Country human development as a moderator between TMT-diversity and Firm Performance

(1)

Country human development as a moderator

between TMT-diversity and Firm

Performance

A comparative study between high developed and medium

developed countries.

Master thesis

MSc International Business & Management

Faculty of Economics and Business

University of Groningen

21

th

of June 2017

By

(2)

Master Thesis

Msc International Business & Management

Country human development as a moderator between TMT-diversity

and firm performance

A comparative study between high developed and medium developed

countries.

21/06/2017

By Jeljer Haarsma

S3000710

Supervisor: Dr. M.H.F. Ridder de van der Schueren

Second assessor: Dr. R.W. De Vries

Faculty of Economics and Business University of Groningen

Duisenberg Building, Nettelbosje 2, 9747AE Groningen, The Netherlands P.O. Box 800, 9700AV Groningen, The Netherlands

(3)

Abstract:

This thesis researched the effect of country human development on the relation between Top Management Team (TMT) Diversity and firm financial performance. The theoretical base for this thesis is the Upper Echelon Theory. Additionally several constructs will be used for TMTs, TMT-diversity and the Human Development Index (HDI). Firm financial performance is measured in Return on Assets (ROA), and TMT-diversity is a

combination of, national, gender and age-diversity made into one variable with the help of the Blau-index. The data was gathered from BoardEx, Bureau van Dijk and the Human

Development Repot (HDR). A regression analysis was used to show the relation between the independent and dependent variables, and the effect of the moderating variable. No significant results were found for the relation between the independent and the dependent variables. And thus the moderator could not be included to check how this influences the relation between the independent and the dependent variable. A sub-sample however did show significant results for TMT-diversity on firm performance.

Keywords: top management teams, diversity, gender diversity, national diversity, age

diversity, firm performance, human development index, emerging markets, developed markets.

(4)

Preface

Finishing this master thesis marks the end of a chapter and the beginning of a new one. I am already looking forward with great excitement as what is to come next. In the last words in the final chapter of my academic career I would like to acknowledge and thank a few people. First off I would like to thank my parents for supporting me in whatever I wanted to do. This freedom gave me the opportunity to develop me in the most complete way.

Additionally I would also like to thank my supervisor dr. Maurice Ridder de van der Schueren for his continued support during the writing of this master thesis. Just having the possibility to spar with him gave me the confidence to push on and keep improving my thesis. Finally I would also like to thank my girlfriend for supporting me during the busy times of writing this thesis.

Jeljer Haarsma

(5)

Tables and Figures

Formula 1: Blau Index (Blau, 1977) ... 26

Figure 1: Strategic choices under effect of bounded rationality (Hambrick & Mason 1984) .. 12

Figure 2: Type and amount of diversity (Klein & Harrison, 2007) ... 13

Figure 3: Values for Blau's-Index of diversity ... 28

Figure 4: Development of TMT-diversity for separate countries during 2013-2015 ... 29

Figure 5: Buildup of TMT-diversity per variable per country per year. ... 30

Table 1: Distribution of final sample ... 23

Table 2: Number of categories and their theoretical maximum for diversity ... 26

Table 3: Descriptive statistics ROA and Control Variables ... 30

Table 4: Correlation Table ... 31

Table 5: Model Summary ... 32

(8)

1. Introduction

The effect of TMT-diversity on a firm’s performance has been classified as an issue in organizational outcomes as well as the financial literature (Adams, Hermalin, & Weisbach, 2010; Milliken & Martins, 1996). Diversity in TMTs is the variety in the composition of the team, and can be measured in many dimensions such as for example age, gender, nationality, education and experience (Campbell & Mínguez-Vera, 2008). First of all diversity of boards can be made into two groups, demographic (observable measures like gender and age), and cognitive diversity (unobservable like education knowledge and values), the latter being harder to measure than the first(Milliken & Martins, 1996).

The main basis for this paper is the upper-echelon theory as first discussed by Hambrick in 1989. This theory initiated research focusing on the relation organizational outcomes have with specific characteristics of an executive (Hambrick & Mason, 1984). According to the upper-echelon theory a TMT/executive’ characteristics influence the direction of a firm, consequently this information is important for strategy development, structure and the actual performance of the company (Boeker, 1997; Hurt, 2015; Murray, 1989; Rau, 2006).

Many previous studies focused on the size and composition of the TMT in the U.S. An increase in the separation of ownership and control is the main reason for this focus, since this decreases the amount of agency costs. Research on MNCs in the U.S. resulted in significant positive effects of diversity on firm performance (Carter et al., 2003). Since then researchers have focused more on this relationship by using several measures for diversity such as gender, age, nationality and other demographics and their impact on firm performance (Carter et al., 2003; Erhardt, Werbel, & Shrader, 2003).

The results of these widening studies have delivered positive effects on firm

(9)

A reason for focusing on TMT-diversity further is

that gender diversity in boards and executive positions is a subject of hot debate. This discussion has countries in a chokehold while finding ways to increase diversity and

specifically female participation in executive and management positions. Some countries in Europe specifically have already incorporated mandatory quotas for a certain amount of female board directors’. Germany for example has made a reform as of 2015 which main goal is to have 30% of corporate boards consist of women(Bertrand, Black, Jensen, & Lleras-Muney, 2014). This is another big reason to increase the field of research.

Moving on, several moderators for the relation between TMT-diversity and performance have already been found. Such as: internationalization, entrepreneurial orientation, experiences and group longevity (Carmen Díaz-Fernändez, Rosario Gonzälez-Rodrïguez, & Simonetti, 2016). These moderators strengthened the relation between diversity and firm performance in their research context. The optimal moderator has however not yet been located.

This thesis will focus on the relationship between TMT-diversity and Firm

performance. TMT-diversity will be measured with age, gender and national diversity within the TMT. Firm performance will be measured in the financial measure return on assets (ROA). Subsequently human development will be added as a new moderator in this field. These concepts lead to the following central research question:

Does country development have an influence on the relation between TMT-diversity and financial performance?

So far the empirical evidence on the relation between TMT-diversity and firm

performance is mainly concentrated on the US. Additionally most of the papers have focused on one single diversity measure using a single year of data. This paper will complement and expand the existing literature on this field by incorporating three measures of diversity; age, gender and nationality into a single measure with the help of a Blau-Index. Additionally this will be done over the course of a three-year period, instead of using a single year as a

reference frame. This should lead to more robust and reliable results, hopefully contributing to the adaption of more firms to a diversification policy for their boards. Finally medium

(10)

Expectations are that the inclusion of the country welfare moderator would cause a significant difference between more and less developed countries. Leading from the assumption that differences between people become more prevalent the more developed a country becomes.

