should have a positive relationship with innovation of a firm. Therefore, the following hypotheses will be tested.
Hypothesis 2a: There exists a positive correlation between gender board diversity and innovation.
Hypothesis 2b: There exists a positive correlation between racial board diversity and innovation.
2.3 Board diversity and Reputation
Corporate reputation is defined as “publics’ cumulative judgement of firms over time” (Fombrun and Shanley (1996), as cited in: Bear et al., 2010, p. 207). It is a perception of the
representation of a firms’ past actions and their future prospects that describes the publics cumulative judgement of firms over time. Research has demonstrated the benefits of having a good reputation (Black, et al., 2000). Some of the benefits include being able to attract a good pool of qualified employees, better stock market and meeting the needs of a global market. It is evident that due to informational asymmetry present in what the public knows and does not know about a firms’ actions and doings, it uses symbols and actions that are public knowledge to judge firms and come to a conclusion about a firm reputation (Brammer et al., 2009). In this empirical research, besides investigating innovation, it will also analyze how board diversity can have reputational impact on companies. Using the signaling theory to further explain the arguments.
Signaling theory identifies that the presence of diversity in a company shows a positive signal to the public and other stakeholders (Certo, 2003). Most importantly, having a diverse board composition aids the image that forms the external perception of the
organizational effectiveness of firms as females influence the perception of external stakeholders or they have a certain skill set that improve board processes (Brammer et al., 2009). Research has shown that increased diversity on boards had improved the workforce motivation and loyalty (Powel, 1999). Exploring this diversity further, Powell (1999)
concluded that female directors give much needed mentoring for young and ambitious lower level women on the job, to allow them to reach their career goals with a higher probability. Furthering this evidence, women have an important role when it comes to tailoring products to really align them with the needs of the consumers (Daily et al., 1999). Firms who put women on the board to close the gap of competence between firm culture and consumer-oriented strategy. Singh and Vinnicombe (2004) conclude that diversity increases there where a specific set of skills is needed.
Further developing the needs of a firm’s consumer base, females and minorities on boards have also lead to more CSR related activities that firms partake in. CSR has been shown to be a mediating variable for diversity and reputation (Bear, et al., 2010). Brammer and Milington (2005) support this research by concluding from their research of UK firms that those with more active philanthropic engagement will shape a better perspective in the stakeholder eyes, and thus increases the positive image build around the company.
Minorities also play a greater role in creating a strong governance mechanism, especially when focusing on promoting equality, fairness and social responsibility (Cook and Glass, 2015). For example, firms signal that a firm follows the norms and positive working
2000). All of these arguments mentioned above coincide with an important signal towards the public, creating their external perception of firms, and thus create a reputation score of the firm.
Diversity can play an important role in forming the boards human, cultural and social capital and thus also play a vital role in shaping the strengths of a board being able to perform its function to the best of their ability (Brammer, et al., 2009). A diverse board with a broad range in backgrounds, cultures, and knowledge are better equipped to make decisions for global companies. They have more knowledge about the market needs and can thus better form the correct strategies. Showing they understand the different needs in the currently globalizing business environment. Possibly, leading stakeholders to give better reputations to the firm as they know exactly how to meet their consumer needs.
Furthermore, the socioeconomic environment with which the firm is embedded with plays a crucial role in reputational judgement made by public and stakeholders (Fombrun & Shanley, 1996). As we are currently living in a more global economy with connections all over the world, addressing diversity within a firm can signal to the world that they are adhering to today’s expectations. They are meeting the needs of a diverse market and show they understand the need to operate in a global environment (Brammer, et al., 2009). Having a gender and racial diverse board composition would signal improvement and thus quality and reputation towards the consumers and stakeholders about the values and ethical principles firms follow. Possibly allowing firms to climb up reputational ranking, such as Fortune’s Most Admired Companies lists when looking at a reputational subcategory scoring. Research conducted by Bernandi et al. (2009) finds that boards with a high percentage of women are more likely be put on ranking lists such as best company to work for or most ethical company to work for.
