The effect of female board members on environmental and social
performance: the moderating effects of power and national culture
Strategy Track Master Thesis
Student: Lotte Horikx Student №: 10340300
Supervisor: Daniel Waeger
Faculty: Economics and Business, University of Amsterdam Study program: MSc. in Business Administration – Strategy Track Submission final draft: 24.06.2016
- 1 - Statement of originality
This document is written by Student Lotte Horikx who declares to take full responsibility for the
contents of this document.
I declare that the text and the work presented in this document is original and that no sources
other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of
- 2 - Table of Contents Abstract……….. 3 1. Introduction……… 4 2. Literature review……… 5 1. Gender……… 6 2. Power………. 8 3. National culture………. 8 3. Theoretical framework………... 10 1. Gender dispersion……….. 10
2. The power of the board of directors……….. 13
3. The national culture of the firm………. 16
4. Methodology……….. 20 1. Sample………20 2. Independent variable……….. 21 3. Dependent variables………... 22 4. Moderator variables………23 5. Control variables……… 24 5. Results……… 25 1. Descriptives………25 2. Normality………... 28 3. Correlations……… 32 4. Regression analysis……… 36 6. Discussion……….. 44 1. Academic relevance………... 45
2. Implications for practice……… 51
7. Limitations and future research………. 52
8. Conclusion………. 54
9. Bibliography………...56
10. Appendix……….. 61
1. Normality analysis………. 61
- 3 - Abstract
Over the years, two trends have slowly started to change the way organizations are structured and
operating. Firstly, women have started to emerge in the upper echelons of the organization and
play an increasingly important part in guiding the strategic direction of the firm, and thereby
impact performance. The second trend is the increased pressure that society exercises on firms in
order to make them more sustainable and transparent. While, previous research has started investigating a possible link between the number of women on the board of directors and a firm’s
environmental and social performance, there is still much that remains to be explored. This
research endeavours to address this gap by looking at the possible relationship between the
proportion of women on the board of directors and a firm’s environmental and social
performance. Furthermore, this thesis also looks into the moderating effects of power of the board and a firm’s cultural background, on the relation between female directors and
environmental and social performance. A sample of 329 firms from the Global Fortune 500 has
been used, for which the necessary data was taken from ASSET4 and Worldscope. The results of
the performed regression analysis show a significantly negative relationship between the percentage of female board members and the firm’s environmental performance, while
controlling for firm size, industry, debt ratio, slack resources and geographical region.
Additionally, the moderator variable power distance has been shown to weaken the positive
relationship between proportion of female board members and social performance. No support
has been found for any of the other moderator variables, specifically board size, CEO duality and
masculinity. Nevertheless, this research has taken an important step in understanding the
relationship between the number of women on the board of directors and a firm’s environmental
- 4 - 1. Introduction
Over the years, the role of women in organizations has received more and more attention. The
role of housewife is no longer the sole occupation of a woman. Instead, they are increasingly
better educated and start families at a later age, making them more prominent on the work floor
than ever. However, breaking through the glass ceiling towards top management positions has
posed some difficulties (Oakley, 2000). This is partly due to the fact that women have other skills
and values than men (Adams & Funk, 2012; Jaffee & Hide, 2000), and that in many corporate
cultures there is still a strong bias towards the merits of men as being the most desirable (Oakley,
2000). However, research has shown that when women share the leadership of the company, the
firm is more likely to succeed and achieve its goals (The Economist, 2015; Carter, Simkins &
Simpson, 2003). Thus, it appears that the presence of women in top management positions may
affect the strategic direction of a firm and consequent outcomes.
The upper echelon theory states that organizations are representations of their top
management, and that executives act in a certain way due to their personal experiences and
values (Hambrick & Mason, 1984). So, the different experiences and values of women,
particularly concerning corporate social responsibility (CSR) issues, are likely to result in other
strategic decisions compared to those of men. More specifically, the tendency of women to focus
on the long term and take into consideration various stakeholder views (Griskevicius, Cantú &
van Vught, 2012; Jaffee & Hide, 2000), may increase the likelihood of adopting a course of
action which positively influences both the environment as well as society. Some countries, such
as Norway, have seen the potential benefit of female executives and changed their legislation
accordingly, instigating quotas for the number of women that need to be present on the board of
directors (The Economist, 2014). However, women still lag behind men in most organizations,
- 5 - executives has to be made more specific. Which is where this research comes in, trying to
investigate the contribution of women more explicitly in order to clarify their effect on the organization’s environmental and social performance. In other words, this thesis strives to
provide clarity of the effect that female board members have on the environmental and social
performance of a firm, and thereby create a better representation of the contribution of women in
top management positions.
2. Literature review
Over the last decade greater attention has been paid to the environmental and social performance
of firms. Society is demanding that firms operate not merely to make a profit, but become
actively engaged in more sustainable practices (Oppenheim, Bonini, Bielak, Kehm & Lacy, 2007;
Marcus & Fremeth, 2009), both environmentally and socially. Thus, in order to remain
legitimate, firms are more than ever focussed on their environmental and social footprint
(Galbreath, 2011; Porter & Kramer, 2006). However, why some firms perform better than others
remains a point of debate. Upper echelon theory focuses in this debate on the top management
level of the organization, and states that executives act in a specific way when facing various
strategic situations due to their personalized interpretations, which are the result of their values,
personalities and experiences (Hambrick, 2007; Hambrick & Mason, 1984). This theory is based
on the assumptions that executives are boundedly rational (Cyert & March, 1963), and that the
organization is a reflection of its top management (Hambrick & Mason, 1984). It, therefore,
follows that executives’ characteristics might influence their strategic decisions and consequently the firm’s CSR outcomes. This relationship is believed to be more pronounced at the group level
than the individual executive level, such as a CEO, because leading a complex organization is
- 6 - a group which has substantial influence and is part of the ultimate strategic decision making
process within the firm (Galbreath, 2011). Thus,the board of directors is believed to be a
powerful force within the company to achieve strategic change, which also affects the
environmental and social performance of the firm. This relationship has been given more
attention in recent literature, however there still remains much to be explored (Horváthová, 2010;
Post, Rahman & Rubow, 2011).