The methodology used in this thesis consists of quantitative research methods. The three different diversity measures were made into one with the help of the Blau-Index. After which this measure together with the independent and control was placed in an

OLS-regression. The data-sources needed to gather this were the Orbis and BoardEx databases, and the Human Development Report (HDR).

To continue the thesis will be structured in the following way: Chapter 2 consists of the theoretical background on the constructs and recent findings, concluding with the hypothesis and conceptual model. Moving on Chapter 3 will describe the dataset, the

(11)

2. Theoretical Background

This section will provide an overview of the concept used and the theories and background information on these concepts and theories. First the relevant theories for the research paper will be discussed. After which the concepts used will be discussed and linked to each other.

2.1 Upper Echelon Theory

The main basis for this paper is the upper-echelon theory as first discussed by Hambrick in 1989. This theory initiated research focusing on the relation organizational outcomes have with specific characteristics of an executive (Hambrick & Mason, 1984). According to the upper echelon theory strategic decisions and organizational effectiveness are based upon the background, experiences, values and personal characteristics (Hambrick, 1989, 2007; Hambrick & Mason, 1984).

All top managers of a firm make up the TMT. These are the people who make the strategic decisions a firm is faced with (Hurt, 2015; Rau, 2006; Ruiz-Jiménez, Fuentes-Fuentes, & Ruiz-Arroyo, 2016). In the end the TMT is responsible for how to allocate companies’ resources, and decide which of the projects should be executed. Moreover they also decide on which objectives and goals to strive for.

Habrick in his original paper mentions the existence of several indicators for his constructs background, experiences and values. Indicators for these constructs can be demographical characteristics such as age, education, functional background and

socioeconomic background of TMT-members (Hambrick, 1989). Researchers capitalizing on Hambricks original papers have found that organization outcomes are explained to a larger extend by a TMT as compared to a single top executive (Hambrick, 2007).

(12)

Due to this concept of bounded rationality, a

manager or even several managers are not able to understand all aspects of an organization or its environment. This three-stage process is depicted in figure 1 below.

According to the upper-echelon theory a TMT or an individual executive’ characteristics influence the direction of a firm (Hambrick, 1989, 2007). And thus the

information on a TMT and executive characteristics are important for strategic development, and the structure and actual performance of a company (Boeker, 1997; Hurt, 2015; Murray, 1989; Rau, 2006). Additionally TMT and executive’ attributes have been found to relate to organizational actions (Finkelstein & Hambrick, 1996) and CSR practices(Chin, Hambrick, & Trevin, 2013).

Previous research regarding the upper-echelon theory mainly focused on developed countries, where it was shown that TMT-diversity has a positive effect on performance (Acar, 2015; Ararat, Aksu, & Tansel Cetin, 2015). Other papers, although less common, have shown that the effects between these two constructs yielded the same results for emerging economies specifically (Acar, 2015; Ararat et al., 2015). There however are no studies which combine both emerging and developed countries within a sample, including country welfare as a moderator will make this comparison possible.

As stated earlier the upper-echelon theory indirectly states that TMT-diversity has positive effects on firm performance. However this relationship’ effect differs over different environmental contexts and researches (Homberg & Bui, 2013; Keck, 1997). Keck (1997) has found evidence that firms’ with heterogeneous TMTs show better performance in more turbulent situations. Complexities in the environment require different perspectives and more diversified problem solving skills (Keck, 1997). On the other hand however homogeneous TMTs perform significantly better in a stable environment (Keck, 1997).

(13)

This theory will provide the foundation for this

research in that it shows that a TMT has an added benefit in explaining performance outcomes as compared to individual top executive characteristics. Additionally this theory has been often used to describe the link between TMT-diversity and performance, which makes it very applicable for this study.

2.2 Diversity

With the upper echelon theory as a base it can be stated that top managers have an influence on a firm’s financial performance. This paragraph will describe the effect of several different types of diversity on firm performance.

Diversity can be described as to how differences in a specific attribute such as tenure, are distributed among members of, in this case, a TMT (Klein & Harrison, 2007). Due to differences in the common attribute three different types of within-unit diversity can be described. The main differences between these types are within substance, pattern,

measurement method and finally their consequences. The figure at the bottom of this page shows the different types and their distribution (Klein & Harrison, 2007)

The three different types of diversity are: disparity, variety and separation. Main difference between them being their pattern and expected impact on a firm’s performance. Additionally the X-axis shows the amount of diversity. The specific combinations between

the amount and type of diversity result in a unique pattern. Each circle in this pattern represents a top manager. In the following

paragraphs the different types and amounts of diversity and its effect on firm performance.

(14)

2.2.1 Separation

One would talk about separation when top managers have differences in opinion in a work-related case. The main reasons for differences in opinion are caused by different values, beliefs, attitudes or a combination of these reasons(Klein & Harrison, 2007). As can be seen in figure 2 on page 13, the minimum amount of diversity is achieved when all managers have the same opinion. According to this theory these managers should have somewhat the same values, beliefs and attitudes to achieve this. Additionally the maximum amount of diversity is achieved when opinions are at the opposite sides of the spectrum but still stacked amongst managers (Klein & Harrison, 2007). High diversity can thus only be achieved when managers are highly diverse in their values, beliefs and attitudes. A moderate amount of diversity is present when managers do have different opinions, but they aren’t as “far” from each other (Klein & Harrison, 2007).

The process theory is very similar to the separation type of diversity. Like separation the process theory states that diversity has an effect on group processes, the main effect of diversity being a change in communications and conflict. With, in another similarity to separation has an effect on firm strategy and performance according to the process theory. The process theory has its base in Schneider’s (1987) similar-attraction-theory (SA-theory) and the social categorization theory (Schneider, 1987; Turner, Hogg, Oakes, Reich, &

Wetherell, 1987). The SA-theory states that companies are more prone to pick employees who fit well into the already established character of the company, causing a more homogenous TMT.

Additionally other negative effects of separation as diversity will appear due to a decrease in communications (Smith et al., 1994), less consensus (Knight et al., 1999), more personal conflicts (Pelled, Eisenhardt, & Xin, 1999) and decision making taking longer (Campbell & Mínguez-Vera, 2008; Hambrick, Cho, & Chen, 1996). This would mean that a more homogeneous group would be more cooperative, with less conflict.

Moving on the social categorization theory suggests that people within a group would rather interact within the group compared to interacting with “outsiders”(Turner et al., 1987). Both theories conclude that diversity as separation in a TMT has a negative influence on group processes and that it decreases performance.

(15)

2.2.2 Variety

Variety as a measure of diversity considers the differences between TMT-members in their knowledge levels and amount of experience (Klein & Harrison, 2007). An example of minimum variety in knowledge would be that every TMT-member has the same level in their educational background. However even differences within a level can cause diversity in variety, for example one TMT-member did a financial study and the other studied

Psychology. On the other hand, one would talk about maximum variety when every possible category in both knowledge and experience is equally represented within the sample.