Furthermore, individuals find reputation to be an important aspect to take into consideration when they are making investment choices. As Helm (2007) found that stock performance is positively affected by the reputational scores of firms. If this signal of having more women on board gives such a strong effect on reputational scores it would be logical to hypothesize that, just like women, racial/ethnic background could also have positive
reputational effect on a firm’s image. These all serve as important signals towards the public and stakeholders about the firm’s reputation and ethical corporate governance.
The preceding arguments suggest that having a diverse board contributes to the external perception of firms, meet the needs in a global environment and investment choices. Therefore, the following hypotheses follow:
Hypothesis 3a: There is a positive correlation between racial diversity and reputation. Hypothesis 3b: There is a positive correlation between gender diversity reputation.
2.4 Mediation of Innovation and Reputation
As mentioned, various studies have focused on the direct relationship between board diversity and financial performance. These have all concluded different outcomes, positive effect, negative effect or no effect at all. Yet, little to none of these studies focus on how this relationship is formed. It could be mediated by various variables that could impact the relationship between the two. As explained above, two of these variables that are believed to impact the relationship of diversity and financial performance are innovation and reputation.
With the theory and empirical evidence argued throughout this section, it is believed that innovation acts as a mediating role which influence the relationship between board diversity and firm performance. An increase in diversity leads to more critical thinking, greater social network and creativity, key to innovation. Furthermore, evidence from previous research concludes that innovation has a positive effect of firm performance (Zahra & Garvis, 2000; Teece et al., 1997; Hilt et al., 1997). Therefore, it is suggested that innovation acts as a mediator between both gender and racial board diversity and firm performance.
Hypothesis 4a: Innovation acts as a mediator for relationship between gender diversity and firm performance.
Hypothesis 4b: Innovation acts as a mediator for the relationship between racial diversity and firm performance.
Additionally, various research has been conducted that shows the positive influence of reputation on firm performance (Roberts & Dowling, 2002; Hall, 1993). With the evidence given above based on theory and previous empirical studies diversity, both racial and gender should increase firm reputation. Therefore, the following hypotheses will be tested on reputation:
Hypothesis5a: Reputation acts as a mediator for the relationship between gender diversity and firm performance.
Hypothesis 5b: Reputation acts as a mediator for the relationship between racial diversity and firm performance also increases.
2.5 The Conceptual Model
Figure 1 below gives a visual overview of the relationships that will be tested in this empirical study. As shown, the dotted line, c’, visualizes the direct relationship between the board diversity (gender and racial) and firm performance. Where the solid lines present the mediating relationship between all variables. The pathway from a to b represents the
mediating relationship in which board diversity (gender and racial) leads to firm performance through reputation. The pathway from c to d represents the mediating effect of innovation: the pathway in which board diversity (gender and racial) lead to firm performance through innovation.
Figure 1. The Conceptual Model.
3. Methodology
This section discusses all aspects of the data used for the empirical research. Firstly, an explanation on how the data is collected and how the database is created, followed by an in-depth explanation of the specific variables and how they are calculated. The section will conclude with an overview of the statistical tests that will be run in section 4.
3.1 Sample
The data used are companies on the Fortune 500 list. These are all American
companies and includes both public and private companies, with the exception of those that do not file with a government agency. These 500 companies represent two-thirds of the U.S. GDP, with a revenue of $12.8 trillion and a market value of $21.6 and employ 28.2 million people around the world (Fortune500). Fortune 500 is a representative sample because the list includes the largest companies present in their sectors and a subset of this list provides
essential variable information necessary for the analysis. This group from the Fortune 500 list published their boards racial background. Furthermore, the database available only publishes board executive demographics of American companies. The use of companies that are too small could lead to many other contextual factors that have to be considered when testing the hypotheses and valuable data information is missing to allow for a full analysis of all
variables. With their global presence, they will be able to generate a larger effect with their innovation and reputational effect. Initially the database contained 500 companies but due to lacking data information the sample size dropped. Initially dropping to 477 companies’ due to missing ticker symbol information to collect the data. This information is crucial to collect the financial and diversity measurements in the Wharton Database. After having collected all the data and merging the various databases the sample size dropped to 261 companies’ due to missing data. The database covers a period of two years. This is a short period, yet the
database was constrained due to the reputation variable, which is only available for the years 2015 and 2016, before this reputation data was not published publically and the data for 2017 is not yet available.
