The influence of a firm’s cultural background on the effects of
cultural diversity within boards on firm performance
by
Maik Bokhove
MSc. International Financial Management
MSc. International Business & Management
June 9th, 2017 Supervisor: C.R.M. Hermes Co-assessor: H.J. Drogendijk Berkenlaan 2/41 9741 JP Groningen 0031 6 13 79 32 65 maikbokhove@concepts.nl student number: 2344726 Abstract
This thesis examines the moderating effect of firm cultural context on the relationship between cultural diversity within boards and firm performance. Feasible generalized least squares (FGLS) regressions, based on a sample of 319 firms from ten countries and 1,463 firm-year observations in the period 2011-2015, provide evidence for the significant negative relationship between cultural diversity within boards and firm performance. No evidence is found for the moderating effect of firm cultural context.
Table of content
1. Introduction ... 3
2. Literature Review ... 6
2.1 Diversity within boards ... 7
2.2 Board roles... 9
2.3 National cultural context ... 9
2.4 Hofstede’s cultural framework ... 10
2.5 Hypotheses ... 12
2.6 Conceptual model ... 15
3. Data and methodology ... 16
3.1 Sample selection and data collection ... 16
3.2 Independent variable – cultural diversity within boards ... 18
3.3 Dependent variable – ROA and Tobin’s q ... 20
3.4 Moderator variable – firm cultural context ... 21
3.5 Control variables ... 21
3.6 Methodology ... 22
4. Results ... 25
4.1 Descriptive statistics ... 25
4.2 The influence of diversity within boards on firm performance ... 29
4.3 The influence of firm cultural context ... 30
4.4 Further analysis ... 34
5. Conclusion ... 35
References ... 37
1. Introduction
The world has been globalizing during the past decades, caused by firms searching for more effective and efficient production opportunities and exploring and expanding to new markets (Ahsan and Sinha, 2016). In other words, they seek for cost efficiencies and new customers. To deal with the challenges that come with globalization and firm presence in a foreign environment, firms aim to acquire specific (human) resources. A trend that is visible is the increase of foreign nationals in corporate boards (Frijns, Dodd and Cimerova, 2016). The foreign directors possess specific expertise and in a supervisory role they are able to monitor if the executive board’s decisions will increase firm performance (abroad).
Furthermore, they can give their advice and opinion on policy development and by doing so support the executive board in the decision-making process.
A significant amount of research has been done studying the various effects of cultural differences and the subsequent impact that these differences have on a firm’s financial
performance. Cultural differences are generally considered challenging, since they are causing both positive and negative effects (see, e.g, Stahl, Miska, Lee and De Luque, 2017; Steigner and Sutton, 2011; Antoine and Kleiner, 2015). The diversity that comes with cultural differences can add value through new perspectives and an increased amount of knowledge about foreign environments. This knowledge is a valuable resource that decreases
environmental uncertainty and makes it easier to connect to a foreign environment (Lynall, Golden and Hillman, 2003). However, it might also lead to communication problems and coordination issues. The previous examples are related to differences between groups and how to make use of the advantages and overcome the associated issues. This thesis will not focus on the effects of diversity between groups, but on the effects of diversity within groups. A group consists of individuals with different cultural backgrounds and/or knowledge traits. These traits might complement each other and lead to improved outcomes, but it is also possible that the different backgrounds lead to conflicts and suboptimal outcomes. Previous studies found that national culture affects financial decision-making in firms and explains the cross-cultural variation in firm practices. The impact of diversity within boards on firm performance is therefore likely to be different in distinct countries due to the particular national cultural traits. This thesis includes firms from ten different countries and investigates both the overall effect of cultural diversity within boards on firm performance and the
An increasing number of foreign nationals in boards automatically leads to an overall increase of cultural diversity in the boards. Since boards have a critical role within the firm it is important that the directors collaborate effectively and efficiently. In their role, the non-executives are guarding the long-term continuity of the firm. Milliken and Martins (1996) already recognize cultural diversity as a ‘double-edged’ sword. Frijns et al.’s (2016) findings show that cultural diversity within boards leads to a decrease in firm performance. This means that the negative effects of diversity outweigh the positive aspects. These findings are
contradicting previous findings of Baker and Anderson (2010), who state that firms choose their board members for their characteristics, mainly to deal with the external environment. This implies that specific knowledge is needed and cultural diversity helps in dealing with the challenges that come with an international environment. Other studies on different types of diversity such as gender diversity, director independence and previous experience did not find conclusive results either. The discussion on the effects of board diversity is therefore ongoing.
An aspect that is not investigated related to the effects of cultural diversity within boards on firm performance is the influence of the firm’s cultural background. Previous studies considered the attitude towards (non-)executives with specific characteristics in different contexts, both from other managers and from other employees. An example is the attitude towards female managers (e.g., Elsesser and Lever, 2011; Tomkiewicz, Frankel and Adeyemi-Bello, 2004). These studies indicate that female managers must overcome more negative attitudes in one environmental context compared to another. This thesis takes a similar approach by analyzing whether the impact of cultural diversity within boards is different across distinct contexts, where the context is measured in terms of culture.
The three major components in this thesis are subsequently cultural diversity within boards, firm performance and firm cultural context. For both cultural diversity within boards and firm cultural context the Hofstede (2001) dimensions are used. To determine cultural diversity within boards a novel metric from Frijns et al. (2016) is applied that is based on the cultural distance formula from Kogut and Singh (1988). Firm context is measured with the scores on the four original dimensions in the Hofstede framework. Lastly, firm performance is measured via return on assets (ROA) and Tobin’s q. Next to these, several control variables are included. These are firm size, leverage, the degree of foreignness, board size, inflation and GDP growth.
diversity within boards in one context, this thesis intends to expand towards a multi-country analysis. Therefore, the research question of this thesis is:
Does a firm’s cultural context influence the effects of cultural diversity within boards on firm performance?
Where previous studies in international business and corporate finance focused solely on either cultural diversity within boards or the effects of cultural differences/distance across countries, this thesis is, to the best of my knowledge, unique in linking these two factors. It is exploring the effects of culture on firm performance by including different levels of analysis. Its first intention is to extend the research done by Frijns et al. (2016). This particular study included firms that are solely from the UK and including firms from a variety of countries will increase the generalizability of the overall relationship between cultural diversity within boards and firm performance. Second, this thesis explores if a similar board composition has different impacts across contexts (i.e. countries). Adding the Hofstede dimensions as
moderators explores possible effects of differences between national cultures on the
relationship between cultural diversity within boards and firm performance. By doing so, the assumptions that cultural diversity is perceived differently and the attitude towards foreign nationals is different across cultural contexts are researched. All in all, the effect of cultural diversity within groups across cultural contexts is investigated.
The structure of this thesis is as follows: first, relevant theories are discussed and hypotheses are formulated. Next, the data collection methods, sample selection and
methodology are explained. Subsequently, the results are provided and discussed. Finally, the conclusion section summarizes the findings and presents limitations and related suggestions for future research.