Variety as diversity has literature backing its positive effect on diversity, the main reason for this being the differences in top managers causing unique insights and thus increasing decision quality (Carter et al., 2003). Additionally, the benefits of variety can be gained in four ways: increased information resources (Kauer, Waldeck, & Schäffer, 2007), greater ability to scan the environment (Keck, 1997; Robinson & Dechant, 1997), enhanced evaluation of strategic decisions (Hambrick & Mason, 1984), enhanced creativity and innovation (Arfken, Bellar, & Helms, 2004) and diminished group-thinking (Bantel & Jackson, 1989).

Additionally in more specific descriptions, variety increased the understanding of a TMT regarding potential consumers and employees (Robinson & Dechant, 1997). And also providing legitimacy for the firm in regard to external and internal constituencies (Carter, Souza, Simkins, & Simpson, 2007).

Concluding, diversity as variety, according to the literature should lead to new insights which increase the decision quality and thereby increase firm financial performance.

2.2.3 Disparity

The third and final type of diversity is disparity, and represents vertical differences. This includes, for example, the difference in rank or status of a TMT-member. When all TMT-members have the same rank or status within the firm there would be minimum

disparity. One would speak of moderate disparity when managers would differ slightly in rank or status. Maximum disparity is achieved when one member would have a significant higher rank or power than the other TMT-members.

(16)

2.3 Findings on Diversity

In this paragraph the empirical findings on diversity will be discussed. This paragraph will start with more general outcomes on diversity, and then a more in depth review of the three specific measures.

First off according to Barkema & Shvyrkov (2007) the actual distribution of diversity within the group is of grave importance. Having argued that subgroups within a TMT are more likely to be formed when people “match” on several demographic characteristics such as for example; age, tenure, gender, nationality and educational background (Barkema &

Shvyrkov, 2007). So even though there might be some diversity within a team this does not have to result in a positive effect. TMT-members with the same characteristics can cause the creation of a subgroup due to its strong fault lines (Lau & Murnighan, 1998).

Additionally what has been shown is that excessive diversity in TMTs leads to sub-optimal decision making, the highest quality of decision making is achieved with neither too much heterogeneity nor homogeneity (Cox & Blake, 1991). This would suggest the effect of TMT-diversity on performance should have an inverted U-like shape.

As long as TMT-diversity is not dominated by a large amount of characteristics and integration is sustained at an acceptable level, the expectations are that gender and nationality will indeed lead to an increase in performance in both emerging and developed countries.

2.3.1 Gender-diversity

The first measure used to determine TMT-diversity in this thesis is gender diversity. There is no clear position which suggests the effect of gender diversity is positive or negative. This paragraph will summarize previous research on gender-diversity.

For a long time male and female have been seen as distinctly different in both physiology as well as psychology, leading to different behavioral patterns (Betz, O’Connol, & Shepard, 1989). According to the stereotype male and female have very different

characteristics (Buckley, Wiese, & Harvey, 1998). A male for example would be seen as “forceful, assertive, aggressive, persistent and decisive”. Whereas a female would be more “passive, submissive, dependent, emotional and indecisive”

When talking about arguments for gender diversity there are two main lines of

(17)

certain group from the upper echelons (Dalton, Daily,

Johnson, & Ellstrand, 1999). Economic arguments however state that the composition of the TMT determines its processes, which partially determine performance.

The empirical results on the economic perspective of gender diversities in TMTs on performance are mixed. A study which researched the effect of female representation in Fortune 1000 companies on Tobin’s Q (financial measure), found a statistically significant positive relationship (Carter et al., 2003). Additionally several researchers have found that a company with a more diverse and thus more correct representation of the consumer-base has positive effects for performance (Bilimoria & Wheeler, 2000; Burgess & Tharenou, 2002). According to the researchers this would give a positive signal to investors and employees, causing higher loyalty and motivation. Moreover other researchers found gender diversity has positive effects on firm performance(Adler, 2001; Carter et al., 2003; Lückerath-Rovers, 2013).

On the other side of the spectrum, other researches claimed the opposite, that either TMT-diversity has no beneficial effect to firm performance (Carter et al., 2007). Or that it actually has a negative effect (Adams & Ferreira, 2009; Lee & James, 2007). According to Adams & Ferreira (2009) more gender diversity however does lead to more monitoring, as well as higher equity based compensation. Lee & James (2007) found that shareholders react negatively on the appointment of a female executive, resulting in the negative effect.

Additionally, others found that gender diversity’ has no significant effect on firm performance whatsoever (Carter et al., 2010; Erhardt et al., 2003; Rose, 2007; Sun, Liu, & Lan, 2011). Literature shows that the reasons for the negative effects might be to the greater effort and time needed to reach consensus (Finkelstein & Hambrick, 1996; Knight et al., 1999).

(18)

2.3.2 National-diversity

The second measure to determine TMT-diversity is the diversity in nationality. The amount of research on this aspect of TMT-diversity is much more scarce compared to gender. According to Carter et al. (2010) there are two reasons which make it crucial to include

different nationalities. The first reason is somewhat ethical stating that people of a different nationality, when they are competent and qualified they should not be excluded. This article states that they could contribute to the firm due to their network, human capital or other characteristics. The second reason suggests that corporate governance would increase with ethnic diversity of board members, which causes the financial outcomes to be more profitable. Additional research states that national diversity actually brings additional value to the

company (Oxelheim & Randoy, 2003). Carter et al. (2010) used ROA and Tobin’s Q as dependent variables for TMT-diversity, but found no significant results. Even after using a 3SLS regression to remove endogeneity, still no results were found.

Another study by Masulis et al. (2012) researched the effect foreign TMT-members might have on corporate governance and a firms’ performance. They found that a MNC with foreign independent directors (FIDs) make significantly better acquisitions when their targets are from the FIDs home country (Masulis, Wang, & Xie, 2012).

It must be noted however that also in non-takeover situations national diversity has a positive effect. Another study found national diversity to be positively related with firm performance, measured by Tobins Q, on a sample of 1.066 US listed firms (Carter et al., 2003). Additionally it’s not only this specific performance measure that finds these results. Erhard Werbel and Shradex (2003) also found evidence for the positive influence of both gender as well as national diversity on firm performance. In this paper firm performance was measured as return on assets (ROA) and return on equity (ROE) (Erhardt et al., 2003).

Moving on the advisory role of a foreign TMT-member has also been shown to be beneficial for boards (Adams et al., 2010). They showed that MNCs with foreign directors are able to establish better foreign networks, and that they can use their advice for expanding into these countries. The foreign directors’ knowledge about regional regulations, cultural and social norms and the market preferences proved most useful.