3.2 Independent variables: Gender and Racial Board Diversity
The independent variables that are used for this thesis are obtained from the ISS Metrics database. This database is the world’s leader in providing data on corporate
governance and responsible investment solution. They focus on key extra-financial datasets, allowing for the examination of key issues and of the board demographics and more. For this thesis, the focus is specifically put on gender and ethnicity diversity aspects of the director’s demographic characteristics available in the database (ISS, 2018). Information on board demographic characteristics in terms of race and gender, among other information is provided.
Two measures of diversity are used; the Blau Index and proportion measurement, both for the gender and racial diversity. Blau stands for the Blau Index of Diversity,
formulated as: 𝐻 = 1 − ∑ 𝑝𝑖2, where H stands for the board diversity measurement focusing on a specific demographic, pi is the proportion of board members in each specific
demographic category of the company. The Blau index is often the measurement of choice for diversity research and it is the ideal measurements for testing diversity (Harrison & Klein, 2007). A Blau measurement can take a value between 0 and 0.75. A score of 0.75 indicates a firm with a very diverse board composition, while a value close to zero indicates minimal board diversity. This measure is calculated for both gender and racial diversity. For gender the two-different pi’s were female or male while for racial the different pi’s were dependent on either being Caucasian or a minority race. The minority race included all other possible raced compared to Caucasian. Having calculated them separately would have ended up with very small sample sizes. The minority races include African America, Middle Eastern, Hispanic, Asian, Indian, and Native American.
Proportional measurement of diversity will be used as a robustness test measurement. The proportion is calculated by dividing the demographic category by the total. For example, for gender diversity, the proportion of the board that is female is calculated as follow: number of females on the board divided by the total number of board members per firm. This
measurement is added to the research to act as a robustness test, to ensure the quality of the results initially obtained by the Blau index measurement of diversity. The same method was used to calculate the racial proportions. Again, as with the Blau index, using majority versus minority group proportion.
Important to note that lagged values of board diversity will be used of t-1. The reasoning being that the effect of diversity might not be directly seen in the financial performance in the same year and it might take some time for the effect to be seen in the results of firms. Therefore, the data of these variables are retrieved for the years 2015 and 2016.
3.3 Dependent variables: Corporate Financial Performance
The COMPUSTAT database has been collecting data since it was established in 1962. Collecting financial, statistical and market data for thousands of global companies around the world. Access to this database was possible through Wharton Database, and the necessary
data points on financial variables were retrieved here. The corporate financial performance of each company will be measured through the accounting based measurement, return on assets, ROA. This financial measurement has been used by various other research that focus on the topic of board diversity and financial performance (Berman et al., 1999; Kang et al., 2010; Tang, et al., 2012). The ROA looks at the efficiency measure of a company use of its total assets throughout a fiscal year, specifically looking at operational excellence. As the COMPUSTAT database does not readily have the data on ROA available, it was calculated manually in Excel for the years 2016-2017. The necessary data points that had to be retrieved from the database were a company’s net income and their total assets during a fiscal year to calculate the return on assets. The following formula was used to calculate each firm’s return on assets:
𝑅𝑂𝐴 = 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠𝑁𝑒𝑡 𝐼𝑛𝑐𝑜𝑚𝑒.
Tobin’s Q is used as another performance measurement for the initial graphical results in section 4. Tobin’s Q measures the growth potential of firm. In other words, it is focused on more long-term performance compared to ROA which focuses more on the short-term performance. This measure is frequently used by researchers as a measure for firm performance (Carter et al., 2003; Wernerfelt, et al., 1988). Tobin’s Q was calculated manually in excel, similar to ROA, the data for this variable was not readily available from the database. The following formula was used to calculate each firm’s Tobin’s Q:
𝑇𝑜𝑏𝑖𝑛′𝑠 𝑄 = 𝑇𝑜𝑡𝑎𝑙 𝑚𝑎𝑟𝑘𝑒𝑡 𝑣𝑎𝑙𝑢𝑒 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠
3.4 Mediators: Innovation and Reputation
The theoretical framework presents the argumentation why certain variables will act as mediators between the relationship of board diversity and firm performance. For both the mediators, similar to the independent variables, the lagged values of the variable will be used to analyze its effect on firm performance, t-1. Similar to the independent variables, the data for these two variables is retrieved for the years 2015 and 2016.