2. Literature Review
Diversity within boards has become increasingly important as a governance issue over the past years and many studies have been done regarding the effects of board diversity on firm performance. Multiple authors focused on the effects of gender diversity (see, e.g., Adams and Funk, 2012; Adams and Ferreira, 2009; Nguyen, Locke and Reddy, 2015; Arfken, Bellar and Helms, 2004) and director independence (see, e.g., Luan and Tang, 2007; Bradley and Chen, 2015; Cavaco, Challe, Crifo and Rebérioux, 2016), but also director ethnicity (see, e.g., Ujunwa, Okoyeuzu and Nwakoby, 2012; Gul, Munir and Zhang, 2016), education (see, e.g., Haynes and Hillman, 2010) and experience and tenure (see, e.g., Kong-Hee Kim and
Rasheed, 2014; Cho and Hambrick, 2006; Thorssel and Isaksson, 2014) have been investigated before.
In this line of research, culture has not been extensively researched as a factor of diversity. However, cultural diversity might be particularly relevant within boards, because culture is an important factor in decision-making. Previous studies that included culture primarily focused on cultural differences between groups. This makes sense, because in many definitions culture is described as a concept that distinguishes groups (i.e. teams, firms, nationalities, etc.) from one another (see, e.g., Hofstede, 2001; Schwartz, 1994). According to Schwartz (1994) members of a culture share a rather similar view on what they find appropriate and/or desirable. House (2004) states that these views are reflected in individuals.
The concept of cultural differences is recognized as a double-edged sword (Milliken and Martins, 1996), which means that there are both positive and negative effects related to cultural differences. In the negative perspective cultural differences are often seen as a liability, because differences cause frictions and conflicts (Stahl et al., 2017). These
environment (Zaheer, 1995). One important aspect of this foreign environment is culture. In order to overcome the differences, also called the liability of foreignness, firms invest in (human) resources to familiarize themselves with the foreign culture.
Cultural differences also have positive aspects. Stahl et al. (2017) argue that the combination of multiple cultures leads to the development of strategic capabilities, better decisions on foreign direct investments and synergies in mergers and acquisitions, because more knowledge is available in the decision-making process. In other words, by combining the best aspects of multiple cultural perspectives through knowledge-sharing, improved decision-making arises with its subsequent improved results.
Besides cultural differences between groups, there are also cultural differences within groups. This means that a group consists of individuals with different cultural backgrounds. It is possible that there is only one individual with a deviant background, but it is also possible that all individuals within the group have distinct backgrounds. Frijns et al. (2016) make the step from the impact of cultural differences between groups to the effects of cultural diversity within groups. The focus is no longer on how a firm from one culture (e.g. Germany) can overcome cultural differences to become successful in another culture (e.g. China). The focus shifts to how cultural differences within one firm affect performance. Frijns et al. (2016) do so in a similar way as others did for other forms of diversity within groups (e.g. gender diversity). This means that they measure the diversity within the board and investigate whether there is a relationship with firm performance. They acknowledge that cultural diversity is this double-edged sword and thus impacts performance in multiple ways.
2.1 Diversity within boards
The degree of task-related and relationship conflict is representing the positive and negative effects that come with the differences between cultures. Maznevski (1994) describes task-related conflicts as the presence of different views and information (Nederveen Pieterse, Van Knippenberg and Van Dierendonck, 2013). It is important to note that conflict is
considered as something positive here, since it might reduce groupthink through the new perspectives and insights. All in all, task-related conflicts enhance the overall decision-making process and the more complex tasks are, the more a group benefits from additional information. In general, decisions made at board level are concerning the whole firm and relatively complex. Therefore, task-related conflicts are very relevant at board level. Next to the overall impact of task-related conflicts in groups, individuals in a group can be very valuable sources of specific information (Maznevski, 1994). Most often, this is specific knowledge about the individual’s own culture that is relevant for the firm’s future plans and survival.
Where the task-related conflicts have a positive effect on group decision-making, relationship conflicts have a negative effect. Relationship conflict within a group leads to decreased commitment to group projects (i.e. board decisions). The decrease in commitment leads to a decrease in overall group knowledge and a subsequent less optimal board decision. This is related to the argument of Mullen and Copper (1994), who state that group members in diverse groups might perceive a lack of similarities that is required for cohesion. Even subgroups might be formed within the group with a bias towards other subgroups (Larkey, 1996).
Next to task-related and relationship conflicts, cultural diversity also has an effect through its impact on trust within a group. Trust is crucial to achieve properly functioning boards. Differences between values and beliefs of individuals and the subsequent behaviour have a negative effect on trust building and cultural diversity within a group therefore supposedly has a negative effect on board functioning. An important factor for building trust is setting clear expectations, which might be harder in diverse groups. Also, the likelihood of
2.2 Board roles
To understand the effects of (cultural) diversity correctly, it is important to properly understand the role of the board in a firm. The paper of Jungmann (2006) specifically explains these roles. On the one side boards are responsible for the day-to-day management of the firm. These directors are the so-called executives. In this role, the directors are responsible for managing short-term company affairs, but also for making long-term strategic decisions. On the other side boards are responsible for supervision. These directors are the so-called non-executives or supervisory directors. Non-non-executives are appointing and dismissing the executive directors. Furthermore, non-executives are mainly monitoring executive decisions and intervene if needed. Nevertheless, it should be clear that the intention of the
non-executives is to guard the long-term survival of the firm and not to be involved in every minor issue. Next to their monitoring role, non-executives function as advisors to the executives on long-term issues, also with the goal to guard the long-term survival of the firm. Because non-executives do not work together on a daily basis, the impact of cultural diversity within the group is likely to be bigger. This is because there is less contact with other directors compared to executives, which makes building trust more challenging and individual directors are less inclined to build relationships with other directors. A lower level of trust increases the likelihood of relationship conflicts as a direct effect of cultural diversity. Furthermore, the percentage of foreign nationals as non-executives tends to be higher than for executives. This implies that more task-related conflicts occur, a direct positive effect from cultural diversity. For the reasons above, this thesis focuses on non-executives only.
2.3 National cultural context
Organizations can be viewed as social entities integrated into the institutional and value structures constituting the culture of a society (Swidler, 1986). Kostova (1997) states that country-level effects can discriminate and explain cross-country variance in
supposed to do (Kostova and Roth, 2002). In line with this reasoning, Hickson and Pugh (1995) state that national cultures are influencing the functioning of organizations. Also, the functioning of organizations is very much based on the corporate culture, where corporate culture is defined as the sharing of similar beliefs and values among members in the organization (Van den Steen, 2010). The beliefs about organizations are formed in the primary socialization phase (Berger and Luckmann, 1966), where organizational structure is an important element. The composition of the board is part of this element. In case the board composition (i.e. degree of diversity) is not according to the values and beliefs of the
organization’s members, this will lead to friction. Also, highly internationalized firms remain deeply rooted in the national business system of the country of origin (Ferner, 1997). Most often, the home country is the primary locus of control and ownership and the influence of the national culture is lasting.
The national cultural values and beliefs towards diversity differ per country. Since previous research mainly focused on different types of diversity in individual countries, it is not surprising that the outcomes differed significantly. The sample that Frijns et al. (2016) use in their research consists of 243 firms from the United Kingdom solely. This means that the values and beliefs within these firms are all very much based on the national culture in the United Kingdom. Also, the attitude towards cultural diversity within boards is very much based on the attitude towards cultural diversity within boards in the United Kingdom.