(19)

Most but not all studies agreed that national

diversity has a positive effect on performance. Even though not all studies found significant results, they all agree that foreign TMT-members have a beneficial effect for the firm. Either in the form of giving knowledge, providing extra information or helping the MNC to get familiar with foreign markets(Adams et al., 2010; Carter et al., 2010).

2.3.3 Age-diversity

The third and final measure for TMT-diversity is the diversity in age. People in different age groups differentiate from each other in terms of capabilities and characteristics, influencing the way they execute a task, and make their decisions. The differences between these groups can affect firm performance. For Kilduff et al. (2000), who researched several TMT-diversity measures, age was the only one to be found to affect firm performance positively. They reported higher firm performance for firms with greater age diversity

(Kilduff, Angelmar, & Mehra, 2000). Additionally others stated that people from different age groups would be complimentary to each other within tasks, once again causing better

performance (Ilmakunnas & Ilmakunnas, 2011).

Young people for example are generally more creative, flexible, innovative and energetic than older people. Whereas the older age groups are generally more knowledgeable and have more responsibility and commitment (Beaver & Hutchings, 2005). Combining these age groups into working teams have showed to increase firm performance (Li, Chu, Lam, & Liao, 2011). Once again, as also found by Hambrick (1986), they found that age diversity specifically creates more points of view, skills and insights resulting in better decision making and problem solving. It has been found that the assets gained from age diversity are unique and hard to imitate, potentially bringing a sustained competitive advantage to a firm (Li et al., 2011).

Furthermore, age diversity was suggested to be beneficial for target markets due to better understanding the demands of their customers (Morrison, 1992; Richard, 2000). For this reason, age diversity might also be important for emerging markets. However, Richard and Schelor (2002) found that diversity in age is somewhat hard to balance in respect to when it has a positive effect on firm performance. They concluded that for low to medium age diversity the effect on firm performance would be positive and significant. However for high age diversity the effect was negative in regard to firm performance (Richard & Shelor, 2002).

(20)

found that the variation in TMTs is the lowest for age of

directors compared to other demographics (PwC, 2014). This might mean that age-diversity is not yet regarded as a conclusive factor for influencing firm performance.

2.4 Country development

In this research paper a new moderator influencing the relation between TMT-diversity and firm performance is used, country human development. Human development will be used to see if the level of human development of a country has an effect on the relation between the independent and dependent variable. Historically measures of a countries well-being have been mostly focused on economic measures such as the Gross Domestic Product (GDP) or employment rates (Chaaban, Irani, & Khoury, 2015). In a response to this the Human Development Index (HDI) made available yearly by the United Nations Development Program (UNDP) was developed (Nübler, 1995). The UNDP’s first report stated that

‘development is much more than just the expansion of income and wealth’ (Sagar & Najam, 1998).

The HDI consists of three combined measures of a countries well-being/development. These three dimensions are (Nübler, 1995; Sagar & Najam, 1998): longevity, which is the life expectancy when someone is born. Knowledge, measured in the adult literacy rate and the enrollment ratios of the country. And finally the standard of living, measured by the adjusted GDP per capita.

These measurement variables can be made comparable to those of other countries by transforming the values to a scale ranging between 0 and 1 (Nübler, 1995; Sagar & Najam, 1998). One of the shortcoming of this type of measurement was that the arithmetic mean of the normalized values was taken, without taking into account any unbalance within single dimensions (Casadio Tarabusi & Guarini, 2016). In 2010 the UNDP started aggregating by using the geometric mean instead. Which means this new HDI would adjust for unbalances among dimensions.

(21)

HDI has been used by many researchers in different

fields of research. The most prevalent being research related to healthcare and happiness (Hu et al., 2016; Musambira & Matusitz, 2015). Hu et al. (2016) for example found that the mortality-to-incidence ratio (MIR) was significantly lower for countries with higher HDI. Additionally Musambire & Matusitz found that countries with higher HDI overall have higher happiness. Since the HDI consist of life expectancy and several other measures this is no different than expected.

There however is no research using HDI as a moderator in the relation between TMT-diversity and firm performance. Using the HDI as a moderator is used to be able to distinguish high developed from medium developed countries, and additionally to see if the link between the variables is strengthened or weakened by the development level of a country.

2.5 Conceptual model

Based upon the literature this section will focus on the hypotheses that must be constructed to help test the conceptual model below:

TMT Diversity Financial Performance

Country Human

Development

H1: TMT-diversity drives higher financial performance and this relation is positively

(22)

3. Methodology

This section will consist of the methodology used for this thesis. This thesis will deal with a causal relationship and the concept that would affect this causality. For this reason the research has a deductive approach seeing that the hypotheses were developed referring to existing theories and previous studies. A deductive approach leads to a quantitative method to test the hypotheses. This will lead to a proposition for the actual statistical methods that will be used for the data analysis.

3.1 Data Sources and Collection

Secondary data will be used for this thesis. The choice for using secondary literature has several advantages. The first being that gathering secondary data yourself can be very costly and time consuming, which both are limiting factors for this master Thesis. While additionally the data from a secondary source can be vastly superior to the data one could have found and obtained by himself.

The demographic data necessary for determining TMT-diversity has already been made available in the BoardEx database, and thus the relevant data will be gathered from this database. BoardEx is a database with information on demographic characteristics of board members and senior executive around the world (BoardEx, 2017). This database has been globally acknowledged and used which makes it appropriate to use this source. The

information used in the BoardEx database is collected from reliable sources and verified true multiple other sources before its put in the database (BoardEx, 2017). One of the main advantages of this database is that it not only consists of data of U.S. based or Western European countries, but also most other countries in the world. For this reason this database will be the most useful for emerging countries.

Additionally, the data needed for determining the ROA of the firms has to be collected. ROA for three separate years will be used as the indicator for firm financial performance. This information will be collected from the Orbis database, which collects data of more than 175 million companies worldwide (Dijk, 2017).

(23)

3.2 Data description

The sample is comprised of firms from two countries of very high human development and three countries of medium human development during the period 2013-2015.

The dataset consists of 102 companies for a total of 306 observations over the three year period. Companies are removed from the sample when they were one of the outliers in the sample due to having extreme values for the dependent variable. Other companies that are left out were often left out for missing information on company executives. A company is

removed when 20% or more of the information on a particular diversity measure of the TMT-members is missing. An assumption based on a sensitivity analysis has been made that 80% of the data will still give a good indication of the specific TMT-diversity measure that has missing data.

