1
T
HE
Q
UEEN BEE PHENOMENON IN THE CORPORATE
WORLD
:
A QUANTITATIVE ANALYSIS
T
HE QUEEN BEE PHENOMENON AMONG FEMALE BOARD
MEMBERS AND FEMALE
CEO’
S
ABSTRACT
S
TUDENT:
D
IETE DEW
AALS
TUDENT NUMBER:
10343830
S
TUDY:
M
SC
E
NTREPRENEURSHIPS
UPERVISOR:
D
R.
R.
P
EREZR
IBAS2
ABSTRACT
Women participation at senior organizational levels lag well behind the male participation. In this study I research the presence of the queen bee phenomenon among the board of directors and the CEO’s by testing the causal effect of the board’s gender diversity on the gender diversity of the lower levels of the company. Hereby, the gender of the CEO is included as mediating variable. The sample of this study consists of 1358 companies listed on the S&P 1500 between 1992 and 2016. The results show that in contrast to my expectations, the board’s gender diversity has a positive effect on the company’s gender diversity. Thereby, there was no mediating effect of the gender of the CEO. However, the company’s gender diversity did appear to decrease once a female is assigned as CEO indicating that there is a possibility that the queen bee phenomenon is applicable to female CEO’s. My findings suggest that organizations can address the scarce representation of women in their organization by focusing on the role of the CEO. Specifically, organizations should motivate the CEO to reinforce gender equality within their organization and hire an equal amount of women and men. My findings contribute to the ongoing policy debate about the value of regulatory interventions to increase female participation at senior organizational levels and underline the presence and importance of the queen bee phenomenon and the awareness that an increasing share of women in the top-management positions do not necessarily lead to more opportunities for junior women.
3
TABLE OF CONTENT
1 Introduction………...4
2 Theoretical background & hypotheses………...6
2.1 Gender inequality in the work place………...…6
2.2 Gender diversity among the top management positions……….…7
2.3 The Queen Bee phenomenon………..8
3 Methodology………...11 3.1 Data………...…11 3.2 Empirical strategy……….…15 4 Results……….………..…18 5 Conclusion……….………23 6 Discussion………...………...24 7 References……….………26
4
INTRODUCTION
Even though in the last years women's participation in the workforce has increased substantially, the actual numbers show that women are still underrepresented at higher organizational levels (European Commission, 2014). Many studies have aimed to explain the persistence of the gender hierarchy. The mere of these studies are however based upon the dominant belief that gender diversity is perpetuated by men, not women and that women help other women and reinforce gender equality within an organization. Based upon these findings, gender quotas have been put in place in order to increase gender inequality and improve the opportunities for junior women (Duguid, 2011; Mavin, 2008).
This study questions this belief and is shedding light on the origin of organizational gender diversity. By removing the dominant belief that gender diversity is perpetuated by men, not women and the assumption that every women behaves according to the female stereotype, another theory is gaining ground in the literature. This theory is referred to as the queen bee phenomenon. Queen Bees are labeled as women leaders who assimilate into male-dominated organizations (i.e., organizations in which most executive positions are held by men) by distancing themselves from junior women and legitimizing gender inequality in their organization (Derks, Van Laar, Ellemers, 2016). This paper is testing this phenomenon
by quantitatively investigating the presence of the queen bee phenomenon among higher organizational levels.
To do so, this research focuses upon the role of the board of directors and the leading CEO positions within the companies. Respectively, the research question is as follows:
Does the queen bee phenomenon exist among senior organizational levels and reinforce the gender inequality within companies?
In order to answer this research question, I test the association between the % of women on the board and the gender diversity in the lower levels of the organization. Hereby, the gender of the CEO will be included as mediating variable between the board’s gender diversity and the company’s gender diversity. For representativeness purposes, this study is looking at companies listed on the S&P super composite 1500 index which combines the S&P mid-cap 400, S&P 500, and the S&P small-cap 600, and together comprise of approximately 90% of the US Market (S&P, 2018). CEO, company, and diversity data is retrieved from the research platform Wharton Research Data Services (WRDS). Consequently, a sample consisting of 1358 companies listed on the S&P 1500 between 1992 and 2016 is ready for testing. In my empirical methods, I make use of mediation steps of Baron & Kenny (1986) and linear and logistic fixed-effect regressions between the variables of interest. Furthermore, this study includes the influence of time by including lags to determine whether effects differ over time. Also, the results are controlled for the overall diversity change, firm size, leverage and company value.
The results show that in contrast to my expectations, the board’s gender diversity has a positive effect on the company’s gender diversity and that there is no evidence to believe that the queen bee phenomenon is present among the board of directors collected in my sample. Thereby, there was no mediating effect of the gender of the CEO. However, the company’s
5
gender diversity did appear to decrease once a female is assigned as CEO, which shows that the queen bee phenomenon might be present at the CEO level. Therefore, my findings suggest that organizations can address the scarce representation of women in their organization by focusing on the role of the CEO. Specifically, organizations should motivate the CEO to reinforce gender equality within their organization and hire an equal amount of women and men.
My findings contribute to the ongoing policy debate about the value of regulatory interventions to increase female participation at senior organizational levels and underline the presence and importance of the queen bee phenomenon and the awareness that an increasing share of women in the top-management positions do not necessarily lead to more opportunities for junior women. Thereby, no studies that I know of have quantitatively aimed to research the presence of the queen bee phenomenon among the board and executive level.
The remainder of this research is organized as follows. First, the literature regarding organizational gender inequality, the role of the Board of Directors and executives and the queen bee phenomenon will be discussed. Accordingly, the hypotheses will be formed. Afterwards, I elaborate upon the methodology of this study whereas the data & the empirical methods are explained. Consequently, the findings are discussed whereas I refer to the current literature. At last, I conclude with the main implications and limitations of this study.
6
THEORETICAL BACKGROUND & HYPOTHESES
In this section, relevant literature is discussed. Hereby a division between the board level and the executive level is made. First, evidence on the persistence of gender hierarchy is discussed. Then, the literature on the gender diversity among the organizational top management positions and its implications for the company are considered. Finally, I introduce the queen bee phenomenon and formulate the hypotheses.
1. Gender inequality in the work place
Even though in the last years women's participation in the workforce has increased substantially, the actual numbers show that women are still underrepresented at higher organizational levels (European Commission, 2014). Globally, women only account for 12% of board seats among the world’s largest companies, with 64% having one women director, and 13% with at least three (MSCI, 2014). When comparing these numbers to the proportion women who attain tertiary education and who graduate in pertinent disciplines, an uneven distribution is found. In the United Kingdom (UK) for example, Women on Boards represent 26%, while 51% of business degrees are held by women. Along the same lines, the US counts 19% of Women on boards while more than 50% of STEM degrees are held by women (OECD, 2016).
Moving on to the global women's participation on the executive level, the numbers further decrease whereas women continue to be under-represented at 19%. According to Catalyst15 (2015) women represent less than 5% of S&P 500 companies’ CEO’s, and hold less than a quarter of senior management roles. Research by Credit Suisse on the FTSE 100 and S&P 500 companies, finds that in the USA men in CEO positions outweigh women by 20, and in the UK, men in executive director positions outnumber women 10 to one. Accordingly, these statistics show that in spite of the increasing numbers of women in the workforce at all levels, the presence of female CEO’s and female board members at the top remains scarce with higher departure levels (Mercer, 2015; RobecoSAM, 2015).