To validate the negative relationship between cultural diversity within boards and firm performance that Frijns et al. (2016) found, this thesis will focus on multiple countries. This expansion will help to answer the research question about the influence of a firm’s cultural background on the relationship between cultural diversity within boards and firm
performance.
2.4 Hofstede’s cultural framework
concepts. Nelson and Gopalan (2003) claim that the model is extremely useful to compare national cultures and study the influence of national culture on firms.
Nevertheless, the framework is not undisputed. Researchers criticize it, because it lacks heterogeneity within country populations, where multiple ethnic groups might be present (Nasif, Al-Daeaj, Ebrahimi and Thibodeaux, 1991). Also, Dimaggio (1997) argues that national borders are not sufficient for determining units of analysis, because culture is fragmented across groups. Furthermore, the methodology is criticized, because the study is executed in one firm only (i.e. IBM). Lastly, it is argued that the study is outdated and its relevance has decreased. However, these points of critique do not hinder the objective of this thesis. The major reason is that the Hofstede dimensions are measuring relative distances between cultures instead of absolute numbers. Culture does not change much over time and the relative distances between cultures have not changed much since Hofstede executed his research (Beugelsdijk, Maseland and van Hoorn, 2015). Furthermore, van Oudenhoven (2001) has confirmed that the original four dimensions are still valid, decades after Hofstede published his work. Therefore, meaningful hypotheses and conclusions can be based on the original definitions/interpretations of the dimensions.
The hypotheses in the next section will be based on the four original Hofstede
dimensions (i.e. individualism, power distance, masculinity and uncertainty avoidance). Later a fifth and sixth dimension were added, but these were not developed in a similar manner as the first four. Combining the arguments of van Oudenhoven (2001), who states that the four original dimensions are still valid, and Magnusson, Wilson, Zdravkovic, Xin Zhou and Westjohn (2008), who state that adding an extra dimension may only lead to a limited
statistical gain, it can be inferred that the original four dimensions are still valid constructs for measuring cultural distance and the subsequent measure for cultural diversity within boards.
Another important aspect that has been discussed extensively in prior studies on applying cultural frameworks is the so-called ecological fallacy. It is the error of assuming that statistical relationships at a group level also hold for individuals in the group (King, Rosen and Tanner, 2004). In other words, the mistake to interpret relationships on an
board members. The key intention of this concept is to represent diversity within the boards. Although individuals might not exactly represent the average characteristics of their country of origin, it is considered as sufficiently capturing the diversity among members. The concept of national cultural context of firms also allocates the Hofstede scores to another unit analysis than the original country level. Firms are integrated in a national context and hence also possess the cultural traits of its home country. The same logic as for the other concept applies: although firms might not exactly represent the average characteristics of its country of origin, the Hofstede scores are sufficiently capturing differences across firms from different contexts. Although firms developed independently from each other it is crucial to, once more, point out that firms are deeply rooted in their national values and practices (Drogendijk and Holm, 2012). Moreover, Harzing and Sorge (2003) state that nationally specific institutions and culture of the country of origin are applicable as facilitator for organizational practices. Using the Hofstede scores to estimate firm cultural context is therefore considered a valid
measurement.
2.5 Hypotheses
The effect that Frijns et al. (2016) find for the relationship between cultural diversity within boards and firm performance is negative. Cox (1994) provides an explanation for this negative effect. Nationality is a cause of diversity that does not change over time, where other factors such as position at work and age change over time. Watson, Kumar and Michaelson (1993) already found that national diversity had a negative effect on group performance, because these heterogeneous groups had less effective group processes than homogeneous groups. It means that the earlier mentioned negative effects (i.e. relationship conflicts and problems with intragroup trust) outweigh the positive effects (i.e. task-related conflicts and high intragroup trust). Since this thesis focuses on non-executives, who do not work together on a daily basis, it is expected that similar issues with relationship conflicts and intragroup trust occur. Furthermore, Watson et al. (1993) find that diverse groups need more time compared to homogeneous groups before they show improvement on process and
performance. This effect will be more visible for non-executive boards than for executive boards, because they do not work together as intensively. This leads to the following hypothesis:
Each firm is incorporated in a macro-environment and the values and beliefs of the country of origin influence the values and beliefs within the firm through firm culture. In other words, there are external (normative) pressures that condition firm behaviour (Ortiz-de-Mandojana, Aquilera-Caracuel and Morales-Raja, 2016), where board activities are part of the firm’s behaviour. The effect of cultural diversity within boards on firm performance is
therefore influenced by the cultural values and beliefs of the country of origin, because firms are deeply rooted in their country of origin. George, Sleeth and Siders (1999) state that firm culture influences the behaviour of employees and Kotter and Heskett (1992) confirm this by stating that the impact of firm culture is visible in the long run. This is an indirect effect of the moderating effect of the national cultural context.
Next to the effect through firm culture, national culture also has a direct effect on the relationship between diversity within boards and firm performance. Stoermer, Bader and Frouse (2016) describe how the workplace climate and acceptance of diversity is (partly) determined by national culture. They state that the national culture will either facilitate or constrain an inclusive climate. Guillaume, Dawson, Priola, Sacramento, Woods, Higson, Budhwar and West (2014) describe the concept of an ‘inclusion climate’, where all group members are included equally. This inclusion increases the number of task-related conflicts, since individual directors have equal opportunities to share their knowledge and opinions. Secondly, it decreases the number and/or intensity of relationship conflicts, since an inclusion climate also causes social integration. Related to this, the intragroup trust increases, because individuals recognize themselves as part of the group and are overall more motivated to deliver high-quality work. Hence, national cultural differences influence the effectiveness of boards. National cultures (partly) determine the inclusion of all directors and in this way the inclusion of foreign nationals. Every Hofstede cultural dimension has its own impact and the hypotheses regarding the moderating effect of the national cultural context are described below.
countries the support for diverse individuals will be stronger than in individualistic countries. From this, the following hypothesis is derived:
Hypothesis 2a: The negative relationship between cultural diversity within boards and firm performance is stronger as individualism increases.
In other words, firms with a similar score on cultural diversity within boards will perform worse in countries where individualism is high, and vice versa.
Power distance deals with the division of power within a country. In countries that score high on the power distance dimension there is a clear hierarchical order that is not very much challenged and accepted as it is. Vice versa, in countries that score low on the power distance dimension, individuals intent to equalize power as much as possible. Hence, in countries that score low on power distance, there is the demand for a fair power distribution. This means that everyone in the board should be treated equally. This leads to more task-related conflicts, which are positively influencing board outcomes. On the contrary, people in countries that score high on power distance tend to be resistant to change and do not see the need for the input of the diverse directors. This leads to the following hypothesis:
Hypothesis 2b: The negative relationship between cultural diversity within boards and firm performance is stronger as power distance increases.
In other words, firms with a similar score on cultural diversity within boards will perform worse in countries where power distance is high, and vice versa.
The third dimension is masculinity versus femininity. Countries with a high score on masculinity are more competitive and focused on rewards. They are more focused on material aspects and strive for success. Feminine countries are more focused on cooperation and caring for the weak. This leads to a more consensus-oriented culture. Hofstede (2001) describes these two opposites as the ‘tough versus tender’ cultures. In this dimension, the more
where relationship conflicts might occur and group trust decreases. All in all, these characteristics lead to the following hypothesis:
Hypothesis 2c: The negative relationship between cultural diversity within boards and firm performance is stronger as masculinity increases.