An example of this is when a TMT would have five members of which of four the nationality is known. The Blau index for age will then be based on four of the five members and thus 80% of the data. This could cause a deviation of on average 0.178. However since total diversity is measured as the weighted average of all three diversity measures the actual effect will be much smaller. Additionally, only about 5% of all companies included had less than 100% availability for the data of TMT-characteristics.

This study only includes executive directors, because in the end these directors take the final decisions, and determine corporate strategy. While non-executive directors are primarily responsible for monitoring and advising the TMT, even though the role of a non-executive director must not be underestimated. However, several papers have shown that

(24)

executive directors have a larger impact on the actual actions of the firm than non-executive directors.

To compare companies with very high human development to companies with medium human development, countries from these categories have been chosen from the HDR. The choice for comparing medium with very high has been made since for countries with even lower development no complete data could be found. Additionally to create a comparable sample three countries of medium development were needed and only two for very high development.

Finally the choice for a three-year period was made to measure how these companies developed in their TMT-diversity measures over these three years. Another reason was to increase the amount of observations and thus increase robustness and reliability of the results.

3.3 Variables

3.3.1 Dependent Variables

The dependent variable in this research is firm performance, which is often defined specifically as firm financial performance (Certo, Lester, Dalton, & Dalton, 2006). Over all research on firm financial performance many measures have been used, however no set of financial ratios can be named as the “ultimate” predictors for performance (Otley, 2007). However the most papers found seem to prefer ROA, Tobins Q or Return on Equity (ROE).

For this thesis one measures of firm performance will be used to check if TMT-diversity has an effect on it: ROA This measures profitability of a company and has seen widespread use.

ROA is the ratio of net profits compared to the total assets of a company (Erhardt et

al., 2003). This ratio gives an idea on how well the total assets a company has and how efficiently these assets were used to generate profits. This ratio is calculated by dividing the net profits by the end of year book value of all assets (Adams, Almeida, & Ferreira, 2005). The higher the ROA means the company performed better.

3.3.2 Independent Variables

(25)

collect (Carter et al., 2003). These measures will all be

converted into a Blau-index, which will be explained further in paragraph 3.4 Methodology. As also discussed earlier, diversity can be measured in several ways, either in variety, separation or disparity. This research will try to focus on diversity as variety. However since these types co-exist it might be hard to specifically focus om this type (Klein & Harrison, 2007). Differences along the vertical and horizontal axes are measured by separation and disparity, whereas variety focusses on within-unit diversity of data (Boone & Hendriks, 2008; Nüesch, 2009).

The collected data on TMT-members for all companies will be coded to be able to calculate the Blau-index for the different diversity measures. These measures will then be normalized and made in to an equally weighed average index for diversity of a specific company for every year of the 3-year period. The way the data was coded to calculate the different Blau-index can be found in Appendix 1: Codes for developing Blau-index. Finally, all independent variables have a 1-year lag to avoid endogeneity and reverse causality.

3.3.3 Control Variables

The control variables used for this study have been based on other studies focusing on the same kind of variables, and that have been shown to have a significant relationship with the dependent variables. The control variables are used to make sure the actual link between the independent and depend variable will be found. This study will make use of three different control variables.

Board size has been controlled for by many previous researchers. The main reason for this being that when the board gets larger information processing also improves (Hambrick & D’Aveni, 1992). Board size on itself might already have an effect on a company’s

(26)

make their boards a representation of society (Marinova,

Plantenga, & Remery, 2015). This variable will be measured by total amount of employees. Tenure, or the amount of time the TMT-members have been the firm. This measure will be used because members with more experience in the firm have been shown to have a significant effect on a firm’s performance.

3.4 Methods

One of the most commonly used measures to use when talking about diversity is the Blau-index. This index invented by Blau (1977) has been often named as the optimal measure to capture variation within a group. This index can be calculated with the help of the

following formula:

Where variety can take k=1, . . . , K possible categories, and p stands for the proportion of board members in the kth category. An increase in board diversity equals in an increase in Blau’s Index.

The minimum score for Blau’s Index is always equal to zero; additionally the maximum score depends on the amount of categories within a variable. To calculate the theoretical maximum the following formula can be used: ( K - 1 ) / K, where K is the amount of categories within a measure. The theoretical maximum for each variable can be found in the table below:

As shown in the table, as the amount of categories within a variable increases the theoretical maximum also increases. However to incorporate the three measures into an average they have to have to be corrected for the amount of categories. This can be done by multiplying the Blau Index scores by K / ( K – 1 ). This will create a standard range from zero to unity for every diversity measure.

For this paper the full equation including the Blau Index is as follows:

Firm Performance = Constant + b1*BlauIndexGender + b2*BlauIndexNationality +

b3*BlauIndexAge + b4*FirmSize + b5*AverageTenure + b6*BoardSize + e

Table 2: Number of categories and their theoretical maximum for diversity

Formula 1: Blau Index (Blau,

(27)

Several regression methods have been used by

previous researchers, most importantly the Two-Stage-Least-Squares (2SLS) and Ordinary Least Squares (OLS) regression method. A 2SLS is used when endogeneity between variables is expected; when this is not the case an OLS is more efficient than a 2SLS. Depending on the existence of endogeneity 2SLS or OLS will be used.

3.4.1 Multicollinearity

An issue that might occur when doing a regression analysis is the issue of multicollinearity. This problem occurs when variables within a multiple regression are strongly related to each other (Zikmund, Carr, & Griffin, 2013). When multicollinearity is found this means the effect of the dependent variables will be arbitrary (Grewal, Cote, & Baumgartner, 2004). In a perfect situation none of the independent variables correlate with each other, while having a strong correlation with the dependent variable. Standard errors will be affected by multicollinearity, meaning that variables will be significant while they should not be. Detecting multicollinearity is most easily done by looking at the ‘Variance Inflation Factor’ (VIF). This measures how much the variance of the estimated coefficient is increased. When a VIF is equal to 1 this means that there is no correlation between the independent variables. Different researchers name different levels of VIF as a maximum. Values ranging from 10 (Kennedy, 1992; Nachtsheim, Neter, & Kutner, 2004; Neter, Wasserman, & Kutner, 1989), to lower than 5(Rogerson, 2001), or lower than 4 (Pan & Jackson, 2008) have all been named. Whereas often higher than 3.5 would mean ‘something’ is wrong 5 means ‘little’ is wrong and higher than 10 means there is a definite case of multicollinearity (Neter et al., 1989). The results from this regression are even lower than the lowest value; this means there is no multicollinearity. See Appendix 2: Multicollinearity Test for the results.