Gender quotas
These staggering numbers are considerable, taking into account the major historical transformations that have occurred; A system that advantages males over females in power, status, and authority has continued in one form or another despite profound structural changes such as industrialization and the movement of production out of the household, women's accelerated movement into the labor force after World War II, and, most recently, women's entry into male-dominated occupations (Hartmann 1976; Reskin and Roos, 1990). Consequently, many studies have aimed to explain this persistence of gender hierarchy in the corporate world. However, as Ridgeway (1997) is making a fair point in her research on gender diversity claiming that there is no single answer to this question.
Governments have reacted to this paucity of women at senior levels by launching interventions such as legislative quotas (Gould, Kulik, Sardeshmukh, 2018). In 2012, 8 countries introduced quotas for women on corporate board of directors: Belgium, France, Spain, Iceland, Italy, the Netherlands, Malaysia & Norwegian (Teigen, 2012). Naturally, these quotas have
7
been put in place to improve opportunities for junior women (Duguid, 2011; Mavin, 2008). The assumption behind the gender quotas is however based on the believe that the presence of female top manager in a particular firm would be self-reinforcing, leading women to cluster there (Dezso, Ross & Uribe, 2016). Whether this belief is true is arguable. Likewise, Terjesen & Sealy (2016) note that although gender quotas seem to work they remain controversial (Terjesen & Sealy, 2016). Hereby, an important question that arises is whether having more women on the board and executives’ level actually reinforces the gender diversity within companies among the lower levels.
2. Gender diversity among the top management positions
The effect of the gender diversity among top management positions (board-and-executive level) on the gender diversity among the lower levels of the company is researched by several authors. In general, these studies underline the traditional belief that the presence of women on a top management team could be expected to reduce the gender inequality in companies. Dezso, Ross & Uribe (2016) outline the behavioral framework behind this assumption and argue that an increasing share of women in top management positions is expected to reduce the doubts of any other woman who might be promoted to the team with, or shortly after, her (Kanter, 1977). Then, as women rise up the corporate hierarchy, we would then expect stronger networks to emerge for senior women managers, greater accommodation of women managers’ parental responsibilities (Bertrand, Goldin, and Katz, 2010; Matsa & Miller, 2011), less uncertainty about women’s suitability for leadership positions by male managers (Aigner and Cain, 1977; Bielby and Baron, 1986; Phelps,1972) or investors (Lee and James, 2007), fewer gendered behaviors in screening job applicants (Fernandez-Mateo and King, 2011), and more role models and mentors for women at lower managerial levels (Williams and O’Reilly, 1998).
In accordance with these theories, many authors aim to confirm a positive relation between the gender diversity at higher organizational levels and the gender diversity among the lower levels of the company. The mere of these studies hereby focusses on the board-level (Kurtulus & Tomaskovic-Devey, 2012; Matsa & Miller, 2011; Gould, Kulik & Sardeshmukh, 2018; Bear et al.,2010) and only some on the CEO (executive)-level (Bloom and Van Reenen, 2007; Blau & DeVaro, 2007; Giuliano, Levine, and Leonard, 2011). For instance, on the board level Gould, Kulik & Sardeshmukh, 2018) find a positive relationship between female board representation and female executive representation, referred to as the trickle-down effect. On the CEO level, Kunze & Miller (2017) look at the relationship between female CEO’s and promotion rates and find positive spillovers across ranks.
Although some of these studies report significant results that could indicate that women on top reinforce gender equality in their organizations, the studies investigating the effectiveness of gender quotas report mixed results (Dubbink 2005, Gopalan & Watson 2015, Szydlo 2015). These scientists insist that quotas are undemocratic, are discriminatory, and undermine merit (Dubbink 2005, Gopalan & Watson 2015).
8
3. The queen bee phenomenon
Although, the mechanisms behind this traditional belief may apply to the reality sometimes, the results of these researches are mixed. Within this light, scientists have questioned the positive association between the gender diversity among the top-management positions and the gender diversity among the lower levels of the company.
The main criticism is pointed towards the underlying assumption that women help other women and reinforce gender equality in the companies they work at. Moreover, due to this assumption, many studies focus on the belief that gender inequality is perpetuated by men and not women. For instance, Durbin & Fleetwood (2010) distinguish internal and external causes of gender inequality in employment whereas none of these causes involve the role women-on-women interaction. Durbin & Fleetwood (2010) argue that externally, negative gender stereotypes and gender division of labor (domestic labor & paid employment) are seen as drivers of gender inequality on lower and executive levels. Internally, the authors point out that the influence of gender of relatively senior managers (who tend to be men) and the recruitment and promotion decisions they make is ought to be an important driver of gender inequality within organizations. As such, Durbin & Fleetwood (2010) state that that male managers and CEO’s exclude female employees from their companies and vice versa. Consequently, recruitment and promotion decisions made by relatively male senior managers and CEO’s continuously favor men, and become both an expression and a cause of gendered inequality in male dominated organizations (Durbin & Fleetwood, 2010).
By taking a step back and removing these existing (dominant) beliefs that for instance gender inequality is only perpetuated by men, not women and that women in general support other women, room is created for other possible causes of the persistence of the organizational gender hierarchy.
The Queen bee phenomenon: the definition & explanation Staines, Tavris, & Jayaratne (1974) were among others one of the first to reflect upon this dominant belief and propose an opposite theory. As such, the Queen Bee Phenomenon has made its introduction into the literature and has set new grounds to develop valuable research upon. In 1974, The Queen Bee effect was referred to as a woman who pursues individual success in male-dominated work settings (organizations in which men hold most executive positions) by adjusting to the masculine culture and by distancing themselves from other women (Kanter, 1977; Staines, Tavris, & Jayaratne, 1974). Since then, studies have researched the presence of the queen bee effect in various settings (Mavin ,2008; Derks et al., 2011; Derks, 2011; Derks et al., 2015; Faniko, Ellemers & Derks, 2016; Faniko, Ellemers, Derks, Lorenzi-Cioldi, 2017). In 2016, Derks, Van Laar & Ellemers (2016), expands this definition and research the consequences for the companies the queen bees are a part of. They find that queen bees affect the diversity climate negatively as they signal to junior women that in order to be accepted they need to deemphasize their gender, possibly leading to even higher turnover rates among women in companies with queen bees in management than companies with no women leaders (Derks, Van Laar & Ellemers, 2016). Consequently, the queen bees attain a new description and are labeled as women leaders who assimilate into male-dominated organizations (i.e., organizations in which most executive positions are held by men) by distancing themselves from junior women and legitimizing gender inequality in their organization (Derks, Van Laar & Ellemers, 2016).
9 The Queen bee phenomenon: evidence
Since the introduction of the queen phenomenon in 1974, scientists have aimed to identify the behavioral mechanisms, presence, and implications of the queen bee phenomenon. However, these studies mainly involve qualitative research (Mavin, 2008; Derks et al., 2011; Derks, 2011; Derks et al., 2015; Faniko, Ellemers & Derks, 2016; Faniko, Ellemers, Derks, Lorenzi-Cioldi, 2017). In 2011, Derks et al. (2011) performs a qualitative research whereas she argues for a social identity explanation and examine organizational conditions that foster the Queen Bee phenomenon. In another study Derks (2011) researches the origin of the queen bee behavior among senior policewomen and argues that the queen bee behavior is a result of the gender bias and social identity threat that produce gender disparities in career outcomes. Drawing upon recent work on the queen bee phenomenon among women at work, Derks et al. (2015) explain self-group distancing as a coping response of low identified minority employees or ethnic minorities who experience social identity threat which suggests that the Queen Bee phenomenon exemplifies a more generic individual mobility response to group disadvantage experienced by minority groups at work. In 2016, Faniko, Ellemers & Derks (2016) research the differences between queen bees and alpha males whereas they find evidence that both female and male managers rated their own masculinity as higher than that of the same gender-junior colleagues. Finally, the last study that has been conducted on the queen bee phenomenon provides a more nuanced picture of what the QB-response is really about, explaining why women managers oppose quotas for junior women, while supporting quotas for women in the same rank (Faniko, Ellemers, Derks, Lorenzi-Cioldi, 2017).