In other words, firms with a similar score on cultural diversity within boards will perform worse in countries where masculinity is high, and vice versa.
Uncertainty avoidance determines how comfortable people feel when they are in situations of uncertainty or ambiguity. According to Hofstede (2001), countries that score high on this dimension generally have rigid codes of belief and behaviour and do not like unorthodox behaviour and ideas. Vice versa, countries that score low on this dimension are more relaxed and focus more on practice than principles. People in countries that score high on uncertainty avoidance tend to see diversity as something uncertain. The new ideas and insights might contradict the current situation and uncertainty avoidant people might therefore have a negative bias towards them. This leads to a situation where task-related conflicts might not have its wanted effects. Also, trust building might be harder, because of the biased and/or negative attitude towards diversity. From this, the following hypothesis is derived:
Hypothesis 2d: The negative relationship between cultural diversity within boards and firm performance is stronger as uncertainty avoidance increases.
In other words, firms with a similar score on cultural diversity within boards will perform worse in countries where uncertainty is high, and vice versa.
3. Data and methodology
3.1 Sample selection and data collectionThe sample used in the empirical analysis consists of listed firms from ten different countries. Firms that were in the major index of the country for at least two years during the period 2011-2015 are included in the final sample. The data related to director-nationality are collected from the Boardex database, which contains director data of firms from all over the world. However, the database is far from complete and the missing values for director
nationality are therefore collected manually. The countries in the sample are chosen based on the level of diversity within boards. Palmer and Varner (2007) found that boards in European and US firms generally have more international experience than boards of Asian firms. Furthermore, Van Veen and Marsman (2008) state that boards in EU countries are marginally diverse. Also, their findings show that there are major differences between the countries. To include a sufficient degree of diversity in the sample, the eight countries with the highest degree of board nationality diversity are chosen. These countries are the Netherlands, the United Kingdom, Germany, Belgium, Sweden, France, Denmark and Finland. Ruigrok, Peck and Tacheva (2007) find a comparable degree of nationality diversity in Swiss boards, which is why Switzerland is also included in the sample. Lastly, the United States is included, since Alli, Chan, Subrahmanyam and Thapa (2010) find that a high percentage of US firms have at least one foreign director in their board.
Table 1 shows the initial sample. It presents the number of firms that were in every index for at least two years in the period 2011-2015. This period is chosen, because it is the most recent period where the majority of director data is available. Furthermore, the firms need to be originally from one of the ten countries that are included in this thesis. The indices included are Dow Jones 30 (United States), FTSE 100 (United Kingdom), AEX (the
Table 2 shows the number of firms per country of origin that were included in the final sample. There are 319 firms left in the final sample, which means that 85% of the initial sample was maintained. Firms that were cross-listed are only included once in the final sample (and categorized based on the country of origin) and firm-years that had missing data about their executives were also excluded. Even when the nationality of one
non-executive was missing, that specific year was excluded because it can bias the independent cultural diversity within boards. Since directors are generally in a board for more than one year, this led to the exclusion of firms from the sample. The final sample consists of 1,463 firm-year observations. This means that on average 4.6 years of the 5-year period are
available per firm. The final sample consists 15,062 director-nationality observations. Table 3 provides an overview of the directors’ country of origin.
Table 2 - Final sample
Country Number of firms
Belgium 20 (6.27%) Denmark 19 (5.96%) Finland 26 (8.15%) France 37 (11.60%) Germany 32 (10.03%) The Netherlands 24 (7.52%) Sweden 20 (6.27%) Switzerland 22 (6.90%) United Kingdom 88 (27.59%) United States 31 (9.72%) Total 319 (100%)
Table 1 - Initial sample
Index Number of firms
AEX 28 BEL 20 23 CAC 40 43 DAX 32 Dow Jones 30 34 FTSE 100 110 OMX Copenhagen 20 23 OMX Helsinki 25 28 OMX Stockholm 30 30 Swiss Market Index 21
The Boardex database was used to determine who the non-executive directors per firm per year are. They are qualified as supervisory directors and clearly separated from the
executive directors. This makes it possible to use the data for both firms that use a one-tier structure and firms that use a two-tier structure (Jungmann, 2006). The missing data on non-executive nationality were gathered using Bureau van Dijk’s Orbis database. Also, firm annual reports, firm websites and LinkedIn were used to complement the data on director nationality. The data on firm financial performance (ROA and Tobin’s q) were gathered with Thomson Reuters’ Datastream database. In case firm financial data was missing, this was completed with information from Orbis and annual reports.
3.2 Independent variable – cultural diversity within boards
Previous measures of cultural differences mainly focused on the primary research objective: investigating differences between groups. Frijns et al. (2016) developed a novel metric that measures diversity within groups.
First, the cultural distance between a pair of directors is calculated following Kogut and Singh (1988). This formula is based on the four original Hofstede dimensions (Hofstede, 2001). Hofstede collected data for a wide variety of countries and calculated a score for each country that ranges from 0 to 100. Imputing the scores of two countries in the following formula will give the score on cultural distance for a pair of non-executives:
CD#$ = .(/0 I(#− I($ */V# /4 ∀ 2 ≠ 4 (1)
where CDij represents the cultural distance between a pair of directors (i and j). Iki is the
culture score on dimension k for director i, and Ikj represents the same for the second director.
Vi is the variance for cultural dimension i (Frijns, et al., 2016). Obviously, in case both
Table 3 - Number of non-executives per country, per year
This table presents the number of non-executives per nationality for the period 2011-2015. The numbers presented in bold are representing the ten countries where the firms in the sample come from.