1. Gender
One such important characteristic which has received more and more attention over the years is
gender, which in terms of the board of directors captures the dispersion between males and
females (Terjesen, Sealy & Singh, 2009). Previous research has shown that women and men have
different ideas and values when it comes to sustainability issues. Jaffee and Hide (2000) show
that women are more likely than men to use care reasoning, which means that they are more
concerned with maintaining good relationships, respond to the needs of others and feel a
responsibility not to inflict damage or hurt. Furthermore, Rosener (1995) has demonstrated that
women are particularly good at solving problems and are better capable to deal with ambiguity
and conflict. This would entail, as shown by Biggins (1999), that women are better able to
understand and represent the needs of multiple stakeholders. When this is combined with the
results of Delmas and Toffel (2008), stating that the receptivity of managers to pressures from market constituents is positively associated with the firm’s adoption of environmental policies,
one can argue that women, who have the needs of different stakeholders in mind, will positively
influence environmental performance. Besides, other research has shown that compared to men,
on average, women place more emphasis on self-transcendent values (Adams & Funk, 2012), have higher ethical standards (Betz, O’Connell & Shepard, 1989), and are outside of work more
- 7 - active in terms of community service and philanthropy (Groysberg & Bell, 2013). This
particularly highlights the influence of women on the firm’s social performance. All in all, it is
logical to assume that female board members will place greater emphasis on environmental and
social responsibility issues than their male counterparts (Galbreath, 2011; Post et al., 2011; Post,
Rahman & McQuillen, 2014).
While some research has been done on this particular relationship, there are
inconsistencies in terms of the results (Galbreath, 2011). While some researchers, such as Post et
al. (2011), find a positive relationship between more female board members and environmental
corporate social responsibility (ECSR) (Post et al., 2014; Bear, Rahman & Post, 2010), Galbreath
(2011) finds a non-significant relationship. However these two studies were conducted in
different countries, the United States and Australia respectively, and differed in the industries that
were examined. Thus, it is plausible that these inconsistencies stem from a difference in sample
and method of analysis. Therefore, there is still a need for further research when it comes to the
different industries and countries where this relationship may be present, in order to show that in
general women actually do have an effect on the environmental and social performance of a firm.
What is more, another important idea to take into consideration when thinking about the
effect of women on the board of directors is critical mass theory, which highlights the idea of
strength in numbers (Konrad, Kramer & Erkut, 2008). A lone woman on the board of directors
will represent a minority voice, and might, due to the social pressure for conformity, not be taken
seriously (Asch, 1955). The study of Bear et al. (2010) has shown a positive linear relationship
between the number of women on the board of directors and CSR strength ratings. It, therefore,
follows that the more women are on the board of directors, the likelier they are of being heard
- 8 - hopefully provide an answer to the general (overall) effect of female board members on a firm’s
environmental and social performance, irrespective of industry or country.
2. Power
However, while previous studies have found a relationship between gender and CSR
performance, the different aspects that might influence this relationship have remained largely
elusive (Galbreath, 2011, Ioannou & Serafeim, 2012). While, the power of the board of directors
has repeatedly been mentioned as a possible factor affecting the making of strategic decisions and policies, thus influencing the firm’s various outcomes, it has as of yet not been linked to the
relationship of gender and CSR performance (Chin, Hambrick & Treviño, 2013). It stands to
reason that when the board of directors has a lot of power in affecting the decisions of the firm,
and more women are present within the board, that their opinions and perspectives might
influence the firm’s CSR policies and performance more drastically than when the board has little
to say over the firm’s direction to begin with. In other words, just because women are on the
board of directors, it does not follow that they will significantly influence the firm’s CSR
outcomes, because the board of directors might lack the power to affect any important or radical
changes.
3. National culture
Another possible factor which may influence the relationship between gender dispersion on the
board of directors and the environmental and social performance of the firm is the company’s
national culture. Culture serves as a reference frame that facilitates the interpretation of
significant common events experienced by the members of that group or community (House,
- 9 - conditions most social practices and processes (Carrasco, Francoeur, Labelle, Lafarga and
Ruiz-Barbadillo, 2015). Besides, as mentioned previously, the experiences and values of board
members influences the way they think, act and interpret different issues (Hambrick, 2007). It
therefore follows, that the behaviour of board members towards their female counterparts and their input, is affected and framed by the firm’s national culture.Hofstede’s measure of
masculinity represents the degree of separate gender roles, where men are assertive and focused
on material success, women should be modest and concerned with personal relationships and
solidarity (Hofstede, 2000). This means that men and women are believed to have certain roles
and behaviours that are appropriate. Research by An and Kim (2006) supports this logic, by
showing that women in Korean advertising campaigns were almost exclusively presented in
recreational and family roles. These distinct roles can create stereotypes (Carrasco et al., 2014), which in turn may result in men not believing a woman’s voice to be appropriate in the setting of
the board of directors and thus are unwilling to listen, irrespective of the issue addressed. Or
perhaps men will think the women are led by their emotions in addressing the issue of CSR and
are of the opinion that these have no place at corporate level, once again resulting in women
being ignored or undervalued.
The same idea applies to the measure of power distance, representing the degree of
acceptance that a country has concerning the unequal distribution of power within organizations
and institutions (Hofstede, 2000). This inequality in power may also be present in the male and
female relationships, where the man has a patriarchal role and beliefs he has the right to control
his women (Mostafa, 2005). Furthermore, patriarchal cultures institute men as having structural
control over political, legal, religious and economic institutions (Glick & Fiske, 1997). Thus,
boards of firms from countries with the beliefs that women should depend on men (Baxter &
- 10 - equals and value their input in the same way. All in all, the national culture of the firm may well
have an important effect on how the women on the board of directors are perceived, treated and
listened to, which can influence the likelihood that female initiatives are realised.
Thus, the goal of this research is to examine the possible relationship of how the dispersion of
males and females on the board of directors affects the environmental and social performance of
a firm. In addition, this thesis endeavours to explore the possible moderating effect of the variables “power of the board of directors” and the “national culture of the firm” on this
relationship.
The findings of this research are likely to be both valuable to theory as well as practice. In
terms of theory an important link has been forged between different aspects influencing the
relationship between female board representation and the environmental and social performance
of firms. This provides further information on the possible outcomes of certain board structures
and may provide new avenues for future research. In terms of practice, this research may offer
insight into the likely effects of board formations and how female directors elicit different
outcomes. This could guide firms in how to create the best boards, so that they are more likely to
reach their desired performance goals, be they environmental or social (Galbreath, 2011).
3. Theoretical framework
1. Gender dispersion
Previous studies have shown that men and women differ markedly in many aspects, for example
values, focus and abilities (Post et al., 2011; Galbreath, 2011; Griskevicius et al., 2012). These
differences are reflected in the way men and women act and reason in their jobs. Applying this
- 11 - differing strategic directions and consequently performances (Post et al., 2014). Following
Griskevicius et al. (2012), the evolutionary theory states that, compared to men, women have a
long term focus, because of sexual selection and the high level of parental investment (Darwin,
1859; Darwin, 1871). The sex with the lower obligatory parental investment in offspring, will be
more competitive. In the case of humans, this is true for men, also leading them to be more risk
taking and aggressive (Griskevicius et al., 2012). This difference in focus, means that men have
steeper future discount rates, thus are more willing to engage in wasteful consumption and
deplete environmental resources in order to gain short term profits (Sundie et al., 2011; Low,
1996; Griskevicius et al., 2012). This entails, as shown by Diamantopoulos, Schlegelmilch,
Sinkovics, and Bohlen (2003), that compared to men women are generally more concerned with
environmental issues and more likely to take action to reduce perceived risk (Post et al., 2011).