The main take away from these studies is that the queen bee phenomenon/effect is well substantiated in the current literature and that we should not expect women among top-positions to only support other women. In the contrary, we need to be open to the idea that women are queen bees and that they pursue their individual success in male-dominated work settings by adjusting themselves to the masculine culture and distancing themselves from other women.
Conclusion: The Queen bee phenomenon in comparison to traditional beliefs (hypotheses)
The main difference between the traditional belief that the presence of women on a top management team could be expected to reduce the gender inequality in companies and the queen bee phenomenon that states otherwise, stems from the different angle taken by the scientists. For the studies that focus upon the traditional belief, the focus is shed on the female stereotype and does not include the influence of her corporate surrounding. Consequently, it is assumed that women behave as stereotype women and that they for instance take care of other females in the company and that consequently gender inequality will decrease.
The literature on the queen bee effect, however, does include the interaction of women in the workplace and their corporate surroundings. As such, the underlying mechanisms of the queen bee phenomenon have everything to do with how a woman deals with her work environment and how this influences her behavior. By adding this interaction, the queen bee
10
Apart from this, the studies that find evidence for positive implications of gender diversity among higher organizational levels is scarce and outnumbered by the literature on gender quotas that report mixed results.
Therefore, I expect to find the queen bee effect on the board-level as well as on the CEO-level. Accordingly, I believe that the women on the board as well as the female CEO’s will behave as queen bees and legitimize gender inequality in their organizations, eventually leading to less gender diversity. Hereby, I include the gender of the CEO as mediating variable between the effect of the board’s gender diversity on the company’s gender diversity and expect this variable to mediate this association. Consequently, I formulate 3 hypotheses according to the mediation steps of Baron & Kenny (1986)
Hypothesis 1
H0: Female board members do not distance themselves from junior women and legitimize gender inequality in their organizations.
H1: Female board members do distance themselves from junior women and legitimize gender inequality in their organizations.
Hypothesis 2
H0: Female board members do not distance themselves from junior women and prefer male CEO’s over female CEO’s. H1: Female board members do distance themselves from junior women and prefer male CEO’s over female CEO’.s Hypothesis 3
H0: Female CEO's do not distance themselves from junior women and legitimize gender inequality in their organizations and mediate the relationship between the board diversity and gender diversity within the company.
H1: Female CEO's do distance themselves from junior women and legitimize gender inequality in their organizations and mediate the relationship between the board diversity and gender diversity within the company.
11
METHODOLOGY
In order to test the hypotheses discussed in the previous section and answer the research question of this paper, quantitative research is conducted. First, the collection and the characteristics of the data are described. In the second section, the research methods that are employed are discussed.
DATA
The data is derived from the research platform Wharton Research Data Services (WRDS). This study is looking at companies listed on the S&P super composite 1500 index. This index combines the S&P mid-cap 400, S&P 500, and the S&P small-cap 600, and together comprise of approximately 90% of the US Market (S&P, 2018).
4. CEO turnover data
For these companies, CEO information per year (CEO's name, gender)is extracted from WRDS’s database Execucomp. Companies whereas a change of CEO occurs are selected. In order to specify the difference between male and female CEO’s and their impact on the company’s gender diversity, only the CEO switches from male to male or from male to female are selected. Furthermore, for representation purposes, a selection is made based on the tenure of the CEO’s: CEO’s with tenure less than 2 years were removed from the dataset.
5. Board diversity data
Also, data on the board’s gender diversity of the selected companies is collected. This information is extracted from the database MSCI ESG KLD STATS (KLD). This database is part of MSCI ESG Research unit that belongs to WRDS. KLD gives access to positive or negative ESG performance indicators for publicly traded companies (WRDS, 2017). Hereby, WRDS presents a binary summary of positive and negative diversity ratings per company per year. In each observation, WRDS assigned a rating in a particular issue (either positive or negative), this is indicated with a value of 1. If the company did not have a strength or concern in that issue, this is indicated with a 0 (WRDS, 2017). For the board diversity (strength), WRDS assigns a value of 1 if more than four seats on the board of directors are held by women (no double counting) or if more than one-third of the board seats are assigned to women if the board numbers less than 12 (WRDS, 2018).
6. Company data In order to test the relative relationship between the board’s gender diversity, CEO turnover (M-FM) and the company’s diversity, I control the outcomes for company characteristics that may influence the causal effect of the independent variables on the dependent variables. The control variables involved are: firm size (amount of assets), leverage (total debt/total equity) and the value of the company. The value of the company is determined by Tobin’s q ratio (Tobin, 1977). Tobin’s Q ratio is calculated as the market value of a company divided by the replacement value of the firm's assets and is found to be a significant determinant of investment (Hayashi and Inoue, 1991).
12
7. Diversity data
Diversity information is also derived from WRDS’s KLD database. In this study, the gender diversity in companies is measured by looking at the following diversity dimensions of KLD stats: women promotion (strength), work-life benefits (strength), women contracting (strength), non- representativeness (concern), and controversies (concern). Likewise the data
on the board’s diversity data, WRDS has assigned a value of 0 or 1 to these dimensions per company per year.
Women promotion (strength)
WRDS categorizes companies in 2 segments: companies that stimulates women promotion (1) and companies that do not (0). Accordingly a company that is given a value of 1 is referred to as an organization that has made notable progress in the promotion of women and minorities, particularly to line positions with profit-and-loss responsibilities in the corporation
(WRDS, 2018).
Work/life benefits (strength)
Hereby, WRDS categorizes companies in the following 2 segments: companies that have outstanding employee benefits and companies that do not. Accordingly a company that is given a value of 1 is referred to as an organization that has outstanding employee benefits or other programs addressing work/life concerns, e.g., childcare, elder care, or flextime (WRDS, 2018).
Women & Minority Contracting (strength)
For this dimension, a company is given a value of 1 if the company does at least 5% of its subcontracting, or otherwise has a demonstrably strong record on purchasing or contracting, with women- and/or minority-owned businesses (WRDS, 2018). Logically a value of 0 Is given to a company if it does not comply to the conditions written above.
Non- Representativeness (concern)
Companies are given a value of 1 for the dimension Non-Representativeness if the company has no women on its board of directors or among its senior line managers. Logically, a company is given a value of 0 if the company does have women on
its board of directors or among its senior line managers.
Controversies (concern)
For the last dimension, a value of 1 is given if the company has either paid substantial fines or civil penalties as a result of affirmative action controversies or has otherwise been involved in major controversies related to affirmative action issues. Accordingly, a company that is given a value of 0 is referred to as an organization that has not paid substantial fines or civil penalties as a result of affirmative action controversies or has otherwise been involved in major controversies related to affirmative action issues (WRDS, 2018).
For the calculation of the company’s yearly diversity score, I summed up the values of the dimensions indicating strength (women promotion, work/life benefits, and women & minority contracting) and subtracted the value of the dimensions indicating concerns (non-representation & controversies). Furthermore, companies whereas one of the dimension values is missing, are removed from the sample.
13
Another variable that I computed from KLD’s diversity data is the overall diversity change per year. By including this variable in the model, I aim to control the results for the overall (macro) change of gender diversity among companies in the USA. I calculated this change by taking the average of the diversity scores of all the companies in the original dataset per year. Then I calculated the change in diversity score per year and included these in the study's model. By doing so, I aim to control the results for possible influences from the overall trend of organizational gender diversity in the USA.