3.4.2 Endogeneity

One would speak of endogeneity (or autocorrelation) when there is autocorrelation of the error terms between the independent and the dependent variables. Autocorrelation can be measured with the help of a Durbin-Watson test, the results of a Durbin-Watson test vary between 0 – 4. Where a score of 2 would mean there is no autocorrelation whatsoever, 0 means there is positive autocorrelation, and values of 4 or close to 4 imply negative

autocorrelation (Montgomery, Peck, & Vining, 2015). The results of this test showed that for this regression the Durbin-Watson score was 1.789, meaning there is little to no

(28)

4. Results

This section will consist of the results of the empirical research. First of all descriptive statistics on TMT-diversity and firm performance will be given. After which a comparison will be made between high and medium developed countries. Last the hypothesis will be confirmed or rejected with the help of the OLS regression analysis.

4.1 Development of TMT-diversity

This section will describe the different diversity variables and their development over the 3-year period for all five countries. First of all the average total diversity for both very high developed countries and medium developed countries show an increase over the 3-year period. Total diversity increased from .344 to .369 to .383 on average for high developed countries, and increased from .297 to .313 to .326 for medium developed countries. During the 3-year period almost all separate variables for both types of countries increased. The only exception of this being age diversity for the medium developed countries, which has shown a decrease yearly from .581 to .571 to a .566 in 2015.

Another interesting fact that can be observed is the higher diversity of gender in medium developed countries compared to high developed countries. Previous research often found the opposite. A reason for this might be that in western companies the data is more readily available and therefore gives a more realistic image. Whereas for the medium

countries only the more known companies have complete data on TMT-team characteristics. Possibly providing additional evidence for the ethical arguments for choosing a more

(29)

TMT-diversity looks different however when looking at each country separately (see figure 4).

Figure 4: Development of TMT-diversity for separate countries during 2013-2015

As can be derived from the above figure, all countries except India have a trend towards having a more diversified TMT. The discussion on board-diversity has been a subject of hot debate for several years now. And the effect of this, next to the possible effect of the research on this subject can be seen on the development of TMT-diversity. What must be stated however that no real trend can be captured in a sample of just three years; it does however show a possible trend. This trend has been confirmed for gender diversity by other researchers. They found that for European countries between 2001 and 2007 gender-diversity increased 1.1% every two years (Heidrick & Struggles, 2007). Additionally the amount of companies with at least one female director in the board increased from 53% in 1990 to 87% in 1999 for Fortune 100 firms (Farrell & Hersch, 2005). This increase however is mainly based on non-executives, whereas this sample only uses executive directors. On the other hand it adds to the literature that the same trend can be seen for executive directors.

(30)

Figure 5 below shows the development of all

diversity variables for all countries over the 3 year period. Some additional interesting things can be seen here. For example nationality diversity for the medium developed countries is pretty much nonexistent for India and The Philippines. Additionally Germany is a country with low diversification of gender, almost just as low as India. Another interesting fact is that South-Africa wins the race for the highest diversity purely by diversifying way more than other countries on gender. A common factor among all countries is the relative high amount of age diversity compared to the other diversity variables.

4.2 Descriptive statistics ROA/HDI and control variables

After discussing the development of the different diversity variables a short summary of the variables ROA and HDI and the control variables will be included. What stands out is the almost double average ROA for medium developed countries. Other than that, the high developed countries have

bigger company size, and a slightly bigger total size of the boards.

Figure 5: Buildup of TMT-diversity per variable per country per year.

(31)

4.2 Correlations

Another important measure to show how variables are linked to each other is a

correlation table as shown in figure 9. This table shows the correlations that exist between the different variables. One would speak of no effect with values between 0-0.1, a small effect from 0.1-0.3, a medium effect for 0.3-0.5 and an effect higher than 0.5 shows a large effect (Field, 2009). When correlation exceeds .9 it becomes harmful for an analysis (Tabachnick & Fidell, 1996).

The results from the correlation analysis show that there aren’t many highly correlated variables. Of course Total Diversity is strongly correlated with its appropriate Blau Indexes. But additionally it is medium positively correlated with two of the three control variables, company size, and total directors on the board. That shows that bigger companies and companies with more directors have significantly more Total Diversity. Additionally the moderator the HDI is slightly positively correlated with Total diversity and slightly negatively with ROA.

4.3 Regression results

As discussed earlier a 2SLS is used when there are endogeneity issues, and an OLS is used when this is not the case. Since there are no endogeneity issues an OLS will be used. The results of the OLS can be seen in figure 10. A stepwise model will be used to test the dataset, this shows how the independent and control variables relate to the dependent variable. In both separate groups as well as bigger groups. In the ANOVA table Appendix 4: ANOVA for

several models it can be seen that two out of three created independent variable groups are

insignificant. Only when adding the moderating variable the regression becomes significant.

(32)

Just the control variables are effective for 1% of the change in the dependent variable, adding the

independent variables combined in Total Diversity are effective for 1.8%. And finally, adding the moderator HDI makes the model effective for 6.7%.

The OLS of model 3 is shown in figure 11 on the side. No significant influence of the independent variable was found on the dependent variable. Additionally none of the control variables seem to influence the dependent variable either. Finally the moderator does seem to influence the dependent variable significantly and negatively with a Beta of -.243. Meaning that having a higher HDI has a significant negative influence on the ROA for the companies in that country. A 1% increase in HDI decreases ROA by 0.00243%.

Since Total Diversity had no effect whatsoever on the sample as a whole a more specific analysis was made where an analysis was done on each country separately. All measures taken to prevent endogeneity and multicollinearity were also taken for these

analyses. This gave some interesting results for Germany specifically which were significant. The moderating variable was excluded here since there is no comparison between countries with a different HDI. In Appendix 5: Regression Results Germanyit can be seen that the total R2 is at 20,8% for the dependent variable. And additionally the results for Total diversity are significant and positive. Showing that when Total Diversity increases by 1%, ROA will increase by 0.00334.

Table 5: Model Summary

(33)

5. Conclusions

Most of the previous empirical research had different results for the effect of TMT-diversity on firm performance. This paper looked at this relation by combining the Blau-scores for age-diversity, gender-diversity and nationality-diversity into one single diversity measure. The dependent variable for firm performance was ROA. Additionally a new moderating variable was included to see this affected this relation, this moderator being the HDI.

Additionally this study follows recent trends in the research by continuing the search for the ultimate moderator in the relation between diversity and firm performance. Moving on it combines two fields of research by using both medium and highly developed countries from several parts of the world.