Firstly, the innovation intensity variable is calculated by deriving the needed data from the COMPUSTAT database, collected together with the financial data that was
as proxy to measure innovation intensity (O’Brien, 2003, Balkin, et al., 2000). The following formula was used to calculate innovation intensity:
𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛 𝐼𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 = 𝑇𝑜𝑡𝑎𝑙 𝑅&𝐷 𝐸𝑥𝑝𝑒𝑛𝑠𝑒𝑠𝑇𝑜𝑡𝑎𝑙 𝑆𝑎𝑙𝑒𝑠 .
For the reputation variable, the Fortune Most Admired Companies (MAC) list was used by manually collecting the overall scores of all the companies for the years 2015-2016. Fortune’s MAC list is one of the best-known measurements used for firm reputation (Black, et al., 2000; Fombrun 1996). The reputation score is collected through a company rating survey which executives have to complete in their own industries. The scale runs from 0 to 10, 0 being poor and 10 being excellent. The rating is based on a total of eight attributes of performance; wise use of corporate resources, responsibility to community and development, ability to attract, develop and retain people, financial soundness, long-term investment value, perceptions of innovativeness and lastly, quality of the products (Korn Group, 2018).
3.5 Control Variables
Important to take into account are other possible variables that could influence the relationship between board diversity and financial performance. The following variables, that could be confounding, will be considered for this analysis, firm size, risk, and industry. These are included to exclude the possibility of biased results. For these variables, no lagged values will be used as the current firm size and risk are believed to take into account the effect of both the previous values and currently. These variables were chosen as other researchers have included these control variables, using current and not lagged values, in the diversity and firm performance analysis as they most often result to have a significant effect on firm
performance (Cater, et al., 2003; Carter, 2010; Campbell & Minguez-Vera, 2008; Mahadeo, et al., 2011).
Fortune 500 by itself is already considered to include only the largest companies. There are no small businesses considered in this sample. Yet, even within such a group there can still be differences in sizes. Therefore, it is important to control for firm size as it can be a confounding variable for the relationship being tested. For this thesis number of employees will be used as a proxy for firm size. The firm size control is calculated as follows:
𝐹𝑖𝑟𝑚 𝑠𝑖𝑧𝑒 = ln(𝐸𝑚𝑝𝑙𝑜𝑦𝑒𝑒𝑠).
Additionally, risk is often considered a confounding variable. Risk in an important factor to take into consideration as the amount of leverage a company has will affect the
amount of money investors are willing to spend on firms and this could lead to possible growth and success of firm. This has been concluded to affect finance performance after having controlled for many factors. This variable is calculated by computing the ratio as follows:
𝑅𝑖𝑠𝑘 = 𝐿𝑜𝑛𝑔 𝑇𝑒𝑟𝑚 𝐷𝑒𝑏𝑡 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠
Lastly, industry is also believed to have an effect on firm performance therefore also important take this into account. However, industry differences are controlled differently than the other two control variables discussed above. The database is structured as panel data hence I will use company fixed effects for all firms. Meaning that industry will already be controlled for as the analysis will focus specifically on the within firm effect instead of across firm effects.
3.6 Statistical analysis
Primarily, graphical results will be visualized to give a general overview of the data. Followed by regression analysis in order to test the hypothesis presented in my theoretical framework. Specifically, panel data regression analysis with fixed effects will be performed between the independent variable, the dependent variable and mediators while controlling for the control variables.
A common method used by researchers to establish a mediating relationship is a four-step process established by Baron and Kenny (1986). Step 1 involves establishing the
relationship between the independent variable and the dependent variable. Path c’ in the conceptual model in section 2.5. Step 2 establishes the relationship between the independent variable and the proposed mediating variables, paths a and c in the conceptual model. Step 3 involves establishing a relationship between the mediators and the dependent variable, paths b and d in the conceptual model. Lastly, step 4 looks at the regression analysis of the
dependent variable on both the independent variable and the mediator together. For the mediating effect to take place the coefficient of the independent variable must be reduced in step 4 compared to step 1 and steps 1-3 must all obtain significant results in the hypothesized direction.