8. Descriptive statistics
After combining the CEO, board, company and diversity data, a total sample of 1358 companies with in total 17744 observations is ready for testing. The data is reported from 1992 to 2016. The descriptive statistics of the variables of this study are presented in table 1 and table 2. In table 1, the descriptive statistics are presented based upon the gender diversity among the board of directors. Table 2 reports the summary statistics of the same variables based upon the gender of the CEO. As shown in both tables, gender diverse boards (1760 observations) as well as female CEO’s (292 observations) have a small share in the sample in comparison to the male-dominated boards (15984 observations) and male CEO’s (17452 observations). This confirms the scarce share women have in the CEO positions in the USA. Furthermore, both tables show high standard deviations indicating a large amount of variation present in the different samples.
Looking at the descriptive statistics of the male dominated boards and the gender diverse boards, the data shows that under the supervision of a gender diverse board, the mean of the amount of CEO turnovers (0,68) and the CEO turnovers from male to female (0,064) are higher in comparison to the output of these variables under the regulation of a male dominated board. Furthermore, in this study’s sample, the company’s gender diversity appears to be higher if the board is gender diverse (0,77) than when the board is male dominated (0,053). However, it must be noted that this value is low in both samples as the company’s gender diversity can attain a maximum value of 3. Looking at the control variables, the output shows that the companies whereas the board is gender diverse, the size of the companies is larger (EUR 76 754) than the companies whereas the board is male dominated (EUR 18 596). Apart from this, no remarkable results are shown. Moving on to the descriptive statistics based on the gender of the CEO, the results show that under the management of a female CEO, the gender diversity is higher (0.43) in comparison to male CEO’s (0,12). However, the overall diversity change appears to be higher under the management of female CEO’s (0,023) in comparison to male CEO’s (-0,02) as well. This finding in combination with the relatively small sample size of female CEO’s, may increase the company’s gender diversity substantially. Apart from this, no remarkable results are shown.
In order to gain more insight in the composition of the company’s gender diversity, the correlations between the total diversity score and the underlying dimensions are shown in table 4. The correlation results show that for all underlying diversity dimensions except controversies (concern), the correlation with the total diversity score per year per company are between 0.49 and 0.67. Therefore it can be assumed that these dimensions play an equal part in the constitution of the total diversity score. For the dimension controversies (concern), a smaller influence on the total diversity score is observed (0.11).
14 Table 1 Descriptive statistics (based on board diversity)
Variable Observations Mean Std dv Observations Mean Std dv Observations Mean Std dv
CEO turnover 17744 0.520 0.500 15984 0.502 0.500 1760 0.685 0.465
CEO turnover (M-FM) 17744 0.016 0.127 15984 0.011 0.105 1760 0.064 0.245
Company's Gender diversity 17744 0.125 0.793 15984 0.053 0.736 1760 0.778 0.976
Overall gender diversity change 17744 0.002 0.104 15984 0.000 -0.103 1760 -0.018 -0.104
Size (€) 15304 24445 122961 13765 18597 99070 1539 76755 244044
Leverage 15274 0.321 0.273 13739 0.324 0.276 1535 0.295 0.238
Company value 15302 1.241 1.219 13763 1.250 1.220 1539 1.161 1.205
Total sample Male dominated board Gender diverse board
Reports the descriptive statistics for the total sample, the sample with a male dominated board of directors, and a gender diverse board of directors. Gender diverse boards comprise of a board whereas more than four seats on the board of directors are held by women (no double counting) or if more than one-third of the board seats are assigned to women if the board numbers less than 12 (Wharton, 2018). For each of these three samples, the mean and standard deviation is reported.
Table 2 Descriptive statistics (based on gender CEO)
Variable Observations Mean Std dv Observations Mean Std dv Observations Mean Std dv
BoD's gender diversity 17744 0.099 0.299 17452 0.094 0.292 292 0.387 0.488
Company's Gender diversity 17744 0.125 0.793 17452 0.120 0.794 292 0.428 0.692 Overall gender diversity change 17744 -0.002 0.104 17452 -0.002 0.103 292 0.023 0.127
Size 15304 24445 122961 15015 24552 124043 289 18915 34858
Leverage 15274 0.321 0.273 14988 0.323 0.273 286 0.267 0.250
Company value 15302 1.241 1.219 15013 1.241 1.222 289 1.198 1.074
Total sample Male CEO's Female CEO's
Reports the descriptive statistics for the total sample, the sample with male CEO's, and female CEO's. The sample with female CEO's comprise of only female CEO's that follow up a male CEO. For each of these three samples, the mean and standard deviation is reported.
Table 4 Correlation table (Gender diversity dimensions)
Correlation table
Total diversity score 1
Non-representation (concern) -0.5988 1
Controversies (concern) -0.1188 -0.0737 1
Women & Minority Contracting (strength) 0.4912 -0.1055 0.1618 1
Work/life benefits (strength) 0.5628 -0.1048 0.1349 0.2738 1
Women promotion (strength) 0.6753 -0.2065 0.0802 0.1081 0.2233 1 Reports the correlations of the company's diversity score and its underlying dimensions.
Total diversity score Non-representation (concern) Controversies (concern) Women & Minority Contracting Work/life benefits (strength) Women promotion (strength)
15
EMPIRICAL STRATEGY
In this section, I explain the empirical strategy I adopt to answer the research question of this study. As stated in the literature review, this research has formulated 3 hypotheses in order to ultimately test whether female CEO's distance themselves from junior women and legitimize gender inequality in their organizations and mediate the relationship between the board diversity and gender diversity within the company. Hereby, the gender of the CEO plays a mediating role in the relationship between board’s gender diversity and the company’s gender diversity. In order to test this, the mediation approach of Baron & Kenny (1986) is applied. Accordingly, these 3 mediation steps are made efficacious in the remainder of this section.
For all the models used in this study, I make use of a linear fixed effect regressions or a logistic fixed effect regression whereas I control for the unobserved heterogeneity within companies if this heterogeneity is constant over time. Thereby, the influence of time is included in all models by making use of lags of the independent variables. By looking at the causal effects of the independent variables of the previous years on the dependent variables in the current year I can determine whether these causal effects differ over time. Furthermore, I make use of the control variables overall gender diversity change, size, leverage and company value (Tobin's Q).
1. The board’s gender diversity & the company’s gender diversity
First, this study looks at the relationship between the board’s gender diversity and the company’s gender diversity in order to confirm that board diversity is a significant predictor of the company’s diversity (Baron and Kenny, 1986). This relationship is tested with the following model:
(1) 𝐶𝑜𝑚𝑝𝑎𝑛𝑦′𝑠 𝑔𝑒𝑛𝑑𝑒𝑟 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 = 𝛽0 + ∑∞ (𝛽 𝑏𝑜𝑎𝑟𝑑′𝑠 𝑔𝑒𝑛𝑑𝑒𝑟 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦) + 𝛽 (𝑜𝑣𝑒𝑟𝑎𝑙𝑙 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 △)
𝑛=5 + β
(size) + β (leverage) + β (company value)
Hereby, the dependent variable is company’s gender diversity that can take a maximum value of 3 or a minimum value of -2. The independent variables are binary variables whereas board (gender) diversity can attain a value of 0 (male dominated board) or 1 (gender diverse board). The control variables are all continuous variables.