Furthermore, another cause for the differences between men and women, particularly in
terms of moral reasoning, is rooted in gender socialization. Traditionally, boys are taught games
that value fairness and rules, while girls are taught those that value inclusion and the
acknowledgement of feelings. Which would explain, why adult men have other ethical values in
the workplace than their female counterparts (Gilligan, 1982). The article by Jaffee and Hide
(2000) has shown support for this reasoning, by stating that women are more likely to use care
reasoning, which entails maintaining relationships, responding to the needs of others and feeling
a responsibility to avoid inflicting hurt. A large body of literature, even goes so far as to say that
women, on average, are more likely to identify situations requiring ethical judgement and behave
ethically (Post et al., 2011; Post et al., 2014; Betz et al., 1989). This tendency of women to build
relationships and respond to the needs of others, allows them to build stronger relations with
multiple key stakeholders (Biggins, 1999). Which in turn allows firms both to better understand
- 12 - concerning sustainability (Galbreath, 2011). Lastly, women emphasize self-transcendent values
and are more interested in activities outside of work regarding philanthropy and community
service (Groysberg & Bell, 2013). All in all, women compared to men, are more concerned about
the welfare and needs of others, making them more attuned to the plight of society and the
environment as a whole. Thus, when female board members are provided with the opportunity to
influence strategic direction they will, guided by their ethical values and stakeholder relations
(Hambrick, & Mason, 1984), put more emphasis on addressing environmental and social issues,
which in turn will improve their respective performance.
However, this may not always be the case, due to lack of support or even open resistance
of men against the female perspective. Women still represent a minority voice within most
corporate boards, which poses the risk of them not being heard or valued by other board members
(Asch, 1955; Post et al., 2014). Social or group pressure encourages the minority to conform to the majority’s opinions (Asch, 1955). What is more, when minorities are widely underrepresented
in a group, they become tokens for their minority group, leading them to be perceived as less
competent and be awarded lower status than the majority members (Post et al., 2011). The same
goes for women, who as tokens are subject to more scrutiny and on-the-job pressure than their
male counterparts, because they are more visible to the rest of the group, which increases
performance pressure (Oakley, 2000). Another issue discussed by Oakley (2000), is the
double-bind faced by women in leadership positions, which entails that they must be tough and
authoritative just like men, to be taken seriously, but when acting too aggressively they will be perceived as “bitches”. The most troublesome double-bind for female board members is the
femininity/competency bind, which means that acting “feminine”, being sensitive to feelings and
basing decisions on the impact on relationships (Oakley, 2000), is associated with incompetence
- 13 - styles. The managing styles of women was characterised by men as less self-confident, less
analytical, less consistent, less emotionally stable and possessing poorer leadership skills than
men (Heilman, Block, Martell & Simon, 1989; Broverman, Vogel, Broverman, Clarkson &
Rosencrantz, 1972). Furthermore, where men were believed to be aggressive, unemotional,
independent, dominant, active, objective, logical, competitive, worldly, self-confident and skilled
in business, all competence related traits, women on the other hand were typically seen as
possessing the opposite traits, and thus viewed as incompetent (Oakley, 2000; Heilman et al.,
1989; Broverman et al., 1972). All these tokens and stereotypes put pressure on women to prove
themselves and voice their opinions. This thesis argues that the more women are on the board of
directors, the less likely they will be seen and treated as a minority group, thereby lowering the
previously mentioned form of tokenism. The consequent increase in diversity combined with the
participative and communal traits of women, facilitates open conversation and a broader
perspective (Bear et al., 2010). Thus, the more women are on the board of directors the likelier
they are heard and valued, improving their input on environmental and social issues and their
respective performance (Galbreath, 2011). Therefore, the first proposition is as follows.
Proposition 1: There is a positive relationship between the percentage of women on a firm’s board of directors and its (a) environmental and (b) social performance.
2. The power of the board of directors
One of the moderator variables used is the power that the board of directors have in terms of
influencing the strategic direction of the firm. As previous research has shown (Goodstein,
Gautam & Boeker, 1994; Coffey & Wang, 1998; Post et al., 2011), power is an important factor
- 14 - change, regardless of the number of women present on the board. So, if women want to change the firm’s environmental and social performance, they will need the board to have a say in the
firm’s strategic decisions to begin with. Power of the board is operationalized by two different
variables, namely board size and CEO duality (Post et al., 2011; Goodstein et al., 1994; Finkelstein & D’Aveni, 1994).
A large body of literature discusses the effect that board size can have on the strategic
decision making process (Goodstein et al, 1994; Post et al., 2011, Eisenberg, Sundgren & Wells,
1998). A larger board provides the advantage of having an increased pool of resources and
expertise (Pfeffer, 1972), as well as the fact that each additional board member comes with their
own unique experiences and values. Thus, larger boards have a wide variety of perspectives
(Pearce & Zahra, 1991), which can translate itself in a strong and well thought out strategic
course of action. However, these advantages can quickly turn into disadvantages, because the difficulty of reaching a consensus does not only lower the CEO’s influence, but that of the entire
board as well. This is due to the fact that larger decision making groups are less cohesive (Shaw,
1981) and pose serious coordination problems due to the large number of interactions and
opinions (Gladstein, 1984; Post et al., 2011). Besides, the large number of individuals increases
the likelihood that factions or coalitions are developed leading to a surge of group conflict (O’Reilly, Caldwell & Barnett, 1989), again making it harder to reach a consensus on various
decisions. The nature and context of the strategic decisions often faced by the board of directors
are complex and ambiguous (Eisenhardt, 1989), which according to Olson (1982), are likely to be
even more unfavourably affected by large group dynamics. These complex issues often divide
board members, thereby creating an environment of conflict which will delay or even inhibit
important changes (Olsen, 1982). This lack of cohesion will, in turn, lower the power of the
- 15 - impact a board’s power, due to the lack of unity, and thereby make it unlikely that the board will
affect any important changes. So, whether there are women on the board with different views on
the strategic direction of the firm is irrelevant, because the board lacks the capacity to affect
strategic change altogether. Therefore, the expectation is that the larger the board becomes, the
more difficult it will be for the board to impact environmental and social performance,
irrespective of female representation.The second proposition is therefore as follows.