The following regression formula will be used to test hypothesis 1 to fulfill step 1 of the Baron & Kenny (1986) method:
(1) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝐺𝑒𝑛𝑑𝑒𝑟𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡−1+ 𝛽2𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽3𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝜀𝑖,𝑡 (2) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝑅𝑎𝑐𝑖𝑎𝑙𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡−1+ 𝛽2𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽3𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝜀𝑖,𝑡
Followed by analyzing the relationship between the independent variables and the mediators. The following formula’s will be used to test the relationships of hypotheses 2 and 3 and step 2 of the Baron & Kenny (1986) method:
(3) 𝑅𝑒𝑝𝑢𝑡𝑎𝑡𝑖𝑜𝑛𝑖,𝑡 = 𝛼 + 𝛽1𝐺𝑒𝑛𝑑𝑒𝑟𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡+ 𝛽2𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽3𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝜀𝑖,𝑡 (4) 𝑅𝑒𝑝𝑢𝑡𝑎𝑡𝑖𝑜𝑛𝑖,𝑡= 𝛼 + 𝛽1𝑅𝑎𝑐𝑖𝑎𝑙𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡+ 𝛽2𝑅𝑖𝑠𝑘𝑖,𝑡 + 𝛽3𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝜀𝑖,𝑡 (5) 𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛𝑖,𝑡 = 𝛼 + 𝛽1𝐺𝑒𝑛𝑑𝑒𝑟𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡+ 𝛽2𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽3𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝜀𝑖,𝑡
(6) 𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛𝑖,𝑡 = 𝛼 + 𝛽1𝑅𝑎𝑐𝑖𝑎𝑙𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡+ 𝛽2𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽3𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝜀𝑖,𝑡
Step 3 involves looking at the relationship between the mediators and the dependent variable. Thus, the relationship between innovation and firm performance and reputation and firm performance. The following regression will be used:
(7) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝑅𝑒𝑝𝑢𝑡𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1+ 𝜀𝑖,𝑡 (8) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1+ 𝜀𝑖,𝑡
Finally, the following regressions will be used to analyze hypotheses 4 and 5 and step 4 of the Baron & Kenny (1986) method to establish if the mediating effect takes place:
(7) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝐺𝑒𝑛𝑑𝑒𝑟𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡−1+ 𝛽2𝑅𝑒𝑝𝑢𝑡𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1+ 𝛽3𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽4𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝜀𝑖,𝑡 (8) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝐺𝑒𝑛𝑑𝑒𝑟𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡−1+ 𝛽2𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1+ 𝛽3𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽4𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝜀𝑖,𝑡 (9) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝐺𝑒𝑛𝑑𝑒𝑟𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡−1+ 𝛽2𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1+ 𝛽3𝑅𝑒𝑝𝑢𝑡𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1 + 𝛽4𝑅𝑖𝑠𝑘𝑖,𝑡 + 𝛽5𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝜀𝑖,𝑡 (10) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝑅𝑎𝑐𝑖𝑎𝑙𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡−1+ 𝛽2𝑅𝑒𝑝𝑢𝑡𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1+ 𝛽3𝑅𝑖𝑠𝑘𝑖,𝑡 + 𝛽4𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝜀𝑖,𝑡 (11) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝑅𝑎𝑐𝑖𝑎𝑙𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡−1+ 𝛽2𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1+ 𝛽3𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽4𝑆𝑖𝑧𝑒𝑖,𝑡 + 𝜀𝑖,𝑡
(12) 𝑅𝑂𝐴𝑖,𝑡 = 𝛼 + 𝛽1𝑅𝑎𝑐𝑖𝑎𝑙𝐷𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦𝑖,𝑡−1+ 𝛽2𝑅𝑒𝑝𝑢𝑡𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1+ 𝛽3𝐼𝑛𝑛𝑜𝑣𝑎𝑡𝑖𝑜𝑛𝑖,𝑡−1 + 𝛽4𝑅𝑖𝑠𝑘𝑖,𝑡+ 𝛽5𝑆𝑖𝑧𝑒𝑖,𝑡+ 𝜀𝑖,𝑡
4. Empirical Results
In the following section the empirical results will be presented and analyzed. Firstly, an overview of the descriptive statistics of the variables is presented in Table I and Table II. Followed by a Pearson correlation matrix and initial graphical results. Concluding with the statistical analysis presented in section 4.6.