2. The board’s gender diversity & CEO turnover (from male to female)
Second, the relationship between the mediators and the independent variable is assessed by regressing the mediating variables on the board’s gender diversity. By doing this, I can determine whether the board’s gender diversity is a significant predictor of the CEO turnover and CEO turnover from male to female within that company. Because the dependent variable is binary, logistic fixed effect regressions are applied. First, I look at the relationship between the board’s gender diversity and the occurrence of a CEO switch (1) and secondly, at the relationship between the board’s gender diversity and the occurrences of a CEO turnover from male to female (3). Accordingly, the following two models apply:
16
(2) 𝐶𝐸𝑂 𝑡𝑢𝑟𝑛𝑜𝑣𝑒𝑟 = 𝛽0 + ∑∞ (𝛽 𝑏𝑜𝑎𝑟𝑑′𝑠 𝑔𝑒𝑛𝑑𝑒𝑟 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦) + 𝛽 (𝑜𝑣𝑒𝑟𝑎𝑙𝑙 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 △)
𝑛=5 + β (size) + β
(leverage) + β (company value)
(3) 𝐶𝐸𝑂 𝑡𝑢𝑟𝑛𝑜𝑣𝑒𝑟 (M − FM) = 𝛽0 + ∑∞ (𝛽 𝑏𝑜𝑎𝑟𝑑′𝑠 𝑔𝑒𝑛𝑑𝑒𝑟 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦) + 𝛽 (𝑜𝑣𝑒𝑟𝑎𝑙𝑙 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 △)
𝑛=5 + β
(size) + β (leverage) + β (company value)
In equation (2), the dependent variable is the CEO turnover that can take a value of 1 if a switch of CEO occurs or a value of 0 if no CEO switch occurs. In equation (3), the dependent variable is the change of gender of the CEO from male to female if a switch occurs. This variable can attain a value of 1 (CEO switch from male to female) or 0 (CEO switch from male to male). The independent variables in both model (2) & (3), board’s gender diversity is binary and attains a value of 1 if the board is gender diverse or 0 if the board is male-dominated. The control variables are all continuous variables
3. The board’s gender diversity, CEO turnover (from male to female) & the company’s gender diversity
Finally, I determine whether female CEO’s behave differently with regards to junior women and gender inequality within their company and mediate the relationship between the board’s gender diversity and the overall gender diversity within the company. This is done by regressing the company’s gender diversity on the board’s gender diversity as well as on the CEO turnover and the CEO turnover from male to female.
As such, I can conclude whether CEO turnover and CEO turnover from male to female mediates the relationship between the board’s gender diversity and the company’s gender diversity. If this is the case, the causal effect of the board’s gender diversity on the company’s gender diversity will decrease once the mediating variables are included in the model as the mediating variables should partly explain the relationship between the dependent and independent variable (Baron & Kenny, 1986). The model for this test is as follows:
(4) 𝐶𝑜𝑚𝑝𝑎𝑛𝑦′𝑠 𝑔𝑒𝑛𝑑𝑒𝑟 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 = 𝛽0 + ∑ (𝛽 𝑏𝑜𝑎𝑟𝑑′𝑠 𝑔𝑒𝑛𝑑𝑒𝑟 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦) + ∑∞ (𝛽 𝐶𝐸𝑂 𝑡𝑢𝑟𝑛𝑜𝑣𝑒𝑟) + 𝑛=1
∞ 𝑛=5
∑∞ (𝛽 𝐶𝐸𝑂 𝑡𝑢𝑟𝑛𝑜𝑣𝑒𝑟 𝑀 − 𝐹𝑀) + 𝛽 (𝑜𝑣𝑒𝑟𝑎𝑙𝑙 𝑑𝑖𝑣𝑒𝑟𝑠𝑖𝑡𝑦 △)
𝑛=1 + β (size) + β (leverage) + β (company
value)
Hereby, the dependent variable is the company’s gender diversity that can take a maximum value of 3 or a minimum value of -2. The independent variables are binary variables whereas CEO turnover as well as CEO turnover from male to female can attain a value of 0 or 1. CEO Turnover gets assigned a value of 0 if no CEO turnover occurs and a value of 1 if a CEO turnover does occur. For the CEO turnover from male to female, a value of 1 is assigned if a CEO changes from male to female. A change of CEO from male to male is assigned value of 0. The control variables are all continuous variables.
17
Concluded, with the mediation steps of Baron & Kenny (1986), I test whether board’s gender diversity has a negative effect on the company’s gender diversity and whether CEO turnover and the CEO turnover from male to female has a mediating effect upon the relationship. In the following chapter, the results of the tests described above are discussed.
18
RESULTS
Throughout this chapter, the results will be discussed and interpreted in the existing literature. This section is divided according to the 3 mediation steps of Kenny & Baron (1986)
1. The board’s gender diversity & the company’s gender diversity
As discussed in the empirical methods, the causal effect of the board's gender diversity on the company’s gender diversity is tested first. Hereby, the control variables overall gender diversity change, size, leverage, and company value are included. Furthermore, I gradually include the lags of board's gender diversity per company per year. The output as well as this process is shown in table 1.
Before looking at the causal effects of the board’s gender diversity on the company’s gender diversity, I determine which model is the best fit for the determination of this relationship. In order to so, I gradually include the lags of the independent variable (board’s gender diversity) and reported this process in table 1. As shown in the characteristics overview of the models, the amount of observations and number of companies decreases once more lags are included. Accordingly, after including a second lag, more insignificant results are presented. Apart from the insignificance of these results, the coefficients (β) also show diverging results that are hard to substantiate after the second lag is included. For selecting the right model, both the amount of observations as well as the significance of the output is taken into account. Accordingly, it becomes apparent that model 2 is the best fit for this research. This is mainly due to the fact that under this model, the output shows substantiate and significant results. Thereby no data is lost as the observations and number of companies is the same as for model 1 (1744 observations). For the remainder of the models that are in need of testing, this model will be applied as well.
Moving on to the results for the first hypothesis of this study, a significant and positive relation between the board's gender diversity and the company’s gender diversity is shown in this study’s sample. More specifically, for the same year, a gender diverse board appears to increase the company’s gender diversity with 20%. When looking at the effect of the gender diversity among the board from the previous year t-1 on the company’s gender diversity in year t, the causal effect decreases but is still significant and positive (0,093). This could indicate that in this sample, the influential power of a gender diverse board has a stronger effect on the company’s gender diversity in that specific year than on the year after that. After including the control variables, these results remain significant (p-value<0,05).
This finding is not in line with my expectations. Likewise, H0 (Female board members do not distance themselves from junior women and legitimize gender inequality in their organizations) is therefore not rejected. The positive association shows that on the board level, there is no evidence to believe that female board member distance themselves from junior women and prefer male CEO's over female CEO's. In contrary, the board's gender diversity is positively influencing the company's gender diversity.