Proposition 2: The higher the number of board members, the weaker the relationship between the percentage of women on a firm’s board of directors and its (a) environmental and (b) social performance.
The other variable operationalizing power is CEO duality, which captures whether the CEO is
also chairman of the board of directors (Post et al., 2011). Finkelstein & D’Aveni (1994) discuss
CEO duality as a double-edged sword, due to the balancing act of avoiding CEO entrenchment,
while upholding the unity of command. It is the job of the board of directors to monitor and
ensure that the interest of the CEO and shareholders are aligned (Fama & Jensen, 1983). In order
to do this the board needs to have the power to actually control the CEO, and where need be
change the strategic direction. However, when the CEO is given the position of chairman on the
board of directors, the power dynamics shift, and the CEO is awarded undivided formal authority,
which can lead to opportunistic and inefficient behaviour at the expense of firm performance
(Jensen & Meckling, 1976). Besides, the CEO is able to dictate both the content and the agenda of the board meetings (Finkelstein & D’Aveni, 1994). Now that the CEO has the power to
influence the board of directors more directly, it is likely that he or she will steer the board towards a course of action that is in the CEO’s best interest or that he/she thinks best. Thereby,
- 16 - increasing the likelihood that the opinions of other board members will carry less weight, and in
combination with their loss of independence in terms of monitoring and disciplining (Bear et al.,
2010), this will result in a clearly diminished level of power of board members (Mallette &
Fowler, 1992). Thus, it becomes largely irrelevant that the female directors have different ideas about the firm’s strategic direction in terms of environmental and social performance, because the
board has no real influence in the strategic decision making process anyway. The third
proposition is therefore as follows.
Proposition 3: When the CEO sits on the board of directors as chairman, the relationship between the percentage of women on the board of directors and the firm’s (a) environmental and (b) social performance will be weaker.
3. The national culture of the firm
The second moderator variable to be analysed in this research is the national culture of the firm.
Previous research by Ioannou and Serafeim (2012) has already shown culture to have an effect on
the managerial decision making process concerning corporate social performance. However, the link between a firm’s national culture and how it influences the likelihood of female board
members affecting environmental and social change has not yet been looked at. Thus, this
research looks at how the cultural traits of the firm’s country of origin affect the possible
relationship between the proportion of women on the board of directors and environmental and
social performance. The national culture of the firm is operationalized by Hofstede’s masculinity
and power distance dimensions (2000).
The masculinity dimension describes to what extend a culture has strong and distinct
- 17 - material success. Thus, he should embody the values for performance, dominance, success,
assertiveness and competitiveness. Women, on the other hand, are supposed to be modest, close
to their emotions, and show concern for personal relationships, the quality of life and solidarity
(Carrasco et al., 2015, Oakley, 2000). These clearly distinct roles and values ascribed to men and
women create gender stereotypes that make generalizations about the characteristics and
competency of each gender group. These stereotypes endorse a strong preference or bias towards
men in the appraisal of their merits and skills, especially in leadership positions (Sealy, Doldor &
Vinnicombe, 2009; Konrad et al., 2008; Oakley, 2000). What is more, the role of men as income
provider and women of homemaker, advocates the idea that men should be dominant and
autonomous, while women should be humble and nurturing (Konrad, Ritchie, Lieb & Corrigall,
2000). However, this poses problems for women in top management positions, because on the
one hand they should act deferential and sensitive to feelings, but in doing so are perceived as
weak and incompetent (Oakley, 2000; Konrad et al., 2000). While, on the other hand, when
female board members act tough and authoritative, they are often perceived as too aggressive
(Oakley, 2000). Unsurprisingly then, women are stereotypically characterized as less competent
managers than men (Heilman et al., 1989; Broverman et al., 1972). Therefore, the expectation is
that boards with a strong masculine background belief women to be out of place on the board of directors, particularly if they express their “feminine” values, and are thus less likely to heed their
ideas and opinions, resulting in lower environmental and social performance. Hence, the
- 18 -
Proposition 4: The higher the firm’s score on the masculinity dimension, the weaker the proposed relationshipbetween the percentage of women on the board of directors and the firm’s (a) environmental and (b) social performance will be.
The other dimension used in this thesis, namely power distance, refers to the degree that society
accepts an unequal distribution of wealth and power (Hofstede, 2000; Carrasco et al., 2015).
Power distance values, as defined by Carl, Gupta and Javindan (2004), are the extent to which
societal members believe that power should be concentrated in the hands of only a few people,
and that these people should be obeyed without question and afforded special privileges
(Waldman et al., 2006). Such societies are prone to manipulative use of power and lack of equal
opportunities for minorities, including women (Carl et al., 2004). Furthermore, Waldman et al.
(2006) have shown that high power distance cultures have a negative effect on corporate social
performance, because the managers are more inclined to care about their own self-interests, than
the concerns of society (Ioannou & Serafeim, 2012). Hence, male managers are less focused on
building long term relationships with stakeholders, and feel less responsible for the welfare of
society (Waldman et al., 2006). This is the exact opposite of female characteristics and values
(Griskevicius et al., 2012; Biggins, 1999), which will in all likelihood create conflict between
board members on environmental and social issues. Hence, it seems unlikely that boards from
high power distance societies should champion the ideas put forth by a minority voice, such as
female board members, in order to improve the firm’s social and environmental performance, a
long term and societal goal. Lastly, Cohen, Pant and Sharp (1996) find that managers from high
power distance cultures are more inclined to view questionable business practices as ethical, in
comparison to low power distance societies. All in all, it seems reasonable to assume that the
- 19 - hailing from high power distance societies, because their primary focus is their own welfare
instead of the greater good, resulting in fewer social and environmental initiatives and lower
performance. Hence, the fifth and final proposition.
Proposition 5: The higher the firm’s score on the power distance dimension, the weaker the proposed relationshipbetween the percentage of women on the board of directors and the firm’s (a) environmental and (b) social performance will be.
So, the proposed model analysed in this thesis is shown below in figure 1.
Figure 1: Proposed model
Control variables: - Firm size - Industry - Debt ratio - Slack resources - Geographical region
- 20 - 4. Methodology
To answer the proposed research question and examine the possible moderating effects of power
and national culture of the firm, a quantitative, empirical research design is applied. The
information on the necessary variables is gathered using the two databases from Thomson
Reuters, namely ASSET4 and Worldscope, because these have all the information necessary for
answering the research question and are relatively easy to access (Ioannou & Serafeim, 2012),
which is an important factor to keep in mind when the research is constrained for time, as this
thesis was (Saunders, Lewis & Thornhill, 2012, pp. 318-323). What is more, they have been
widely used by researchers, which makes it possible to compare results and improve the overall
validity of the data. The variables taken from ASSET4 are gender, and the environmental and
social performance of the firms. Worldscope provides data concerning the firm’s country of
origin as well as the company fundamentals to measure the control variables, such as sales to
measure size, industry, debt ratio and slack resources. Following the data gathering, a regression
analysis is performed with the collected data, as well as other statistical tests, such as normality
tests, in order to make sure that all the requirements for parametric testing are met and the data is
representative of the sample (Saunders et al., 2012, pp. 524-525).