Country 2011 2012 2013 2014 2015 Total % Total Country 2011 2012 2013 2014 2015 Total % Total Australia 24 22 19 21 22 108 0.72% Lebanon 0 0 1 0 0 1 0.01% Austria 27 34 33 33 28 155 1.03% Malaysia 1 1 2 2 1 7 0.05% Bahrain 0 0 1 1 1 3 0.02% Mexico 8 8 11 11 10 48 0.32% Belarus 1 1 1 1 1 5 0.03% Morocco 1 1 1 0 0 3 0.02%
Belgium 133 135 125 152 145 690 4.58% Netherlands 134 140 129 122 115 640 4.25%
Brazil 5 6 8 9 11 39 0.26% New Zealand 4 4 4 6 5 23 0.15% Bulgaria 1 1 0 0 0 2 0.01% Norway 16 18 19 19 20 92 0.61% Canada 33 33 34 39 41 180 1.20% Pakistan 0 0 0 1 0 1 0.01% Chili 4 5 5 7 7 28 0.19% Philippines 0 0 1 1 1 3 0.02% China 11 12 16 14 15 68 0.45% Poland 1 1 0 0 0 2 0.01% Colombia 0 0 0 1 1 2 0.01% Portugal 1 1 1 2 2 7 0.05% Cyprus 0 0 1 1 0 2 0.01% Qatar 5 5 4 4 4 22 0.15% Czech Republic 0 0 0 1 2 3 0.02% Romania 0 1 1 1 2 5 0.03%
Denmark 134 123 107 128 128 620 4.12% Russia 7 9 9 8 6 39 0.26%
Egypt 1 1 1 1 1 5 0.03% Saudi Arabia 1 0 0 0 0 1 0.01%
Finland 141 144 141 141 126 693 4.60% Singapore 2 4 6 12 13 37 0.25%
France 452 450 420 434 409 2165 14.37% Slovakia 0 0 0 0 1 1 0.01%
Germany 468 490 458 471 462 2349 15.60% Slovenia 0 0 0 1 0 1 0.01%
Ghana 1 1 1 1 1 5 0.03% South Africa 19 23 22 23 23 110 0.73% Greece 1 1 1 2 3 8 0.05% South Korea 2 2 1 2 2 9 0.06% Hong Kong 0 0 0 1 2 3 0.02% Spain 14 13 15 16 18 76 0.50% Hungary 0 0 1 1 1 3 0.02% Sweden 180 184 189 197 190 940 6.24%
India 15 16 11 11 14 67 0.44% Switzerland 96 97 100 84 85 462 3.07%
Iran 0 1 1 2 2 6 0.04% Tanzania 0 0 0 0 2 2 0.01% Ireland 14 12 12 13 14 65 0.43% Turkey 0 0 1 1 2 4 0.03% Italy 27 27 33 34 29 150 1.00% United Kingdom 491 513 512 546 525 2587 17.18%
Jamaica 1 0 0 1 0 2 0.01% United States 470 508 510 503 492 2483 16.49%
Second, to calculate the cultural diversity within a board (instead of cultural distance between two directors), the novel metric of Frijns et al. (2016) is used. They define this as the average of cultural distances between all pairs of board members:
CD BOARD9: =
CD#$,9:,
#,$
<(< − 1)/2 ∀ 2 < 4
where CD BOARDnt is the cultural diversity within a board of firm nin year t. The measure is
scaled for the number of pairs of board members, where m is the number of board members. This part is included to normalize the outcome for board size (Frijns, et al., 2016).
3.3 Dependent variable – ROA and Tobin’s q
In this thesis, the influence of cultural diversity within boards on firm performance is measured and follows Frijns et al.’s (2016) approach for measuring firm performance by using ROA and Tobin’s q.
ROA is an indicator that measures how efficiently assets are used to generate earnings. It can be considered as a measure that is used for comparing both operating performance and profitability, since it shows how efficiently the firm converts its investments (i.e. assets) into net income. An increase in ROA therefore indicates an increased ability to make profits with the same amount of assets. It is measuring internal performance, although it is also very useful for investors.
BCDEFG HG IJJCDJ BKI = LCD MGNH<C OHDPQ IJJCDJ
Tobin’s q measures a firm’s market value compared to its book value. This measure includes an external component, namely market value. It therefore represents the perception of investors. This perception is based on a firm’s performance (e.g. operating performance, profitability and leverage) and subsequently Tobin’s q is an adequate measure of a firm’s overall performance.
OHR2GSJ T =OHDPQ IJJCDJ − UHHV WPQEC XTE2DY + [PFVCD WPQEC XTE2DY
OHDPQ IJJCDJ (4)
3.4 Moderator variable – firm cultural context
Firm cultural context is measured by the score for each Hofstede dimension (i.e. individualism, power distance, masculinity and uncertainty avoidance). The Boardex database is used to determine the firm’s country of origin. Subsequently, the Hofstede scores of the ten countries in this thesis are connected to the firm. The scores are added as separate moderators to the regression analysis. In table 4 the Hofstede scores per country are shown.
Table 4 - Hofstede scores per dimension, per country
Country Individualism Power distance Masculinity Uncertainty avoidance
Belgium 75 65 54 94 Denmark 74 18 16 23 Finland 63 33 26 59 France 71 68 43 86 Germany 67 35 66 65 The Netherlands 80 38 14 53 Sweden 71 31 5 29 Switzerland 68 34 70 58 United Kingdom 89 35 66 35 United States 91 40 62 46 3.5 Control variables
In the empirical analysis several control variables are included that might affect firm performance. To test the relative relationship between the independent variable and the dependent variable, control variables at several levels are included.
assets and expected to have a positive effect on firm performance (Vedd and Yassinski, 2015). Also, in previous research on diversity this positive effect was found (Frijns, et al., 2016). Leverage is measured as total debt divided by total debt and shareholder equity and expected to negatively influence firm performance (Frijns, et al., 2016). The degree of foreignness is measured as foreign sales divided by total sales and the effects are mixed across different studies. Frijns et al. (2016) find that a higher degree of foreignness has a negative effect on firm performance, but foreign presence (as measured by NYSE listings) has a positive effect. Lu and Beamish (2004) confirm these mixed results. They argue that the total costs of internationalization are exponential, whereas the benefits have an inverted U-shape. The costs increase due to increased coordination costs, whereas benefits do no longer increase, because potential foreign resources (e.g. knowledge) are used up. At national level, inflation and GDP growth are included. These are included, because firms from ten countries are included and to exclude possible effects that national conditions have on firm
performance. The effects of inflation are expected to be negatively related to firm performance, whereas GDP growth is positively influencing firm performance.
3.6 Methodology
To test the hypotheses in this thesis, a regression analysis is performed. To test the relationship between cultural diversity within boards and firm performance, the following empirical model is used:
BKI#,: = \] + ^0_` UHPFa#,:+ ^*b2F< J2cC QHd #,: + ^efCgCFPdC#,: + ^.bHFC2dG JPQCJ#,:+ ^hUHPFa J2cC#,:+ ^iMGjQPD2HG#,: + ^kl`m lFHnDℎ#,: + p#,:+ E#,:
where ROAi,t represents return on assets as firm performance of firm i in year t. In case firm
performance is measured with Tobin’s q, ROAi,t is replaced by Tobin’s qi,t which represents
Tobin’s q for firm i in year t. The constant is represented by \] and the independent variable is CD Boardi,t. It represents the cultural diversity within the board of firm i in year t. This
number is based on all the individual scores on the Hofstede dimensions for each director. Firm size (log)i,t is the logarithm of firm total assets in US dollars, Leveragei,t is calculated as
total liabilities divided by the sum of total liabilities and total shareholder’s equity and Foreign Salesi,t is representing the degree of foreignness and calculated as the percentage of
foreign sales relative to total sales. Board sizei,t represents the number of board members in
the board of firm i in year t. Inflationi,t is the inflation rate in firm i’s country of origin in year
t and GDP Growthi,t is the GDP growth percentage in firm i’s country of origin in year t.
Finally, p#,:represents the individual-specific error term, which covers the deviation of the
observed values from the (unobservable) expected value and is time invariant and ui,t is the
remainder error component.
Before running the regression, the data is checked for several conditions. An important data issue is the presence of heteroscedasticity. The Breusch-Pagan test was used to check for heteroscedasticity. It regresses the error term on the independent variables. The outcome of the F-test is significant and therefore heteroscedasticity is present in the dataset. Also, a scaled version of the Breusch-Pagan test, the Pesaran test, was performed to test for cross-sectional correlation. The outcome was significant and therefore cross-sectional correlation is present. Furthermore, the Wooldridge test for autocorrelation (AR (1)) was executed and found significant at 0.01. This means that autocorrelation is present and might cause unreliable results. Another common problem that researchers face is endogeneity. To check for the presence of endogeneity the Hausman test was executed. The test confirmed the presence of endogeneity among the variables.