4.1 Summary Descriptive Statistics
Within Table I one can find information concerning the variables mean, standard deviation, minimum and maximum value the variables have. For racial diversity Blau index the mean value is 0.28 with a standard deviation of 0.10 and for gender diversity the mean Blau index is 0.32 with a standard deviation of 0.09. Both Blau mean indexes are quite low considering the highest possible value for a Blau index is 0.75. However, when looking at the maximum value of both indexes we can observe that some firms do have a 0.5 value for diversity within their firm. Furthermore, noteworthy to mention that the minimum Blau diversity measurement is 0.12 for both, meaning all firms have at least some measure of diversity.
Looking at the descriptive statistics of the mediators, reputation has a mean value of 6.62. This is a high average considering the maximum value of any firm is 8.8 for the database. Looking at the innovation variable, which measures innovation intensity of a firm there is a large range within the database. Looking at a minimum of 0.02% and a maximum of 39.9% innovation intensity. Thus, interesting to deepen the analysis to see if this has any meaning for firm performance with the relationship to board diversity.
Table I. Summary Descriptive Statics
Variable Observations Mean Std. Dev. Minimum Maximum
ROA 522 0.046 0.0691 -0.4588 0.3194
Tobin's Q 522 1.3589 1.1771 0.000 7.4796
Ethnic Blau Index 445 0.2781 0.0965 0.1172 0.5000
Gender Blau Index 514 0.3267 0.0885 0.1172 0.5000
Proportion Minorities 445 0.1772 0.0823 0.0620 0.5000 Proportion Female 514 0.2184 0.0860 0.0625 0.6667 Innovation 260 0.0568 0.0703 0.0003 0.3994 Reputation 366 6.6153 0.6979 4.4000 8.800 Risk 522 0.2810 0.1811 0.0000 1.7753 Size 518 1.6027 0.4963 -0.6882 3.3617
Following the descriptive statistics of the data an overview of the industries that are included in the database. Table II indicates there is an even spread of firms from the
Industrials, Consumer Discretionary, Health Care and Information Technology industries. With Consumer Discretionary having the biggest presence in the database with 18.20%. Telecommunications is the least present industry in the database accounting for only 0.77% of all firms in the database.
Table II. Industry sectors
GIC Sector Frequency Percentage Cumulative
10 - Energy 24 4.60% 4.60% 15 - Materials 35 6.70% 11.30% 20 - Industrial 89 17.05% 28.35% 25 - Consumer Discretionary 95 18.2% 46.55% 30 - Consumer Staples 53 10.15% 56.70% 35 - Health Care 70 13.41% 70.11% 40 - Financials 54 10.34% 80.46% 45 - Information Technology 68 13.03% 93.49% 50 - Telecommunication 4 0.77% 94.25% 55 - Utilities 24 4.6% 98.85% 60 - Real Estate 6 1.15% 100.00% Total 522 100,00% 4.2 Correlation Matrix
In this sub section the Pearson Correlation matrix will be discussed of all the variables used for the analysis. A Pearson correlation coefficient of the various independent and
dependent variables will be presented. The value of a correlation coefficient can have a value between -1 and 1. -1 indicates there is a negative linear relationship between two variables, whereas +1 indicates a positive linear relationship between two variables. Two variables are considered to be strongly correlated when their correlation coefficient is between the
following ranges. For a strong positive correlation, the coefficient has to be in the range of 0.7 to 1.0, while a strong negative correlation falls in the range of -0.7 to -1.0 (Field, 2013). Table III, below, presents the correlation coefficient of all the variables used in this empirical study.
The first conclusion that can be drawn from the table is that there are no variables in this empirical study that are strongly correlated, implying there is no multicollinearity. Taking a closer look at the mediators’ various results are concluded. Firstly, the correlation between reputation and the diversity measurement, is negatively correlated and insignificant, -0.0043 and -0.0095, respectively. The opposite of what was expected based on the
argumentation in the theoretical framework. For innovation, it is positively correlated with both gender and racial Blau indexes, 0.0273 and 0.0662, respectively. However, just as reputation, these are also insignificant correlation coefficients.