1. Sample
The firms to be examined in this research are taken from the Global Fortune 500. Until now
research has been done in a limited number of industries and has focused on either the United
States or Australia. Various researches have therefore, expressed the importance of including a
wider range of industries and countries in future research (Galbreath, 2011; Bear et al., 2010;
Post et al., 2011; Post et al., 2014). Thus, this study responds by looking at firms which operate
- 21 - increasing the generalizability of the results (Saunders et al., 2012, pp. 265). The inclusion of
different industries means that all manner of firms are analysed, with widely differing
environmental and social issues and impacts. For instance, industries working with fossil fuels
have a markedly different environmental footprint than internet service industries, and therefore
face other stakeholder pressures (Spar & La Mure, 2003). The same argument can be made for a firm’s country of origin, an originally Chinese company may face different stakeholder pressures
than an American firm, due to variations in culture and regulations. So, the Global Fortune 500
provides a geographically and characteristically balanced sample, which is expected to be a fair
representation of the population.
The data concerning the independent, moderator and control variables were collected for
the year 2013, from the Global Fortune 500 companies of that year. The environmental and social
performance of said companies was collected for 2014, allowing for a one-year time lag. The
Global Fortune 500 companies were then checked for the availability of environmental and social
performance, and relevant data concerning the moderator variables in ASSET4. A total of 151
firms were eliminated due to missing information, resulting in the final list of 349 firms.
2. Independent variable
The gender dispersion on the board of directors is operationalized as the number of female
directors divided by the total number of board members, which is named the proportion of
women on corporate boards (POWB). This measure has been used in previous studies by
Carrasco et al. (2015) and Galbreath (2011), thereby increasing reliability and making it possible
- 22 -
3. Dependent variables
The two dependent variables in this research are environmental and social performance. There
are many different aspects within these two pillars that are important. This research has followed
the example of Ioannou and Serafeim (2012) in their included topics, which were based on
extensive previous literature, and provides the opportunity of comparing results. The categories
used for operationalizing environmental performance are resource reduction, emission reduction
and product innovation. While, the categories included in the social performance pillar are
employment quality, health and safety, training and development, diversity, human rights,
community and customer/product responsibility (Ioannou & Serafeim, 2012, pp.863-864). The
environmental and social performance are measured as the average performance of their
respective categories, resulting in the environmental and social scores taken from ASSET4. There
is as of yet no clear and definite ranking order concerning these categories, and therefore the
categories are all believed to be of equal importance, resulting in the equal weight assigned to
each category (Ioannou & Serafeim, 2012; Waddock & Graves, 1997, Waldman et al., 2006).
The data is taken from ASSET4 for the year 2014. In order to observe an effect of the gender
composition of the board of directors on environmental and social performance, a one year time
lag was incorporated in the research design. This particular time-frame was chosen, because there
is no consensus in current literature on what the exact time frame should be, and it would be
likely that after a second year, new board members are appointed and consequently affect the firm’s environmental and social performance. Bear et al. (2010) used the same time lag, under the
assumption that board members should be in their roles for some time in order to affect CSR
decisions. Therefore, a one year time lag is deemed appropriate, to allow female board members
time to influence the other board members on environmental and social decisions, yet minimize
- 23 -
4. Moderator variables
There are four moderator variables analysed in this research, two concerning the power of the
board of directors, namely board size and CEO duality, and two in terms of the national culture of
the firm, namely masculinity and power distance. Both the power moderator variables were
collected for the year 2013 from ASSET4, while the level of masculinity and power distance
were derived from Hofstede (2000) for the stated nationality of the firm found in Worldscope.
The power of the board of directors measures their ability to influence the strategic
direction of the firm. Board size is incorporated and measured as the total number of board
members. The other variable influencing power of board members, is CEO duality and measures
whether the CEO presides over the board as chairman. To use CEO duality in the analysis, a dichotomous dummy variable is created, where “0” means that the CEO is not the chairman of
the board of directors, and “1” entails that he/she is.
The other moderator variables concern the national culture of the firm and the
accompanying cultural differences. The national culture is operationalized through linking the firms’ countries of origin, found in Worldscope, to their respective levels of masculinity and
power distance, taken from Hofstede (2000). Both masculinity as well as power distance can
have a value between 1 and 100, where a low value means a feminine culture and low power
distance, and a high value means a masculine and high power distance culture (Hofstede, 2000).
These cultural dimensions are used, because they are still believed to be the best proxy for culture
(Carrasco et al., 2015). Furthermore, only the dimensions masculinity and power distance are
analysed in this study, because research by Carrasco et al. (2015) and Waldman et al. (2006), has
shown them to have the most significant impact on the appointment of female board members
- 24 -
5. Control variables
Five variables have been chosen to control for any variations in outcome, namely firm size,
industry, debt ratio, slack resources and the firm’s geographical region. All the aforementioned
variables have been shown in previous literature to affect the environmental and social
performance of a firm, thus making them more reliable (Galbreath, 2011; Post et al., 2011; Post
et al., 2014; Bear et al., 2010; Carrasco et al., 2015; Barnett & Salomon, 2012). All the control
variables are gathered for the year 2013 from Worldscope.
Firm size is controlled for, because even though all firms in the sample are relatively
large, there are still sizeable differences, which may affect their environmental and social
performance. For instance, a larger organization has more resources to engage in CSR and thus
might have a better environmental and social performance than a smaller firm. Firm size is measured as a firm’s total assets, following Galbreath (2011) and Ioannou and Serafeim (2012).
Another control variable included in this research is industry. Since, there is a large variation of
industries within the chosen sample, it is important to control for any effect on environmental and
social performance caused by differences in structure, technology, environmental issues and
stakeholder pressure (Post et al., 2011). The general industry classification of Worldscope
distinguishes between six different industry classifications, namely industrial, utility,
transportation, bank/savings and loan, insurance and other financial. In order to control for
industry in the analysis, various dummy variables were created. The reference group is the
industrial industry category, because this was the largest group in the sample (Hardy, 2013).