Taking into account the data conditions, the feasible generalized least squares method (FGLS) is used. This method can be applied when N (i.e. number of firms) is larger than T (i.e. number of time-periods) and this condition is met for the dataset. FGLS provides robust standard errors in case the data suffer from heteroscedasticity, cross-sectional correlation and autocorrelation (Parks, 1967; Verbeek, 2004). The significant outcome of the Hausman would normally lead to the application of a fixed-effects (FE) approach. However, the moderator variable is time invariant (i.e. cultural context is unchanged over time) and using the FE approach would therefore fail to measure its effect. Since the assumption is that differences across countries are influencing firm performance (dependent variable), a random effects (RE) approach is chosen. FGLS is a method that uses the RE approach and is able to estimate coefficients for invariant variables (Bell and Jones, 2015).
Next to the overall effect of cultural diversity within boards on firm performance, this thesis focuses on the effects of the firm’s cultural context (i.e. hypothesis 2a, 2b, 2c and 2d). The following empirical models include the four different moderator variables (i.e.
BKI#,: = \]+ ^0_` UHPFa#,:+ ^*MGa2g2aEPQ2J<# + ^e_` UHPFa
∗ MGa2g2aEPQ2J<#,: + ^.b2F< J2cC QHd #,: + ^hfCgCFPdC#,: + ^ibHFC2dG JPQCJ#,:+ ^kUHPFa J2cC#,:+ ^rMGjQPD2HG#,: + ^sl`m lFHnDℎ#,:+ p#,:+ E#,:
BKI#,: = \]+ ^0_` UHPFa#,:+ ^*mHnCF a2JDPGNC#+ ^e_` UHPFa
∗ mHnCF a2JDPGNC#,:+ ^.b2F< J2cC QHd #,:+ ^hfCgCFPdC#,:
+ ^ibHFC2dG JPQCJ#,:+ ^kUHPFa J2cC#,:+ ^rMGjQPD2HG#,: + ^sl`m lFHnDℎ#,:+ p#,:+ E#,:
BKI#,: = \]+ ^0_` UHPFa#,:+ ^*[PJNEQ2G2DY# + ^e_` UHPFa
∗ [PJNEQ2G2DY#,: + ^.b2F< J2cC QHd #,:+ ^hfCgCFPdC#,: + ^ibHFC2dG JPQCJ#,:+ ^kUHPFa J2cC#,:+ ^rMGjQPD2HG#,: + ^sl`m lFHnDℎ#,:+ p#,:+ E#,:
BKI#,: = \]+ ^0_` UHPFa#,:+ ^*tGNCFDP2GDY PgH2aPGNC# + ^e_` UHPFa
∗ tGNCFDP2GDY PgH2aPGNC#,:+ ^.b2F< J2cC QHd #,:
+ ^hfCgCFPdC#,: + ^ibHFC2dG JPQCJ#,:+ ^kUHPFa J2cC#,:
+ ^rMGjQPD2HG#,:+ ^sl`m lFHnDℎ#,:+ p#,: + E#,:
where Individualismi,t, Power distancei,t, Masculinityi,t and Uncertainty avoidancei,t enter the
model as independent variables. They represent the scores on each of the dimensions for firm
i’s country of origin. Since this number is fixed during the sample period, there is no time
component present here. Furthermore, the interaction terms CD Board*Individualismi,t, CD
Board*Power distancei,t, CD Board*Masculinityi,t and CD Board*Uncertainty avoidancei,t are
added. They represent the moderating effect of the firm’s cultural context on the relationship between cultural diversity within boards and firm performance. They consist of the scores on the cultural context (i.e. the Hofstede dimensions) multiplied by the scores on cultural
diversity within boards in firm i in year t. The time component is present, because the score on cultural diversity within boards possibly changes over the years, although the score on cultural context remains unchanged. Lastly, also the error terms p#,: and E#,: are added to account for the deviation of the observed value from the expected value. In the four models above, ROAi,t is representing firm performance. When firm performance is measured with
Tobin’s q, ROAi,t is replaced by Tobin’s qi,t.
Because the correlation matrix and the subsequent Variance Inflation Factor (VIF) tests indicate that there is high multicollinearity between the moderator variables and the independent variable, the results are possibly unreliable. Multicollinearity influences the coefficients, the t-values and the subsequent significance levels of the variables. Aiken and West (1991) developed a methodology that creates reliable results, while using the same
(6)
(7)
(8)
dataset, called mean-centering. Multicollinearity is decreased by subtracting the variable mean of the observed value. Positive and negative values are created that together have a 0 mean. In the models that include interaction terms the interaction terms as well as the variables that form the interaction terms are mean-centered.
4. Results
4.1 Descriptive statistics
This section describes the descriptive statistics. Since several variables included extreme values, they are winsorized at 5%, which means that both the upper and bottom 2.5% values are replaced. This excludes the possible effect of spurious outliers.
Table 5 provides an overview of the descriptive statistics. There are 1,463 firm-year observations, spread over the 319 firms in the period 2011-2015. ROA and Tobin’s q are the two dependent variables that represent firm performance. ROA has a mean of 0.0612 and both positive and negative values, which indicates that there are firm-year observations where firms made losses and firm-year observations where firms made profits. Furthermore, the position of the median relative to the mean shows that the data are skewed towards the lower values. It means that the majority of the firms have an ROA lower than the average. The value of Tobin’s q theoretically ranges from 0 to infinity. In this dataset, the average score on Tobin’s q is 1.6639, which means that that the profits made in the firm are higher than the costs of the firm’s assets. Since the median is lower than the mean, the data are skewed towards the lower values. The independent variable in the analysis is CD Board, which
represents the cultural diversity within boards. The mean score for this variable is 0.5972. The data are skewed towards the lower values, because the median is lower than the mean. An explanation for this is that 15.58% of the firms have a score of 0 on the CD Board variables. This information is provided by the variable CD dummy, which has a score of 1 when there is at least one foreign director in the board and a score of 0 when all directors have the same nationality. A score of 0 on CD dummy therefore automatically leads to a score of 0 on CD Board. The highest score on CD Board is 2.5091, which represents the board with the highest degree of cultural diversity.
assets. The differences in firm size are immense, as is indicated by the range. Furthermore, the standard deviation is higher than the mean, which means that there are very large firms included in the sample that bias the outcome. Therefore, the log of firm size is included. The degree of foreignness is represented by foreign sales. Important to note here is that there are firms that have no foreign sales (represented by the 0 at minimum) and firms that have almost all their sales abroad (represented by the 0.9930 at maximum). Together with the mean of 0.5915 and the median of 0.6564 it can be stated that the sample is very diverse regarding the degree of foreignness. At board level, board size is included. The sample consists of boards between 3 and 23 directors, but on average around 10 people are in a board. Furthermore, inflation and GDP growth represent the control variables at national level. Both variables have positive and negative values, which means that circumstances at national level have contributed to firm performance as well as caused a negative impact.
The scores on the four Hofstede dimensions represent the cultural context of firms, which is the moderator variable. From the numbers, it can be extracted that most variety is present in the masculinity and uncertainty avoidance dimension. The range for both dimensions is quite wide and the standard deviations are highest. Vice versa, the variety between the cultural contexts is lowest in the individualism dimension, which has the lowest standard deviation and a narrow range compared to the other dimensions.