There is no strong or significant correlation between the mediators, reputation and innovation, with return on assets. However, when looking at Tobin’s Q as a performance measurement both reputation and innovation have a positive significant coefficient, 0.2518 (p<0.05) and 0.2847 (p<0.05) respectively. Indicating a positive linear relationship between the variables. Tobin’s Q is considered to be a measurement that focusses more on long term performance and thus could be a possible reason why the long-term performance
measurement is significantly correlated to the mediator variables while they are not
significantly correlated with the short-term performance measure indicator, return on assets. Additionally, racial and gender Blau measurements are positively and significantly correlated with the both measurements for firm performance.
4.3 Graphical Results
Graph I. Diversity and Firm Performance
Graph II. Diversity and Firm Performance (2)
Initial graphical results between board diversity and firm performance are shown above in Graphs I and Graph II. Important to note, Graph I indicates Return on Assets as the firm performance measurement wile Graph II uses Tobin’s Q. For both performance
measurements, the left graph is based on racial diversity and the right graph is based on the gender diversity Blau index values. All graphs show an upward sloping fitted line indicating there is a positive relationship between diversity and firm performance. The data used for these graphs include all the possible industries that are accounted for within the database. As visualized in the graphs above, the slope for the relationship with Tobin’s Q is much steeper than that with the ROA measurement. This gives a very general overview, allowing one to only conclude with caution that there is a positive relationship between board diversity and firm performance. Deeper analysis within the data will be needed to confirm the developed hypotheses.
Graph III. Diversity and Mediators
Secondly, Graph III visualizes the relationship between the two Blau diversity measurements and the mediators, reputation and innovation. The relationship between reputation and both the Blau index for gender and race is almost flat-lined and seems to have almost no linear relationship. However, for the innovation mediator there seems to be a very slight and minimal linear relationship for both gender and racial Blau score. No definitive conclusions can be drawn from these graphs. Deeper analysis, with regression results, will allow for more conclusive results.
Graph IV. Robustness Test: Diversity and Firm Performance
Graph IV tests the exact same relationship as Graphs I and Graph II, however, now a different diversity measurement is used to see if the results of the first graphs are reliable. In these graphs, proportions are used as measurement for diversity, as explained in my
methodology section. As can be seen from these four graphs they have very similar trends as those of the first graphs that use the Blau index as diversity measurement. Allowing for only a cautionary conclusion that there is a positive relationship between board diversity and firm performance. Deeper analysis within the data will be needed to confirm the developed hypotheses.
4.4 Regression Results
In the following section the regression results are explained. The research method used for the regression analysis is panel data with fixed effects for the period of 2016-2017 of financial data and 2015-2016 for the independent variables and the mediators. This method will control for different types of omitted variables without actually observing them. This is because the analysis will study the dependent variable, firm performance, over time, with the possibility of eliminating the effect of any possible omitted variables that could differ across firms but are constant over time (Stock and Watson, 2015, p. 396). The dataset is an
unbalanced panel dataset. There are 261 companies in the dataset with data from over two periods, resulting in a total of 522 observations. This is also shown in Table I, which presents the descriptive statistics of the dataset.
The graphs in section 4.3 show the initial results of the relationship, however this analysis only takes into account possible between company effects. The panel data regression analysis with fixed effect will allow for a deeper analysis on within company differences. Furthermore, the correlations presented earlier in this section do not automatically imply causation. Even with this regression analysis no causal relationship between the variables can be concluded as there might be other variables not taken into account in this analysis that could affect the relationship or the results of the statistical test are a random coincidence. Furthermore, this empirical analysis is not a randomized experimental study, the most powerful tool to establish a cause and effect relationship (DeVoto & Zweig, 2015). Hence, only careful associations between the variables can be made. The rest of this section will focus on answering the hypotheses developed in section 2. In terms of reading the table, the first model per variable is tested without taking into account control variables followed by a regression that does include them.