Another variable which is controlled for is debt ratio, measured as total debt/total assets, because
when a firm has a lot of debt it is unlikely to engage in extensive environmental and social
initiatives (Barnett & Salomon, 2012; Waddock & Graves, 1997). The same logic applies for
- 25 - likely to engage in environmental and social initiatives (Waddock & Graves, 1997; Post et al.,
2011). Slack resources are measured by net profit (Galbreath, 2011). The geographical region of
the firm might also influence the environmental and social performance of firms, because
European countries, for instance, have different environmental standards than Asian countries to
which firms need to adhere (Zadek & MacGillivray, 2007; Post et al., 2014). These differences
are controlled for by incorporating the continent from which the firm originates. To use this
categorical variable in the analysis, dummy variables were created. The chosen reference group is
the continent Europe, because it represented the largest group in the sample (Hardy, 2013).
5. Results
5.1 Descriptives
In this study, data was collected from all Global Fortune 500 companies of 2013, but due to
missing data the final list used in the analysis was brought down to 349 firms. However, the
normality analysis required an elimination of extreme cases, which further reduced the number of
firms ready for analysis down to 329. For those 329 firms all relevant data for the independent,
dependent, moderator and control variables was available.
Four categorical variables were used in this research, namely moderator variable CEO
duality, country of origin, and control variables industry and geographical region. Dichotomous
dummy variables were created for each, with the exception of country of origin, because this
variable was only used to find data on the moderator variables masculinity and power distance as
well as form the basis for the control variable geographical region. In terms of CEO duality, there were more firms with CEO’s who were not chairman of the board (62.7%), than those who were
(37.4%), as shown in table 1. Most firms in the sample originate from the United States of
- 26 - Mexico and Thailand were some of the least heavily represented countries (0.3%). Table 1 shows
the complete list of home countries represented in the sample. The majority of the firms operated
in the industrial industry (67.2%) followed by bank/savings and loan (10.9%), while on the other
hand the fewest companies operated in other financial (1.8%). Lastly, in terms of geographical
region, most firms hail from Europe (33.7%) followed closely by North America (33.4%), while
South America (1.8%) and Oceania (2.4%) are the least represented.
Table 1. Descriptive statistics for categorical variables
Variable Level N %
CEO duality CEO is not Chairman 206 62.6
CEO is Chairman 123 37.4
Country of Origin Australia 8 2.4
Austria 1 .3 Belgium 3 .9 Brazil 6 1.8 Canada 7 2.1 China 20 6.1 Denmark 1 .3 Finland 1 .3 France 24 7.3 Germany 18 5.5 Hong Kong 8 2.4 Hungary 1 .3 India 6 1.8 Ireland 1 .3 Italy 7 2.1 Japan 46 14.0 Luxembourg 1 .3 Mexico 1 .3 Netherlands 7 2.1 Norway 1 .3 Russia 4 1.2 Saudi Arabia 1 .3 Singapore 1 .3 South Korea 3 .9 Spain 7 2.1 Sweden 2 .6 Switzerland 9 2.7
- 27 - Taiwan 3 .9 Thailand 1 .3 Turkey 1 .3 United Kingdom 27 8.2 United States 102 31.0 Industry Industrial 221 67.2 Utility 25 7.6 Transportation 11 3.3
Bank/Saving & Loan 36 1.9
Insurance 30 9.1
Other financial 6 1.8
Geographical region Europe 111 33.7
North America 110 33.4
South America 6 1.8
Asia 94 28.6
Oceania 8 2.4
Besides the four categorical variables, nine continuous variables have been used in this research,
specifically the dependent variables environmental and social performance, the independent
variable POWB, the moderator variables board size, masculinity, power distance, and the control
variables net income, total assets and debt ratio. The descriptives of these variables are shown
below in table 2.
The dependent variable environmental performance ranged between 8.46 and 97.08, with
a mean of 78.58 and a standard deviation of 23.38. While, the other dependent variable, social
performance, ranged between 3.66 and 96.67, with a mean of 75.22 and a standard deviation of
22.50. The values for the POWB, or proportion of women on the board of directors, ranged
between 0 and 46.67, with a mean of 15.95 and a standard deviation of 11.48.
Then concerning the moderator variables, board size values ranges between 5 and 23,
with a mean of 12.75 and a standard deviation of 3.55. The range of masculinity, however,
showed a range of 5 to 95, with a mean of 62.69 and a standard deviation of 17.67. Lastly, power
- 28 - Finally, the control variables net income, total assets and debt ratio. Net income values
ranged between -124567000 and 426118222, with a mean of 31770867.59 and a standard
deviation of 76334993.93. The values of total assets were ranging between 7078719 and
21325640422, with a mean of 1626326869.82 and a standard deviation of 3735307454.99. And
debt ratio consisted of values between 0 and 0.98, with a mean of 0.24 and a standard deviation
of 0.15.
Table 2. Descriptives statistics for continuous variables
5.2 Normality
In order to measure the adequacy of the nine continuous variables, a normality test was
conducted for environmental and social performance, POWB, board size, masculinity, power
distance, net income, total assets and debt ratio. The descriptives of the normality tests can be
found in the appendix, specifically table 1. The outcome of the Kolmogorov-Smirnov test can be
found in table 2 of the appendix.
First of all, the dependent variable environmental performance had a substantial negative
skewness of 1.794 and a positive kurtosis of 1.993, meaning that the distribution of values was
clustered to the right and peaked in the centre, as shown in the histogram (appendix, graph 1).
The variable was transformed using the formula Y = Log10 (K – X), improving skewness to
0.705 and kurtosis to -0.564. Despite the transformation, the variable Log Environmental
Variable N Min Max Mean Std. Deviation
Environmental performance 329 8.46 97.08 78.58 23.38
Social performance 329 3.66 96.97 75.22 22.50
POWB 329 .00 46.67 15.95 11.48
Board size 329 5 23 12.75 3.55
Masculinity score 329 5.00 95.00 62.69 17.67
Power Distance score 329 11.00 95.00 49.37 15.52
Net Income 329 -124567000 426118222 31770867.59 76334993.93
Total Assets 329 7078719 21325640422 1626326869.82 3735307454.99
- 29 - performance still had a significantly different distribution than normal, as shown in the
Kolmogorov-Smirnov test. However, the skewness and kurtosis are within the boundaries of +2
and -2, set by George and Mallery (2010) for an acceptable normal distribution, and thus are not
expected to have a drastic effect on the results. The Q-Q plot (appendix, graph 2) supports this
belief, by showing a slightly skewed and platykurtic distribution, but still reasonably close to a
straight line.