The analysis was done on 102 companies from 5 different countries, 2 high developed countries; The Netherlands and Germany, and 3 medium developed countries; India, South Africa and the Philippines for the three year period 2013-2015. The research controlled for three variables used by many previous researches, including; average tenure, board size and firm size. The regression analysis was not significant for analyzing this relation until adding the moderating variable, showing this variable does have a strong influence on the dependent variable. It however cannot be shown if the strength of the relation between the independent end dependent variable is influenced by the moderator. Moving on to the hypothesis:

H1: TMT-diversity drives higher financial performance and this relation is influenced by country development

Due to not having a significant effect of the independent variable on the dependent variable the hypothesis can neither be rejected or confirmed. What has been shown however is that when only looking at Germany the effect of the independent variable on the dependent variable is positive and significant. However when looking at a single country the mediator could not be included. The results being significant and positive for a western country has strong overlap with several other papers (Bilimoria & Wheeler, 2000; Burgess & Tharenou, 2002; Morrison, 1992; Richard, 2000). Additionally, not finding a significant relation also happened several times before (Adams et al., 2010; Carter et al., 2010; Erhardt et al., 2003; PwC, 2014).

(34)

provide evidence of this being due to the positive effects of

diversity on firm performance. It might be that the pressure to become more of a reflection of society is causing them to do this.

Moreover the results show that different countries diversify on different aspects of diversity. Where India and the Philippines have almost no diversification on gender, South Africa has the highest score on this measure. Additionally, for all countries more than half of the diversity came from the measure diversity. Wat must be noted however is that age-diversity historically had the least positive effects on firm performance.

5.1 Managerial Implications

Because the initial possibilities were strongly correlated with some sort of outcome, it is hard to find any managerial implications for a research which had no significant outcomes. However it makes sense to companies to set targets for TMT-diversity in regard to what is realistic and taking into account the needs of the company. First of all they should take into account the critical mass theory, stating that only at a certain point of diversity the diversity could become useful. Additionally even though the effect of diversity on firm performance for just Germany was positive, this sample was so small that this should not be seen as a definite signal to increase diversity for performance reasons.

This does however give additional information as to how the market is currently situated in several countries which have had very little attention in previous research. For the India market you can distinguish yourself from your competitors by focusing on gender-diversity. In previous research Gender-diversity has often proven to have a positive effect on firm performance. Since nearly no companies in these countries have gender-diversity this is a way to outperform and distinguish yourself from your competitors. What must be kept in mind however is the previously mentioned critical mass theory, stating that something only has a positive effect when a certain critical mass has been achieved.

(35)

5.2 Limitations and Future Research

Like any research paper this thesis too has a few limitations. First of all the

information on companies from medium developed countries is very incomplete. Leaving only very few companies to include in the sample, it makes common sense that specifically the more diversity companies would report their diversity. This means that the companies were left in final sample, already were the most diversified. This could have caused a biased sample, resulting in outcomes that would not be true for the country as a whole. For higher developed countries the information on all companies is much more complete, with reporting on gender, age, nationality and several other characteristics of board members even being law in the Netherlands.

Moving on this paper only focused on demographic diversity of a TMT-team. Other characteristics of board members, including education, language, knowledge and skill-diversity were not included. These characteristics might also have an effect on firm performance, but these were left out due to the fact that this information is hard to obtain.

Future researchers should include more measures on diversity to make a more

complete measure of total diversity to link with firm performance. For repetition of a similar paper the sample size for each of the countries should also be increased. Additionally using several measures for firm performance instead of one might also have a beneficial and more complete effect.

Another limitation on this study is the sample size, even though 102 companies is a reasonable sample size. Since these samples are then split into medium and highly developed countries the actual sample size per subsample is much smaller. Where the literature

prescribes a higher sample size to effectively measure a group.

Additionally something which might be interesting for future researchers is researching the critical-mass-theory further in this field of research. There would be large managerial implications if it could be found what the critical mass would be for certain measures, and if this theory actually affects all diversity measures.

Finally, to find real trends, future research should also focus on longer time-spans for their sample. What however must be kept in mind is to not go overboard, since it was already noted in this thesis that increasing the amount of years has a big effect on the amount of companies available. Due to them not yet being active in these years, or missing data for just one of the total years.

(36)

6. Discussion

The aim of this thesis was to research the effect of TMT-diversity on firm performance and if this relation was affected by country development. By doing this the idea was to expand on the body of literature by helping find the ultimate moderator. Additionally by

incorporating the diversity measures into one overall measure for diversity we hoped to rid the inconsequent results in the field.

This study found that for the entire sample there were no significant results for diversity on firm performance. This is in concordance with results found by several other researchers (Carter et al., 2010; Erhardt et al., 2003; PwC, 2014; Rose, 2007). However looking back there are several things which might have caused the results to be insignificant. First of all the different types of diversity discussed in the literature review (Klein & Harrison, 2007). It could be that using age, nationality and gender somehow include diversity in

separation and disparity instead of variety. The incorporation of these measures into a single overall diversity measure might have caused the insignificance. However this would need to be researched further to see which demographic would capture variety in the best possible way.

Additionally the choice to incorporate the three measures into a single “Total

Diversity” measure removed the option to see what a single diversity measure might do to the dependent variable. The total explanatory power of the model also shows that these three measures only capture a very low % of the variation in firm performance. Adding additional measures might increase the explanatory power, even though for this research more variables were hard to add. This thesis failed to incorporate more measures due to the low availability of other diversity variables.

Moving on the sample size was hugely affected by missing data, either in the variables or non-existence of firms during certain years. This caused the sample to become too small to have actual meaning for the world market or even specific countries.

(37)

After finding no significant results in the overall

analysis several sub-analyses were made to see if for these samples there would be a significant result. After conducting all sub-analyses on every country only one country was found to have a significant relation between the independent and dependent variable. Positive results for western countries are in concordance with several other papers (Burgess &

Tharenou, 2002; Carter et al., 2003; Erhardt et al., 2003). However for The Netherlands individually the results were once again non-significant.

Something else that might have affected the results for the “Total Diversity” measure is the relatively large size of age-diversity within the “Total Diversity” measure. As

discussed in the literature review, scholars have found the least evidence for the positive effect of age-diversity on firm performance (Richard & Shelor, 2002). This measure might have affected the results for “Total Diversity” especially since it was such a big part of the final measure.

Additionally where a recent study found that there isn’t much diversification on age, gender and ethnicity in developing countries, the results from this paper do not agree. Apart from India, the lower developed countries do not lag too far behind the high developed countries. In fact, South-Africa has the highest diversity of all countries in the sample. However what once again must be stated is that the sample for specific countries was very small. And the information on TMT was scarce, resulting in a possible bias of the results.

Another factor that might have been of influence on the results being insignificant is the context and Industry researched in this paper. Whereas other researchers categorized their sample into different industries, and excluded some industries, I did not due to time

constraints. This might have caused a bias due to overrepresentation of certain industries which have values that differ significantly from other industries.

Subsequently something that I already discussed in the results was the lack of

information on companies in India, South-Africa and The Philippines. There were only a few companies in these countries that had enough information on TMT-characteristics. It is possible that only companies that have relatively high diversity in their teams report their characteristics. This would mean that the results in this thesis could be biased towards the more diversified companies within a country.