19
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6
Number of observations 17,744 17,744 15,041 13,723 12,426 11,150
Number of companies 1,358 1,358 1,316 1,295 1,274 1,231
average observations per company 13.1 13.1 11.4 10.6 9.8 9.1
Total diversity score β SE t P>|t|
BoD's gender diversity 0.247 0.022 11.110 0.000 0.203 0.290 Overall gender diversity change 0.356 0.047 7.610 0.000 0.264 0.447 Size 0.000 0.000 -5.450 0.000 0.000 0.000 Leverage 0.063 0.039 1.620 0.105 -0.013 0.139 Company value 0.001 0.007 0.160 0.873 -0.013 0.016 Constant 0.069 0.019 3.670 0.000 0.032 0.106 BoD's gender diversity 0.207 0.026 7.890 0.000 0.155 0.258 BoD's gender diversity n-1 0.093 0.027 3.460 0.001 0.041 0.146
Overall gender diversity change 0.380 0.048 7.950 0.000 0.286 0.473 Size 0.000 0.000 -5.460 0.000 0.000 0.000 Leverage 0.062 0.041 1.490 0.135 -0.019 0.143 Company value -0.003 0.008 -0.330 0.738 -0.019 0.013 Constant 0.077 0.020 3.780 0.000 0.037 0.116 BoD's gender diversity 0.215 0.027 8.040 0.000 0.162 0.267 BoD's gender diversity n-1 0.028 0.030 0.930 0.350 -0.031 0.088
BoD's gender diversity n-2 0.154 0.028 5.480 0.000 0.099 0.208
Overall gender diversity change 0.376 0.049 7.750 0.000 0.281 0.471 Size 0.000 0.000 -5.540 0.000 0.000 0.000 Leverage 0.085 0.044 1.930 0.054 -0.001 0.171 Company value 0.005 0.009 0.620 0.537 -0.012 0.022 Constant 0.065 0.022 3.010 0.003 0.023 0.107 BoD's gender diversity 0.219 0.027 7.990 0.000 0.165 0.273 BoD's gender diversity n-1 0.031 0.031 1.020 0.309 -0.029 0.092
BoD's gender diversity n-2 0.159 0.032 4.900 0.000 0.095 0.222
BoD's gender diversity n-3 -0.004 0.030 -0.120 0.903 -0.062 0.055
Overall gender diversity change 0.361 0.050 7.290 0.000 0.264 0.458 Size 0.000 0.000 -5.590 0.000 0.000 0.000 Leverage 0.079 0.047 1.690 0.092 -0.013 0.171 Company value 0.007 0.009 0.800 0.423 -0.011 0.026 Constant 0.079 0.023 3.410 0.001 0.034 0.124 BoD's gender diversity 0.219 0.028 7.750 0.000 0.164 0.275 BoD's gender diversity n-1 0.033 0.032 1.050 0.294 -0.029 0.095 BoD's gender diversity n-2 0.157 0.033 4.730 0.000 0.092 0.223 BoD's gender diversity n-3 0.006 0.036 0.160 0.872 -0.064 0.076 BoD's gender diversity n-4 -0.001 0.032 -0.040 0.971 -0.064 0.062
Overall gender diversity change 0.336 0.051 6.640 0.000 0.237 0.435 Size 0.000 0.000 -5.480 0.000 0.000 0.000 Leverage 0.065 0.050 1.280 0.201 -0.034 0.163 Company value 0.009 0.010 0.910 0.365 -0.011 0.029 Constant 0.099 0.025 3.950 0.000 0.050 0.148 Gender diversity board of directors 0.215 0.029 7.310 0.000 0.157 0.272 Gender diversity board of directors n-1 0.030 0.033 0.920 0.360 -0.034 0.094
Gender diversity board of directors n-2 0.165 0.034 4.810 0.000 0.098 0.232
Gender diversity board of directors n-3 -0.002 0.037 -0.040 0.965 -0.074 0.071
Gender diversity board of directors n-4 0.041 0.038 1.070 0.283 -0.034 0.116
Gender diversity board of directors n-5 -0.073 0.034 -2.160 0.031 -0.139 -0.007
Overall gender diversity change 0.333 0.052 6.410 0.000 0.231 0.435 Size 0.000 0.000 -5.340 0.000 0.000 0.000 Leverage 0.055 0.055 1.000 0.319 -0.053 0.162 Company value 0.008 0.011 0.700 0.481 -0.014 0.030 Constant 0.122 0.028 4.410 0.000 0.067 0.176 Model 4 Model 5 Model 6
Reports the results of a fixed-effect linear regression whereas the overal diversity score per company per year is regressed upon the gender diversity of the board and additional control variables. The lags are included gradually and reported in model 1-6.
Table 5. The causal effect of the board's gender diversity on the company's diversity
95% Conf. Interval Model 1
Model 2
20
2. The board’s gender diversity & CEO turnover (from male to female)
The results for the second hypothesis, which investigates the causal effect of the board’s gender diversity on the CEO turnover and the CEO turnover from male to female, are reported in table 6 (The causal effect of the board's gender diversity on CEO turnover) and table 7 (The causal effect of the board's gender diversity on the CEO turnover (M-FM)).
In order to determine whether CEO turnover (from male to female) mediates (the positive and significant) causal effect of the board’s gender diversity on the company’s gender diversity, the board’s gender diversity needs to be a significant predictor of the mediator(s). Logically, if the mediators are not associated with the board’s gender diversity, then they couldn’t possibly mediate the relationship between the board’s gender diversity and the company’s gender diversity (Baron & Kenny, 1981).
CEO turnover
Regarding the relationship between the board’s gender diversity and CEO turnover (table 6), significant results are found for the direct effect of the board’s gender diversity on the CEO turnover within that same year. After controlling the results for size, leverage, and company value, the odds ratio stays positive (1.74) and significant (p = 0 <0,05). Within this sample, this finding shows that, if a company’s board of directors becomes gender diverse, more CEO switches will occur that same year.
CEO turnover (M-FM)
Moving on to the causal effect of the board’s gender diversity on the CEO turnover from male to female, no significant results are reported (table 7) after the control variables are included in the model. This finding shows that in this study’s sample, the gender diversity among the board has no causal effect on the occurrence of CEO gender switches from male to female. Therefore, I am not able to reject H0 of hypothesis 2 (female board members do not distance themselves from junior women and prefer male CEO’s over female CEO’s). Additionally, this finding shows that CEO turnover from male to female cannot mediate the relationship between the board’s gender diversity and the company’s gender diversity.
Number of observations 13287
CEO switch Odds Ratio SE t P>|t|
Gender diversity BoD 1.686 0.194 4.530 0.000 1.345 2.114
Gender diversity BoD n-1 2.429 0.293 7.340 0.000 1.917 3.078
CEO switch Odds Ratio SE t P>|t|
Gender diversity BoD 1.740 0.316 3.050 0.002 1.219 2.483
Gender diversity BoD n-1 1.082 0.203 0.420 0.675 0.749 1.561
Overall diversity change 28.065 7.819 11.970 0.000 16.256 48.454
Size 1.000 0.000 12.820 0.000 1.000 1.000
Leverage 1.402 0.293 1.620 0.106 0.931 2.111
Company value 0.786 0.035 -5.410 0.000 0.720 0.857
95% Conf. Interval Table 6. The causal effect of the board's gender diversity on CEO turnover
95% Conf. Interval
Reports the results of a fixed-effect logistic regression whereas the CEO turnover per company per year is regressed upon the gender diversity of the BoD and additional control variables.
21
3. The board’s gender diversity, CEO turnover (from male to female) & the company’s gender diversity
Finally, the main research question of this study is answered by interpreting the results of the tests of the final hypothesis. As such it will become clear whether Female CEO's distance themselves from junior women and legitimize gender inequality in their organizations and mediate the relationship between the board diversity and gender diversity within the company? As stated earlier the relationship between the board’s gender diversity and the company’s gender diversity is positive (0.20) and significant (SE<0,05). In order to report the impact of the mediating variables on the relationship between the board’s gender diversity and the company’s gender diversity, the mediating variables are gradually included. The results are reported in table 8.
CEO turnover
The first mediating variable that is included in the model is CEO turnover. As stated in earlier, CEO turnover is associated with the board’s gender diversity leaving the possibility of a possible mediation open. Looking at table 8, the results show that once the variable CEO turnover is included in the model, the causal effect of the board’s gender diversity on the company’s gender diversity decreases from 0.185 to 0.178. This is also reflected in the coefficients of CEO turnover and CEO turnover n-1 that have a value of positive and significant value of 0,05 and 0,049. Accordingly, we can confirm that within this study’s dataset, CEO turnover mediates the positive relationship between the board’s gender diversity and the company’s gender diversity. This finding shows that the CEO actually might have an influence on the company’s gender diversity.