Besides environmental performance, social performance also showed a skew to the right
(-1.379) and a relatively peaked centre (1.050). However, the logarithmic transformation similar
to environmental performance, improved the distribution drastically to a negligible negative
skewness (-0.067) and a somewhat flat centre (-0.825), as shown in the histogram (appendix,
graph 4). Nonetheless, the Kolomogorov-Smirnov test was still significant at a level of 0.032,
meaning that the distribution is significantly different from normal. Yet, the Q-Q plot (appendix,
graph 5) shows a distribution close to a straight line and the skewness and kurtosis of Log Social
performance are within the boundaries set by George and Mallery (2010), thus the expectation is
that distribution is close to normal and will not drastically affect the outcome.
Then, in terms of the independent variable POWB (proportion of women on the board of
directors), there is slight skew to the left (0.152) and a relatively flat centre (-0.847). No
transformation could change the distribution for the better. However, despite the moderate
kurtosis and the significant Kolmogorov-Smirnov test, the Q-Q plot (appendix, graph 8) shows an
almost straight line. Furthermore, the distribution is well within the boundaries of George and
Mallery (2010), thus the expectation is that POWB is close to a normal distribution.
While POWB did not respond well to transformation, moderator variable board size did.
- 30 - 0.704 to 0.366, and changed kurtosis from 0.134 to -0.135. The histogram (appendix, graph 10)
thus shows a small skew to the left and a slightly flat centre. Even though the
Kolmogorov-Smirnov test is significant, the Q-Q plot (appendix, graph 11) roughly exhibits a straight line and
the values are within the limits of George and Mallery (2010). Consequently, the distribution is
expected to be close to normal.
Moderator variable, masculinity showed a negative skew of 0.204 and a positive kurtosis
of 1.340. No transformation or deletion of extremes seemed to improve the distribution, and since
the values are within the boundaries of +2 and -2 (George & Mallery, 2010), they are not
expected to greatly affect the results. Supported, by a mean of 62.690 and a 5% trimmed mean
value of 63.310, indicating that the extreme scores did not have a significant impact on the mean
(appendix, table 1). Besides, the Q-Q plot is somewhat close to a straight line (appendix, graph
14). So, despite the significantly different distribution from normal, as shown by the
Kolmogorov-Smirnov test, the expectation is that this will not drastically affect the results.
The last moderator variable, namely power distance did, somewhat, improve from the
transformation Y = √(𝑋). Where there previously was a moderate clustering of values to the left
(0.863) and a slightly flat centre (-0.144), Sqrt Power Distance had smaller positive skewness
(0.581) and a marginally flatter centre (-0.268) (appendix, graph 16). The Q-Q plot (appendix,
graph 17) presents a reasonably straight line. Nonetheless, the Kolmogorov-Smirnov test is
significant, indicating a distribution that is not normal. Yet, the skewness and kurtosis are well
within stated limits (George & Mallery, 2010), and thus not expected to greatly impact the
analysis.
Then in term of control variables, firstly net income. Before transformation, there was a
- 31 - slightest of improvements after eliminating the most extreme cases from the data, was the square
root formula. Sqrt Net Income had a marginally better skewness (2.663) and kurtosis (7.253).
These values far exceed the boundaries posed by George and Mallery (2010), indicating a
significantly different distribution than normal, as supported by the Q-Q plot (appendix, graph
20) and the Kolmogorov-Smirnov test. However, no other transformation or the further
elimination of extreme values has improved this distribution. So, even though everything has
been done to minimize skewness and kurtosis, Sqrt Net income is still different from a normal
distribution. Nonetheless, it is the best possible representation of net income.
The control variable total assets also had an extreme tendency to cluster to the left (3.362)
and a strong leptokurtic distribution (11.548). The elimination of extreme values and a
logarithmic transformation improved the distribution markedly to a small skew to the left (0.469)
and a somewhat flat centre (-0.858), as shown in the histogram (appendix, graph 22). According
to the Kolmogorov-Smirnov test the distribution is still significantly different from normal,
however the Q-Q plot (appendix, graph 23) shows a reasonably straight line and the distribution
values are now well within the boundaries set by George and Mallery (2010). So, the Log Total
Assets distribution is not expected to drastically affect the results.
Finally, the logarithmic transformation of debt ratio, markedly improved the skewness
from 0.933 to 0.490, and the kurtosis from 1.974 to 0.496. This translates into a slightly skewed
distribution to the left and somewhat peaked centre, as shown in the histogram (appendix, graph
25). Even though, the Kolmogorov-Smirnov test signifies a significantly different distribution
from normal, the Q-Q plot (appendix, graph 26) is close to a straight line and the skewness and
kurtosis are within the +2 and -2 boundaries, thus signalling a close to normal distribution. Thus,
the slightly abnormal Log Debt Ratio distribution is not expected to have a drastic effect on the
- 32 -
5.3 Correlations
Correlations between the various variables used in this research, were investigated using the
Pearson correlation coefficient. As previously mentioned, CEO duality is a dichotomous dummy variable where “1” signifies that the CEO of the firm is also the chairman of the board of
directors, while “0” means the CEO is not. Dummy variables were also created for the control
variables industry and geographical region, where the categories industrial and Europe served as
the reference groups, because they each were the largest group in the sample (Hardy, 2013). In
the following paragraphs only some of the most notable significant correlations will be discussed.
The overall results of the correlation analysis are presented in table 3 at the end of this section.
Firstly, the independent variable POWB is negatively and significantly correlated to both
Log Environmental performance (r = -0.213, N = 329, Sig. = 0.000) and Log Social performance
(r = -0.223, N = 329, Sig. = 0.000). Both correlations contrast with the hypothesized direction of
the relationship between the proportion of women on the board and environmental as well as
social performance. Log Environmental performance and Log Social performance exhibited a
positive and significant correlation to each other (r = 0.656, N = 329, Sig. = 0.000), which might
be cause for concern in terms of multicollinearity. However, the threshold for possible
multicollinearity, presented by Field (2013, p. 325), is 0.8, which means that the two variables are
unlikely to measure the same construct.
Secondly, the moderator variable Sqrt Board size is significantly correlated to the
independent variable POWB (r = 0.192, N = 329, Sig. = 0.000), as well as to the dependent
variables Log Environmental performance (r = -0.159, N = 329, Sig. = 0.004) and Log Social
performance (r = -0.190, N = 329, Sig. = 0.001). This entails, that the larger the size of the board,
the higher the percentage of women on the board, but also lowering environmental and social
- 33 - any of the dependent and independent variables. On the other hand, the moderator variables
concerning national culture, namely masculinity and power distance, were. Masculinity was
significantly correlated with Log Social performance (r = 0.149, N = 329, Sig. = 0.007) and
POWB (r = -0.377, N = 329, Sig. = 0.000), but not to Log Environmental performance. Sqrt
Power Distance was significantly correlated to all three, specifically Log Environmental
performance (r = 0.198, N = 329, Sig. = 0.000), Log Social performance (r = 0.119, N = 329, Sig.