(38)

should be women to reach positive results. Due to this the

(39)

7. Appendixes

Appendix 1: Codes for developing Blau-index

Codebook

Variable Code Intepretation

Year year

Company name company ID-nr id-nr Director name director

(40)

Appendix 2: Multicollinearity statistics Coefficientsa Model Unstandardized Coefficients Standardize d Coefficients t Sig. Collinearity Statistics

B Std. Error Beta Tolerance VIF

1 (Constant) 11,573 2,153 5,375 ,000 Average Time in company ,040 ,051 ,046 ,784 ,433 ,936 1,068 Number of Employees 4,056E-6 ,000 ,053 ,754 ,451 ,632 1,583 Total Diversity -2,907 2,119 -,085 -1,372 ,171 ,831 1,203 Total Directors ,092 ,081 ,086 1,135 ,257 ,545 1,836 Human Development Index -10,139 2,562 -,243 -3,957 ,000 ,840 1,191

a. Dependent Variable: ROA using Net income

Appendix 3: Durbin- Watson Score

Model Summaryb Model R R Square Adjusted R Square Std. Error of the Estimate Durbin-Watson 1 ,259a ,067 ,051 5,87482 1,789

(41)

Appendix 4: ANOVA for several models ANOVAa Model Sum of Squares df Mean Square F Sig. 1 Regression 108,694 3 36,231 ,996 ,395b Residual _10806,604 ₂₉₇ _36,386 Total _10915,297 ₃₀₀ 2 Regression 193,281 4 48,320 1,334 ,257c Residual _10722,016 ₂₉₆ _36,223 Total _10915,297 ₃₀₀ 3 Regression 733,813 5 146,763 4,252 ,001d Residual _10181,485 ₂₉₅ _34,514 Total _10915,297 ₃₀₀

a. Dependent Variable: ROA using Net income

b. Predictors: (Constant), Company Size, Average Tenure, Total Directors c. Predictors: (Constant), Company Size, Average Tenure, Total Directors, Total Diversity

d. Predictors: (Constant), Company Size, Average Tenure, Total Directors, Total Diversity, HDI

(42)

8. References

Acar, F. P. (2015). The effects of top management team composition on SME export performance: an upper echelons perspective. Central European Journal of Operations

Research, 24(4), 833–852. http://doi.org/10.1007/s10100-015-0408-5

Adams, R. B., & Ferreira, D. (2009). Women in the boardroom and their impact on governance and performance. Journal of Financial Economics, 94(2), 291–309. http://doi.org/10.1016/j.jfineco.2008.10.007

Adams, R. B., Hermalin, B. E., & Weisbach, M. S. (2010). The Role of Boards of Directors in Corporate Governance: A Conceptual Framework and Survey. Journal of Economic

Literature, 48(1), 58–107. http://doi.org/10.1257/jel.48.1.58

Adams, Almeida, H., & Ferreira, D. (2005). Powerful CEOs and their impact on corporate performance. Review of Financial Studies, 18(4), 1403–1432.

http://doi.org/10.1093/rfs/hhi030

Adler, R. D. (2001). Women in the executive suite correlate to high profits. Harvard Business

Review, 79(3), 1–8.

Ararat, M., Aksu, M., & Tansel Cetin, A. (2015). How board diversity affects firm

performance in emerging markets: Evidence on channels in controlled firms. Corporate

Governance (Oxford), 23(2), 83–103. http://doi.org/10.1111/corg.12103

Arfken, D. E., Bellar, S. L., & Helms, M. M. (2004). The ultimate glass ceiling the presence of revisited: Women boards. Journal of Business Ethics, 50(2), 177–186.

http://doi.org/10.1023/B:BUSI.0000022125.95758.98

Bantel, K. A., & Jackson, S. E. (1989). Top Management and Innovations in Banking : Does the Composition of the Top Team Make a Difference ? Strategic Management Journal,

10(May), 107–124. http://doi.org/10.1002/smj.4250100709

Barkema, H. G., & Shvyrkov, O. (2007). Does top management team diversity promote or hamper foreign expansion? Strategic Management Journal, 28(7), 663–680.

http://doi.org/10.1002/smj.604

Beaver, G., & Hutchings, K. (2005). Training and developing an age diverse workforce in SMEs. Education and Training, 47(8/9), 592–604.

(43)

Bertrand, M., Black, S. E., Jensen, S., & Lleras-Muney, A.

(2014). Breaking the Glass Ceiling? The Effect of Board Quotas on Female Labor Market Outcomes in Norway. SSRN Electronic Journal, (August).

http://doi.org/10.2139/ssrn.2488955

Betz, M., O’Connol, L., & Shepard, J. M. (1989). Gender Differences in Proclivity for Unethical Behavior. Journal of Business Ethics, 8(5), 321–324.

Bilimoria, D., & Wheeler, J. V. (2000). Women corporate directors: Current research and

future directions. London: Sage.

Blau, P. (1977). Inequality and Heterogeneity. New York: The Free Press.

BoardEx. (2017). BoardEx. Retrieved March 3, 2017, from http://corp.boardex.com/data/ Boeker, W. (1997). Strategic change: The influence of managerial characteristics and

organizational growth. Academy of Management Journal, 40(1), 152–170. http://doi.org/10.2307/257024

Bøhren, Ø., & Strøm, R. Ø. (2007). Aligned, Informed, and Decisive: Characteristics of Value-Creating Boards. SSRN Electronic Journal, (February 2007).

http://doi.org/10.2139/ssrn.966407

Boone, C., & Hendriks, W. (2008). Top Management Team Diversity and Firm Performance: Moderators of Functional-Background and Locus-of-Control Diversity. Management

Science, 55(2), 165–180. http://doi.org/10.1287/mnsc.1080.0899

Brammer, S., Millington, A., & Pavelin, S. (2007). Gender and ethnic diversity among UK corporate boards. Corporate Governance, 15(2), 393–403. http://doi.org/10.1111/j.1467-8683.2007.00569.x

Buckley, M. R., Wiese, D. S., & Harvey, M. G. (1998). An Investigation Into the Dimensions of Unethical Behavior. Journal of Education for Business, 73(5), 284–290.

http://doi.org/10.1080/08832329809601646

Burgess, Z., & Tharenou, P. (2002). Women Board Directors : Characteristics of the Few Board Directors : of the Few Characteristics panies. Journal of Business Ethics, 37(1), 39–49. http://doi.org/10.1023/A:1014726001155

Campbell, K., & Mínguez-Vera, A. (2008). Gender Diversity and Firm Financial in the Boardroom Performance. Journal of Business Ethics, 83(3), 435–451.