CEO turnover (M-FM)
Then, in order to answer the ultimate research question, I included the variable CEO turnover from male to female in the model. Earlier, no association between the board’s gender diversity and the CEO turnovers from male to female appeared to be present. When reviewing table 8, we find results that are in line with this finding. By including the gender change of the CEO turnover, the causal effect between the board’s gender diversity and the company’s gender diversity increases again from 0.178 to 0.185. Accordingly, I can confirm that CEO turnover changes from male to female do not mediate the
Number of observations 13287
CEO switch Odds Ratio SE t P>|t|
Gender diversity BoD 1.969 0.519 2.570 0.010 1.174 3.301
Gender diversity BoD n-1 2.070 0.554 2.720 0.007 1.225 3.497
CEO switch Odds Ratio SE t P>|t|
Gender diversity BoD 1.686 0.500 1.760 0.078 0.943 3.014
Gender diversity BoD n-1 0.922 0.287 -0.260 0.794 0.501 1.696
Overall diversity change 353.082 357.306 5.800 0.000 48.584 2566.018
Size 1.000 0.000 3.310 0.001 1.000 1.000
Leverage 6.009 6.287 1.710 0.087 0.773 46.702
Company value 0.696 0.129 -1.950 0.051 0.483 1.002
Reports the results of a fixed-effect logistic regression whereas the CEO turnover from male to female per company per year is regressed upon the gender diversity of the BoD and additional control variables.
95% Conf. Interval Table 7. The causal effect of the board's gender diversity on the CEO turnover (M-FM)).
22
relationship between the board’s gender and the company’s gender diversity. In spite of this, the gender of the CEO does seem to have a significant (p>0,05) and negative (-0.30) effect on the company's gender diversity after adding the control variables which is in line with the literature on the queen bee phenomenon. Therefore, I can partly reject H0 of the third hypothesis (Female CEO's do not distance themselves from junior women and legitimize gender inequality in their organizations and mediate the relationship between the board diversity and gender diversity within the company). The negative causal effect of the CEO turnover from male to female on the company’s gender diversity shows that the queen bee phenomenon might be present among female CEO’s. Accordingly, these females might distance themselves from junior women and legitimize gender inequality in their organization. Therefore, the first part of the hypothesis 3 (Female CEO's do not distance themselves from junior women and legitimize gender inequality) is rejected. Moving on to the second part of hypothesis 3 (Female CEO's mediate the relationship between the board diversity and gender diversity within the company), no evidence is found to reject this.
Number of observations 13287
Number of companies 1333
Average observations per company 10
Diversity score β SE t P>|t|
BoD's gender diversity 0.186 0.024 7.800 0.000 0.139 0.232
BoD's gender diversity n-1 0.101 0.024 4.160 0.000 0.054 0.149
Constant 0.105 0.005 20.430 0.000 0.095 0.115
Diversity score β SE t P>|t|
BoD's gender diversity 0.179 0.024 7.520 0.000 0.132 0.226
BoD's gender diversity n-1 0.090 0.024 3.680 0.000 0.042 0.138
CEO turnover 0.050 0.021 2.450 0.014 0.010 0.091
CEO turnover n-1 0.049 0.020 2.520 0.012 0.011 0.088
Constant 0.054 0.009 6.140 0.000 0.037 0.071
Diversity score β SE t P>|t|
BoD's gender diversity 0.186 0.024 7.840 0.000 0.139 0.232
BoD's gender diversity n-1 0.096 0.024 3.950 0.000 0.048 0.144
CEO turnover 0.059 0.021 2.840 0.004 0.018 0.099 CEO turnover n-1 0.059 0.020 3.030 0.002 0.021 0.098 CEO turnover (M-FM) -0.303 0.078 -3.870 0.000 -0.456 -0.150 CEO turnover (M-FM) n-1 -0.218 0.082 -2.650 0.008 -0.379 -0.056 Constant 0.051 0.009 5.860 0.000 0.034 0.068 Diversity score β SE t P>|t|
BoD's gender diversity 0.212 0.026 8.120 0.000 0.161 0.263
BoD's gender diversity n-1 0.094 0.027 3.480 0.000 0.041 0.146
CEO turnover 0.066 0.024 2.760 0.006 0.019 0.113
CEO turnover n-1 0.028 0.023 1.220 0.224 -0.017 0.072
CEO turnover (M-FM) -0.375 0.081 -4.610 0.000 -0.534 -0.215
CEO turnover (M-FM) n-1 -0.194 0.084 -2.300 0.021 -0.358 -0.029
Overall gender diversity change 0.374 0.048 7.790 0.000 0.280 0.468
Size 0.000 0.000 -5.770 0.000 0.000 0.000
Leverage 0.062 0.041 1.500 0.133 -0.019 0.143
Company value -0.001 0.008 -0.070 0.942 -0.017 0.015
Constant 0.031 0.023 1.370 0.171 -0.013 0.075
Table 8. The mediation effect of CEO turnover (M-FM)
95% Conf. Interval
95% Conf. Interval
95% Conf. Interval
95% Conf. Interval
23
C
ONCLUSIONBy means of this study, I aim to investigate the presence of the queen bee phenomenon among higher organizational levels. To do so, I looked at the relationship between the board’s gender diversity, the gender of the CEO and the gender diversity among the lower levels of the organization. Hereby 3 hypotheses are formulated based upon the mediating steps of Baron & Kenny (1974). Below the main conclusions of these 3 steps are given:
The board’s gender diversity & the company’s gender diversity After selecting the right model, which eventually turns out to be the model whereas one lag is included in the model, the results show that the relationship between the board’s gender diversity and the company’s gender diversity is significant and positive for both year t (0,20) as well as year t-1 (0,093). The latter finding shows that the influential power of a gender diverse board has a stronger effect on the company’s gender diversity in that specific year than on the year after that. With these findings, that go against my expectations, H0 of hypothesis 1 (Female board members do not distance themselves from junior women and legitimize gender inequality in their organizations) is therefore accepted.
The board’s gender diversity & CEO turnover (from male to female) Then, in order to determine whether CEO turnover (from male to female) mediates (the positive and significant) causal effect
of the board’s gender diversity on the company’s gender diversity, the board’s gender diversity needs to be a significant predictor of the mediator(s) (Baron & Kenny, 1981). However, the results show that the board’s gender diversity is only a significant predictor of the mediator turnover (for year t) and not on the turnover from male to female. Consequently, H0 of hypothesis 2 cannot be rejected (female board members do not distance themselves from junior women and prefer male CEO’s over female CEO’s).
The board’s gender diversity, CEO turnover (from male to female) & the company’s gender diversity
Ultimately, the impact of the mediating variables turnover (from male to female) is assessed. This is done by regressing company’s diversity score on the board’s gender diversity and the turnover from (male to female). The results show that only turnover appears to mediate the effect between the board’s gender diversity and the company’s gender diversity. When the turnover from male to female is included in the model, the significance of the effect of the board’s gender diversity decreases. Accordingly, I can conclude that CEO turnover from male to female does not mediate the relationship between the board’s gender diversity and the company’s gender diversity. However, I do find a significant and negative relationship between the CEO turnover from male to female and the company’s gender diversity for both year t and year t-1, which is in line with the literature on the queen bee effect. Consequently, I can partly reject H0 of hypothesis 3 (Female CEO's do not distance themselves from junior women and legitimize gender inequality).