The analysis will start by looking the results for step 1 of the Baron & Kenny (1986) method. Hypothesis 1a and 1b hypothesize that both, gender and racial diversity, have a positive relationship with firm performance. Results are presented in Table IV, where
multiple regressions are summarized. For ethnic diversity and firm performance, presented in Model 1a and 1b, the coefficients are negative, -0.0085 and -0.0088 respectively, and
insignificant. Meaning that when diversity increases by 1 unit return on assets will decrease by 0.88%. This is the opposite direction compared to what was expected in the relationship. Allowing us to conclude that hypothesis 1b is rejected. For gender diversity, the results are presented in Model 2a and 2b. Without control variables, the coefficient (0.0150, p<0.05) is positive and significant. This result remains when adding the control variables into the regression the coefficient (0.0151, p<0.05), also still positive and significant. Indicating that when gender diversity increases by 1 unit return on assets increases by 1.51%. Hypothesis 1a is therefore not rejected. Risk is the only control variable with significant positive results on ROA. In other words, when leverage increases return on assets increases 0.3746 (p<0.01) and 0.3272 (p<0.01) points, with their respective diversity measurements. The F-tests for all models are significant with a 1% level with the exception of the first and fourth model.
Table IV. Results of Regression: Board Diversity and Mediators on Firm Performance
Variables Return on Assets (ROA)
Model 1a Model 1b Model 2a Model 2b Model 3 Model 4 Ethnic Blau Index -0.0085 -0.0088
(0.0062) (0.0060)
Gender Blau Index 0.0150** 0.0151**
(0.0065) (0.0063) Reputation -0.0125** 0.0138 (0.0056) (0.0288) Innovation Risk Control 0.3746*** 0.3275*** (0.0776) (0.0770) Size Control -0.1191 -0.0927 (0.0723) (0.0710)
Fixed Effect Yes Yes Yes Yes Yes Yes
R2 0.0092 0.1249 0.0220 0.0990 0.0072 0.0061
F-Test 1.88 9.37*** 5.40** 8.61*** 5.08** 0.23
Notes. Standardized Coefficients. Standard error in parentheses. * p<0.10; ** p<0.05, ***p<0.01
The analysis will now move to step 2, establishing a relationship between the proposed mediators and the independent variables in Table V and VI. Hypothesis 2a and 2b
hypothesized that gender and racial diversity have a positive relationship with innovation intensity. Model 5a and 5b display the results for hypothesis 2b. The corresponding coefficients for the Blau Indexes are -0.0128 and -0.0132 both of which are insignificant. Although the controls, risk and size, added in Model 5b do result in significant result, -0.1232 (p<0.0) and 0.1506 (p<0.01), respectively. Meaning that if racial diversity increases the innovation level will decrease by 0.0132. Furthermore, no significance was found therefore rejecting hypothesis 2b. Hypothesis 2a is also rejected, as can be concluded from Model 6a and 6b. The corresponding coefficients for the gender Blau index are 0.0123 and 0.0119. In contrast to racial diversity, if gender diversity increases innovation will increase by 0.0119 points. The effect is positive yet insignificant. Their robustness tests, presented in Model 7 and Model 8, find similar results as those in Models 9 and Model 10, -0.0132 and 0.0077 respectively, and also insignificant.
Table V. Results of Regression: Innovation and Diversity
Variables Innovation
(Robustness test)
Model 5a Model 5b Model 6a Model 6b Model 7 Model 8
Ethnic Blau Index -0.0128 -0.0132 (0.0296) (0.0301)
Gender Blau Index 0.0123 0.0119
(0.0328) (0.0332) Proportion Minorities 0.0077 (0.0329) Proportion Female -0.0132 (0.0287) Risk Control -0.1232** -0.1595 -0.1212 -0.1633 (0.3547) (0.3395) (0.3547) (0.3394) Size Control 0.1506*** -0.1875 0 .1460 -0.1859 (0.3741) (0.3262) (0.3739) (0.3264)
Fixed Effect Yes Yes Yes Yes Yes Yes
R2 0.0019 0.0047 0.0012 0.0064 0.0048 0.0058
F-Test 60.22*** 55.93*** 78.66*** 75.08*** 55.97*** 75.36*** Notes. Standardized Coefficients. Standard error in parentheses. * p<0.10; ** p<0.05,
***p<0.01
Moving on the second mediator, reputation, with statistical results for hypotheses 3a and 3b are presented in Table VI. Hypothesizing that reputation is positively related to gender and racial diversity. As shown in the table, however, both hypotheses are rejected. Model 9a