Table 6 shows the correlation matrix, including the interaction terms. The correlations between CD Board and both dependent variables are negative and therefore in line with the expected negative relationship. However, it is not possible to draw final conclusions from the correlation matrix.
Table 5
Descriptive statistics
This table presents the descriptive statistics of the 1,463 firm-year observations for variables included in this thesis. Variables with * are winsorized at 2.5th and 97.5th percentile
Variable Observations Mean Standard Deviation Median Maximum Minimum
ROA* 1,463 0.0612 0.0612 0.0508 0.1979 -0.0315 Tobin’s q* 1,463 1.6639 0.7811 1.4077 3.7860 0.9322 CD Board 1,463 0.5972 0.5656 0.4925 2.5091 0.0000 CD dummy 1,463 0.8442 0.3628 1 1 0 Leverage* 1,463 0.6356 0.1844 0.6151 0.9570 0.3306 Firm size* 1,463 104,000,000 204,000,000 24,800,000 841,000,000 2,064,800
Firm size log 1,463 17.0874 1.6669 17.0261 20.5501 14.5405
Table 6 Correlation matrix
This table presents the correlations between the dependent and the independent variables, based on the 1,469 firm-year observations. The variables ROA, Tobin’s q, Firm size, Leverage and Foreign sales are winsorized at the 2.5th and 97.5th percentile.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
1 ROA 1.0000
2 Tobin's q 0.5828 1.0000
3 CD Board -0.1107 -0.0055 1.0000
4 Firm size (log) -0.4083 -0.3346 0.0094 1.0000
4.2 The influence of diversity within boards on firm performance
The first hypothesis suggests that there is a negative relationship between cultural diversity within boards and firm performance. Table 7 shows the results of the FGLS regression that was run to investigate this relationship. Model 1 indicates that there is a significant negative effect of cultural diversity within boards on firm performance when firm performance is measured as ROA. Model 3 shows a negative relationship between cultural diversity within boards and firm performance when firm performance is measured as Tobin’s q, but this effect is not significant. This suggests that increasing diversity within boards does negatively impact overall firm performance. These findings are in line with the negative relationship that Frijns et al. (2016) found in their study and (partly) confirm hypothesis 1.
Table 7
Feasible generalized least squares regressions with determinants of firm performance, excluding moderators.
ROA and Tobin’s q are the dependent variables, measuring firm performance. CD Board is the independent
variable that represents cultural diversity within boards. The other variables are the control variables, where
Firm size (log) is the logarithm of total assets, Leverage is the amount of debt divided by the total of debt and
shareholder’s equity, Board size is the number of board members, Inflation the inflation rate of the firm’s country of origin and GDP growth the GDP growth of the firm’s country of origin. Standard errors are presented between parentheses and ***, ** and * indicate the statistical significance at 0.001, 0.01 and 0.05 level,
respectively.
(1) (2) (3) (4)
ROA ROA Tobin’s q Tobin’s q
CD Board -0.00744** -0.0426
(0.00261) (0.0353)
Firm size (log) -0.0113*** -0.0116*** -0.132*** -0.134***
(0.00109) (0.00109) (0.0147) (0.0146) Leverage -0.0847*** -0.0854*** -0.522*** -0.526*** (0.00867) (0.00869) (0.117) (0.117) Foreign sales -0.000166*** -0.000206*** 0.00111 0.000885 (0.0000481) (0.0000462) (0.000649) (0.000622) Board size -0.00000612 0.000199 -0.00808 -0.00690 (0.000454) (0.000450) (0.00614) (0.00607) Inflation 0.00333** 0.00367*** -0.0925*** -0.0906*** (0.00111) (0.00111) (0.0150) (0.0149) GDP growth 0.00748*** 0.00791*** 0.0666*** 0.0690*** (0.00124) (0.00123) (0.0168) (0.0166) Constant 0.308*** 0.308*** 4.322*** 4.324*** (0.0160) (0.0161) (0.217) (0.217) Observations Groups 1,463 319 1,463 319 1,463 319 1,463 319 Wald 495.96 485.19 285.28 283.54 Prob > chi2 0.000 0.000 0.000 0.000
4.3 The influence of firm cultural context
The results in table 8 show that none of the interaction terms is significant. This indicates that it cannot be stated that a firm’s cultural background influences the relationship between diversity within boards and firm performance as measured by ROA. As the results are insignificant it is not possible either to interpret the directions and size of the effects. Nevertheless, the inclusion of the moderators also influences the other variables in the models. For example, the significance of the independent variable CD Board decreases when a moderating effect is included. Also, its coefficient becomes smaller in all models compared to model 1 in table 7. This implies that the effect of cultural diversity within boards on performance is still present, but that it might be overestimated in the model without moderators. Furthermore, the inclusion of the moderator variables also influences the significance and size of the effects of some control variables. The effects of firm size, leverage and foreign sales on ROA do not change much when moderators are included. However, the sign of board size changes from negative to positive. Again, this might relate to the mixed and inconclusive results that previous studies found. The inclusion of moderators also changes the significance and direction of the effect of inflation. Without moderators, it was positive and significant, but in table 8 the results are mixed. This implies that the effect of inflation on ROA is partly explained by the difference in culture and not an effect that is similar in all contexts. Lastly, the significance of GDP growth decreases when individualism and the moderator are included. However, the sign does not change and the overall effect of GDP growth remains positive.
Table 9 is similar to table 8, but the dependent variable changed from ROA to Tobin’s q. Compared to table 7, without interactions, no significant effects of cultural diversity within boards on firm performance were found. Furthermore, no significant moderator effects were found, which is similar to table 8. Overall, it is therefore not possible to confirm hypotheses 2a, 2b, 2c and 2d. The control variables firm size, leverage, inflation and GDP growth do not change compared to model 3 in table 7. The effect of foreign sales becomes significant when individualism and its moderator are included and board size becomes significant when uncertainty avoidance and its moderator are included.
Table 8
Feasible generalized least squares regressions with determinants of firm performance, including moderators.
ROA is the dependent variable, measuring firm performance. CD Board is the independent variable that
represents cultural diversity within boards. The moderator variables are CD Board * Individualism, CD Board *
Power distance, CD Board * Masculinity and CD Board * Uncertainty avoidance. Also, the Individualism, Power distance, Masculinity and Uncertainty avoidance are included to account for their direct effects. The
other variables are the control variables, where Firm size (log) is the logarithm of total assets, Leverage is the amount of debt divided by the total of debt and shareholder’s equity, Board size is the number of board members, Inflation the inflation rate of the firm’s country of origin and GDP growth the GDP growth of the firm’s country of origin. Standard errors are presented between parentheses and ***, ** and * indicate the statistical significance at 0.001, 0.01 and 0.05 level, respectively.
(1) (2) (3) (4)
ROA ROA ROA ROA
CD Board -0.00559* -0.00720** -0.00226 -0.00649* (0.00253) (0.00267) (0.00263) (0.00268) Individualism 0.00186*** (0.000174) CD Board * Individualism 0.000174 (0.000270) Power distance -0.000181 (0.000112) CD Board * Power distance -0.0000243 (0.000209) Masculinity 0.000741*** (0.0000714) CD Board * Masculinity -0.0000152 (0.000101) Uncertainty avoidance -0.000290*** (0.0000813) CD Board * Uncertainty avoidance 0.0000830
(0.000128) Firm size (log) -0.0133*** -0.0115*** -0.00925*** -0.0124***
Table 9
Feasible generalized least squares regressions with determinants of firm performance, including moderators.