= 0.031) and POWB (r = -0.376, N = 329, Sig. = 0.000). So, when masculinity and power
distance are high, the percentage of women on the board is lower, and environmental and social
performance improve. None of the correlations are above 0.8, thus multicollinearity is unlikely to
be a problem.
In terms of control variables, there is a significant correlation between Log Total Assets
and the independent variable POWB (r = -0.407, N = 329, Sig. = 0.000), the dependent variable
Log Environmental performance (r = -0.127, N = 329, Sig. = 0.021), and the moderator variables
Masculinity (r = 0.387, N = 329, Sig. = 0.000) and Sqrt Power Distance (r = 0.398, N = 329, Sig.
= 0.000). Thus, when a firm’s total assets is high, they have a lower percentage of women on the
board of directors and a lower environmental performance. It is more likely that a firm has higher
total assets when it hails from a masculine and high power distance country. Furthermore, the
dummy variable for geographical region concerning Asia shows a significantly negative
correlation with POWB (r = -0.646, N = 329, Sig. = 0.000), meaning that proportionately fewer
women are on the board of directors in Asian firms. The same is true for South American firms (r
= -0.120, N = 329, Sig. = 0.029), however on the other hand North American companies have a
higher percentage of women on their boards (r = 0.274, N = 329, Sig. = 0.000). Additionally,
Asia is significantly correlated with Sqrt Power Distance (r = 0.644, N = 329, Sig. =0.000), just
- 34 - Sig. = 0.000). This is not surprising given that Asian countries are on average higher on power
distance. Lastly, the Bank/Savings & Loans industry was significantly correlated with Sqrt Board
Size (r = 0.174, N = 329, Sig. = 0.002) and CEO duality (r = -0.170, N = 329, Sig. = 0.002). So,
in the bank/savings and loans industry boards are larger and there is less likelihood of the CEO
also being the chairman of said board.
Even though, all the correlations are below 0.8 and thus unlikely to be cause for concern,
more subtle forms of multicollinearity might be overlooked. Therefore, to get a clearer idea, the
variance inflation factor (VIF) is used. According to Field (2013, p. 325), the VIF may not be
higher than 10 and the tolerance level should not be lower than 0.2, otherwise multicollinearity
might pose a problem. None of the variables used in this research has a VIF higher than 10 or a
tolerance level lower than 0.2. Consequently, multicollinearity is unlikely to pose a problem and
- 35 - Board Size (r = 0.174, N = 329, Sig. = 0.002) and CEO duality (r = -0.170, N = 329, Sig. =
- 36 -
5.4 Regression analysis
A hierarchical linear regression analysis was conducted to test the stated propositions and answer
the research question. In the regression Log Environmental performance and Log Social
performance were added as the dependent variables. The control variables, firm size, industry,
slack resources, debt ratio and geographical region were entered into the first model of the
regression. Then, in the second model, the independent variable POWB was included. For
propositions 2 to 5, concerning the moderating effects, the PROCESS application of Hayes
(2012) was used. Thus, model 1 for simple moderation was applied, as shown in appendix section
2 graph 1, where Y exhibits the dependent variable, X the independent variable, M the moderator
variable and XM shows the interaction effect.
The results of the regression regarding proposition 1a/b, for both environmental and social
performance, are presented in table 4 at the end of this paragraph. The results show that 24.3% of
the variance concerning environmental performance was explained by the control variables,
while for social performance this was even 31.6%. Model 1 was significant for both
environmental (F = 8.443, Sig. = 0.000) as well as social performance (F = 12.165, Sig. = 0.000).
When independent variable POWB was added in model 2, an additional 2.3 % of variance was
explained in terms of environmental performance, which was a significant improvement (F =
8.753, Sig. = 0.002). The regression coefficient for POWB was negative and significant (B =
-0.008, Sig. = 0.002), meaning that there is a relationship between POWB and environmental
performance, but this relationship is negative instead of positive as expected. Thus, proposition
1a remains unsupported. However, the second model did not improve the amount of variance
which could be explained for social performance (F = 11.464, Sig. = 0.122), the regression
coefficient was negative and insignificant for POWB (B = -0.004, Sig. = 0.122). So, POWB does
- 37 -
Table 4. Results of hierarchical linear regression analysis for proposition 1a and b
The results for the regression regarding proposition 2a and 2b are shown below in tables 5 and 6.
The regression coefficient for XM concerning Log Environmental performance is b3 = -0.003
and insignificant, t (313) = -0.869, Sig. = 0.386 (table 5). Thus, the effect that the percentage of
women on the board of directors has on environmental performance does not depend on the size
of the board. Consequently, proposition 2a remains unsupported. The same is true for Log Social
performance, where the interaction effect accounts for 0.11% of the variance explained, with a
regression coefficient of -0.003 which is statistically insignificant (t (313) = -0.727, Sig. = 0.468).
So, the size of the board does not affect the relationship between the percentage of female board
members and environmental and social performance, meaning that both propositions, 2a and 2b,
lack statistical support.
Environmental performance Social performance
Variable Model 1 Model 2 Model 1 Model 2
B Sig. B Sig. B Sig. B Sig.
Control variables
Log Total Assets -.327 .000 -.332 .000 -.169 .000 -.171 .000
Industrial vs. Utilities .002 .981 .019 .813 -.119 .156 -.110 .189
Industrial vs. Transportation -.011 .923 -.013 .908 .076 .524 .075 .529
Industrial vs. Bank/Savings & Loans .334 .000 .388 .000 .174 .052 .202 .027
Industrial vs. Insurance .453 .000 .449 .000 .347 .000 .345 .000
Industrial vs. Other Financial .083 .611 .072 .657 .522 .002 .516 .002
Log Debt Ratio 1.199 .012 .939 .048 .938 .052 .805 .100
Sqrt Net Income 3.639E-05 .105 3.519E-05 .112 -4.019E-05 .079 -4.081E-05 .074
Europe vs. North America .262 .000 .256 .000 .440 .000 .437 .000
Europe vs. South America .266 .101 .143 .386 .168 .309 .105 .537
Europe vs. Asia .660 .000 .535 .000 .765 .000 .700 .000 Europe vs. Oceania .035 .802 .033 .812 .268 .062 .267 .063 Independent variable POWB -.008 .002 -.004 .122 𝑅2 .243 .265 .316 .321 𝑅2 change .243 .023 .316 .005 F 8.443 8.753 12.165 11.464 Sig. F change .000 .002 .000 .122