Concluded, the results show that only among the female CEO’s the queen bee effect could be applicable. In contrary, the board’s gender diversity appears to have a positive effect on the company’s gender diversity. Accordingly, these findings show that at higher organizational levels the queen bee phenomenon might be present but only at the CEO-level.
24
DISCUSSION
Within this section, the results are discussed and interpreted in the existing literature. Then, I discuss the main limitations of this study and suggest new insights and research implications.
Results
The results show that the queen bee phenomenon might be present at higher organizational levels but only on the CEO level. On the board level, opposite results are found as the board’s gender diversity turns out to have a positive effect on the company’s gender diversity. The findings on the CEO level are in line with the studies on the queen bee phenomenon and underline that gender inequality is not only perpetuated by men and that we should accept the idea that women can also reinforce gender inequality within their organizations. In the contrary, I suggest that organizations can address the scarce representation of women in their organization by focusing on the role of the CEO. More specifically, organizations should motivate the CEO to reinforce gender equality within their organization and hire an equal amount of women and men.
However, the findings on the board level are surprising as my expectation was to find the queen bee effect on both levels. This expectation was based on the principles of the queen bee phenomenon and the impact of a female’s interaction with a male dominated work-environment as well as on the existing literature on the effectiveness of gender quotas. As such I hypothesized that in order for a female to attain a high position, she would become more masculine and eventually distance herself from female co-workers and legitimize gender inequality in her organization. A possible reason for these unexpected results might be the fact that decisions by the board are not only made by the female members of the board but also by the other male members. As such, it is hard to determine whether these female board members are queen bees and how they influence the gender diversity in the lower levels of the company. Accordingly, the effect of the board’s gender diversity on the company’s diversity might be affected by other board characteristics which could eventually lead to the reported results within this study. Limitations
The main challenges of this study stems from the data scarcity, reversed causality and the influence of external factors not included in the model. Regarding the data, the scarcity of the diversity data that is present leads to a significant decrease in the amount of observations within the sample. Thereby, the outcomes of this study are subject to the expertise, subjectivity and selections of Wharton when they are assigning a value of 1 or 0 to the earlier described dimensions.
Another possible limitation is the presence of reversed causality. Although I aimed to filter only the diversity dimension indicating the gender diversity among the lower level of the companies (women promotion, work-life benefits, women & minority contracting), reversed causality is still possible due to the diversity dimension representation.
25
Finally, I acknowledge that in spite of the included control variables and careful selection of the CEO turnover from only male to female or from male to male, the results might be influenced by external factors. This problem especially arises among the results regarding the effect of the board’s gender diversity. Hereby no control-variables for the board’s characteristic are taken into account.
Insights & research implications
My findings contribute to the ongoing policy debate about the value of regulatory interventions to increase female participation at senior organizational levels and underline the presence and importance of the queen bee phenomenon and the awareness that an increasing share of women in the top-management positions do not necessarily lead to more opportunities for junior women. Thereby, no studies that I know of have quantitatively aimed to research the presence of the queen bee phenomenon among the board and executive level. Thereby, no studies that I know of have quantitatively aimed to research the presence of the queen bee phenomenon among the board and executive level. Implications for further research could build up on this research by improving the empirical model, researching the presence of the queen bee phenomenon outside of the USA and collecting more and precise data.
26
REFERENCES
Aigner, D. J., & Cain, G. G. (1977). Statistical theories of discrimination in labor markets. ILR Review, 30(2), 175-187.
Baron, Reuben M. and David A. Kenny (1986), “Moderator-Mediator Variables Distinction in Social Psychological Research:
Bear, S., Rahman, N., & Post, C. (2010). The impact of board diversity and gender composition on corporate social responsibility and firm reputation. Journal of Business Ethics, 97(2), 207-221.
Bertrand, M., Goldin, C., & Katz, L. F. (2010). Dynamics of the gender gap for young professionals in the financial and corporate sectors. American Economic Journal: Applied Economics, 2(3), 228-55.
Bielby, W. T., & Baron, J. N. (1986). Men and women at work: Sex segregation and statistical discrimination. American journal of sociology, 91(4), 759-799.
Blau, F. D., & DeVaro, J. (2007). New evidence on gender differences in promotion rates: An empirical analysis of a sample of new hires. Industrial Relations: A Journal of Economy and Society, 46(3), 511-550.
Bloom, N., & Van Reenen, J. (2007). Measuring and explaining management practices across firms and countries. The Quarterly Journal of Economics, 122(4), 1351-1408.
Catalyst. (2015). “New global 2014 catalyst census: Women board directors.”Conceptual, Strategic, and Statistical Consider- ations,” Journal of Personality and Social Psychology, 51(6), 1173–82.
Credit Suisse Research Institute. (2014). “The CS gender 300: Women in senior management.” Zurich: Credit Suisse.
Derks, B., Ellemers, N., Van Laar, C., & de Groot, K. (2011). Do sexist organizational cultures create the queen bee? British Journal of Social Psychology, 50, 519–535.
Derks, B., Ellemers, N., Van Laar, C., & De Groot, K. (2011). Do sexist organizational cultures create the Queen Bee?. British Journal of Social Psychology, 50(3), 519-535.
Derks, B., Laar, C., Ellemers, N., & Raghoe, G. (2015). Extending the queen bee effect: How Hindustani workers cope with disadvantage by distancing the self from the group. Journal of Social Issues, 71(3), 476-496.
27
Derks, B., Van Laar, C. & Ellemers, Naomi (2016). The Queen Bee Phenomenon: Why Women Leaders Distance Themselves from Junior Women. Leadership Quarterly, 27 (3), (pp. 456-469).
Dezso, C. L., Ross, D. G., & Uribe, J. (2016). Is there an implicit quota on women in top management? A large‐sample statistical analysis. Strategic Management Journal, 37(1), 98-115.
Dubbink, W. (2005). Democracy and private discretion in business. Business Ethics Quarterly, 15(1), 37-66.
Duguid, M. (2011). Female tokens in high-prestige work groups: Catalysts or inhibitors of group diversification? Organizational Behavior and Human Decision Processes, 116, 104–115..
Durbin, S., & Fleetwood, S. (2010). Gender inequality in employment: Editors' introduction. Equality, Diversity and Inclusion: an international journal, 29(3), 221-238.
European Commission (2014). Report on equality between women and men 2014. Retrieved from http://ec.europa.eu/justice/gender-equality/document/index_en. htm#gender
Faniko, K., Ellemers, N., & Derks, B. (2016). Queen Bees and Alpha Males: Are successful women more competitive than successful men?. European Journal of Social Psychology, 46(7), 903-913.
Faniko, Klea, Ellemers, Naomi, Derks, Belle & Lorenzi-Cioldi, Fabio (2017). Nothing Changes, Really: Why Women Who Break Through the Glass Ceiling End Up Reinforcing It. Personality and Social Psychology Bulletin, 43 (5), (pp. 638-651) (14 p.).
Fernandez-Mateo, I., & King, Z. (2011). Anticipatory sorting and gender segregation in temporary employment. Management Science, 57(6), 989-1008.
Giuliano, L., Levine, D. I., & Leonard, J. (2011). Racial Bias in the Manager-Employee Relationship An Analysis of Quits, Dismissals, and Promotions at a Large Retail Firm. Journal of Human Resources, 46(1), 26-52.
Gopalan, S., & Watson, K. (2015). An agency theoretical approach to corporate board diversity. San Diego L. Rev., 52, 1.
Gould, J. A., Kulik, C. T., & Sardeshmukh (2018), S. R. Trickle‐down effect: The impact of female board members on executive gender diversity. Human Resource Management.