Tobin’s q is the dependent variable, measuring firm performance. CD Board is the independent variables that
represents cultural diversity within boards. The moderator variables are CD Board * Individualism, CD Board *
Power distance, CD Board * Masculinity and CD Board * Uncertainty avoidance. Also, the Individualism, Power distance, Masculinity and Uncertainty avoidance are included to account for their direct effects. The
other variables are the control variables, where Firm size (log) is the logarithm of total assets, Leverage is the amount of debt divided by the total of debt and shareholder’s equity, Board size is the number of board members, Inflation the inflation rate of the firm’s country of origin and GDP growth the GDP growth of the firm’s country of origin. Standard errors are presented between parentheses and ***, ** and * indicate the statistical significance at 0.001, 0.01 and 0.05 level, respectively.
(1) (2) (3) (4)
Tobin’s q Tobin’s q Tobin’s q Tobin’s q CD Board -0.0253 -0.0299 -0.0469 -0.00873 (0.0348) (0.0354) (0.0368) (0.0348) Individualism 0.0165*** (0.00241) CD Board * Individualism -0.00384 (0.00372) Power distance -0.0113*** (0.00148) CD Board * Power distance -0.00233 (0.00277) Masculinity -0.00124 (0.00100) CD Board * Masculinity 0.000682 (0.00142) Uncertainty avoidance -0.0120*** (0.00106)
CD Board * Uncertainty avoidance 0.00224
(0.00166) Firm size (log) -0.153*** -0.148*** -0.136*** -0.179***
4.4 Further analysis
Several additional tests are conducted to validate the results from the core analysis. A quantile regression is performed that tests for the impact of CD Board and the moderators across every 10th quantile of ROA and Tobin’s q. The results can be found in table A.1 and A.2 in the appendix. Previous studies on diversity indicated heterogeneity with respect to the impact on performance (Dang and Nguyen, 2016). The results of the quantile analysis confirm that the impact of cultural diversity on ROA also differs across quantiles, but is always negative. Furthermore, marginal effects are found for the effect of CD Board on Tobin’s q for the 50th, 60th and 70th quantile, which shows that the impact on Tobin’s q also differs. Also, a significant negative impact is found for the moderators Individualism and Masculinity on Tobin’s q from the 10th to the 40th quantile. This implies that high scores on individualism and masculinity strengthen the negative relationship between cultural diversity within boards and firm performance (as measured with Tobin’s q) for firms with a low Tobin’s q. Moreover, FGLS regressions were run where the 88 firms from the United Kingdom are excluded. As is described in the data and methodology section, the number of firms is more or less equally spread over the countries. However, only the number of firms from the United Kingdom is significantly larger than the number of firms from the other nine countries. To exclude possible selection bias due to an unequally balanced sample, the sample was limited to the remaining 231 firms. The results are presented in table A.3 and A.4 in the appendix and indicate that the size of the effect of cultural diversity within boards on firm performance increases. Also, the effect of CD Board on Tobin’s q becomes marginally
significant. Overall, this does not change the conclusion regarding hypothesis 1. In contrary to the core analysis, the moderator Individualism becomes significant (for ROA) and is in line with the predicted negative direction. Also, the moderator Masculinity is negative and
becomes marginally significant (for ROA). Running the FGLS regression to test hypothesis 1 on firms from the United Kingdom only (unreported), indicates that there is no relationship between cultural diversity within boards and firm performance. This clarifies the increased significance of both CD Board and the moderators Individualism and Masculinity. Lastly, a regression is run where the firms without a foreign non-executive in the board (and thus a score of zero on CD Board) are excluded. The results are presented in table A.5 and A.6 in the appendix and confirm the findings of the core analysis; there is a negative relationship
5. Conclusion
The number of foreign nationals in corporate boards has increased over the past years.
Previous studies already focused on several aspects of diversity within boards that affect firm performance. However, culture as a form of diversity has been under researched. The limited number of previous studies all focused on firms from one country (see, e.g., Frijns, et al. 2016; Van Veen and Marsman, 2008). The aim of this thesis is to empirically test the relationship between cultural diversity within boards and firm performance in multiple countries and find evidence for the moderating effect a firm’s cultural context has on this relationship. The empirical analysis included 319 firms (and 1,463 firm-year observations) from ten different countries. The FGLS method was applied to run the regressions.
The regression outcomes show that there is a significant negative relationship between cultural diversity within boards and firm performance when firm performance is measured as ROA. The relationship is also negative when performance is measured as Tobin’s q, but insignificant. ROA as a measure of firm performance represents an internally focused aspect, namely the efficient usage of assets. On this aspect, the negative effects of cultural diversity within boards outweigh the benefits. Connecting this to the causes of positive and negative influences of diversity, it can be stated that relationship conflicts and a lack of trust seem to outweigh the task-related conflicts. A possible explanation for this finding is that the added value of foreign directors is related to handling unfamiliarity with the foreign environment and their specific market knowledge (Zaheer, 1995) and less to their experience on
operational aspects. A possible explanation for the insignificant negative effect on Tobin’s q is that Tobin’s q is more externally focused and representing firm value. The specific
knowledge of foreign directors in task-related conflicts regarding market entry and presence has more influence on market value (as a part of Tobin’s q) than on operational efficiency (as represented by ROA). However, this still does not explain the, although insignificant,
negative direction of the relationship.
are still rooted in the cultural values of the country of origin, they have grown over the years and gained foreign insights. The effects of the cultural context possibly decreased, because of the exposure to many foreign cultures over multiple years. The quantile regression robustness test provides a more nuanced view of the impact of the cultural context across firm
performance. Although it can overall be concluded that the cultural context does not influence the relationship between cultural diversity within boards and firm performance, future
research can focus on the influence of individualism and masculinity. These two dimensions are influencing the relationship, for firms that perform below average only.
The contribution of this thesis to academic literature is its focus on a previously under researched part of board diversity, namely culture. Furthermore, it contributes by expanding towards a multi-country study, where earlier studies focused only on one country. It
specifically validates the results that Frijns et al. (2016) found and the inclusion of ten countries also leads to an increased ability to generalize the results. Lastly, the moderating effect has, to the best of my knowledge, not been studied before. From a managerial
perspective, the results give insights in the effects of board composition on firm performance. The lack of a moderating effect implies that this effect is valid in all ten countries. When appointing a new board member, the influence of a new board member on the team can be assessed more accurately and the other board members can be prepared beforehand. The subsequent increased board effectiveness will lead to improved performance.
Also, this thesis has several limitations that should be acknowledged. Although it expands the scope of previous studies, ten Western countries is still a limited number in a cross-country study. Even though the Boardex database is far from complete, it can be used to expand this thesis by including an increased number of countries. Secondly, this thesis
focusses on non-executive (supervisory) directors. This term is rather broad and board systems differ significantly across countries. The specific tasks of non-executives therefore vary across countries. Future studies can focus on specific subgroups within the supervisory board of directors, such as the auditing committee and the employee representatives. Third, national culture is used as a proxy for both cultural diversity within boards (based on director nationality) and the firm’s cultural background. However, the values and beliefs of
