WOMEN ON BOARD AND FIRM PERFORMANCE:
AN EUROPEAN PERSPECTIVE
Laura Gutium University of Groningen Faculty of Economics and Business Supervisor: Drs. Boris van Oostveen
June 2016
Abstract
The following study uses an European dataset during 2008-2015 and takes a multi-country approach in examining the relationship between board gender diversity and firm performance. The obtained results indicate that the extent of women representation on company boards is negatively associated with firm performance in the short run, while in the long run the effect is insignificant. This supports the argument that more diverse boards lead to lack of communication, more conflicts and less effective decision-making. Moreover, the negative relationship and insignificant results might indicate the presence of gender stereotypes and tokenism on European boards. The main implication related to tokenism is that women are appointed on boards in order to meet the legislation requirements rather than qualifications criteria. Thus, the low authority in management decisions and the lack of necessary skills provide a rationale why women might detract from firm performance. Also, when considering cultural aspects in the key relationship, the results support the argument that in societies that have a more favourable attitude towards the engagement of female directors in the public life, a positive effect of female directors on firm performance is more likely to be seen. The following suggests that cultural aspects cannot be neglected and should be accounted for in both future research and policy-making.
1
1. Introduction
The most recent financial crises and corporate scandals raised an important question: would have losses been anticipated and prevented if more women were leading companies from all over the world (Adams and Funk, 2012)? Theory suggests that the answer might be affirmative as more gender diverse boards lead to a stronger corporate governance and respectively to an increase in firm performance. However, empirical evidence provides mixed results. This study aims to shed more light to the existing literature by examining the impact of board gender diversity on firm performance in 28 European countries in the period 2008-2015. Moreover, it takes into account national cultural differences and investigates if the benefits of board gender diversity are higher in countries that have a more positive and supportive attitude towards gender equality. In order to provide more accurate estimates with respect to the relationship between board gender diversity and firm performance, endogeneity issues are addressed by employing a panel data methodology and instrumental variables approach (IV).
An increasing number of academic publications indicate that gender diversity within the boardroom is an area of interest and development for many interested parties such as boards, investors and academics. According to Doldor et al. 2012, more women directors on boards is desired as it contributes to a stronger corporate governance, wider talent pool, better responsiveness to the market and improved performance. Many countries try to encourage a greater female board representation by imposing gender quotas through legislation. Among the European countries, Norway takes the top position as it has 42 percent of women on boards1. From 2010 till 2014, the largest percentage increase in European countries was registered in France (20 %), Italy (19.6%), Belgium (11.9%), Germany (11.8%), the United Kingdom (10.8%) and Slovenia (10.1 %)2. The main benefit of gender quotas is that they increase the speed with which women are appointed as board members. However, gender quotas may not always be the best option towards an increased performance as they can encourage tokenism on the board3. In other words, women might be appointed on boards in a forced manner just to satisfy legislation requirements and very often they might lack the necessary skills or decision making authority for a good management.
The existing literature on the relationship between women on boards and firm performance provides no clear results. A positive relationship between board gender diversity and firm performance tends to prevail more in US studies, while in Europe the results are more controversial. According to some researchers, some of the main reasons behind the mixed results are related to limitations in terms of methodology, diverse control variables, a short time frame period and a small sample size (Wang et
1 European Commission, 2012. Women in Economic Decision-Making in the EU: Progress Report. European Commission. 2
Delloite. 2014. Women in the Boardroom: A Global Perpective-4th edition. Deloitte Touche Tohmatsu Limited.
3
2 al., 2009). Another important reason that many studies fail to take in consideration is related to the endogeneity issue that may arise as a result of measurement error, omitted variables and simultaneity. This research is unique in several ways. First, it extends the existing literature on board gender diversity and firm performance by employing non-US data from 28 European countries using a time frame of eight years (2008-2015). The majority of existing studies in Europe examine the impact of the extent of female directors on firm performance only in one or two countries and the time span is smaller. Second, this analysis employs a panel regression with fixed effects approach and includes more control variables related to board characteristics. This can be considered as an improvement to the pooled OLS approach as the firm fixed effects help to eliminate unobserved heterogeneity bias, and year fixed effects help to control for the yearly fluctuations in the state economy. Third, this study is one of the few attempts to account for the role of cultural differences in the relationship between board gender diversity and firm performance. Particularly, it is examined whether in countries characterized by a more supportive environment for women a more positive effect of gender diverse board on firm performance prevails. This could be the case of fewer gender stereotypes with respect to women in managerial roles and lower practice of tokenism on boards. Last, the endogeneity issue between women directors and firm performance is encountered not only by using the fixed effects approach but also by using the instrumental variable specification (IV). The main model is estimated via the two-stage least squares (2SLS) method.
The study results provide evidence that greater female representation on company boards is negatively associated with firm performance in the short run, while in the long run the effect is insignificant. The negative results indicate that board gender diversity leads to communication barriers, more conflicts and less effective decision making and hence, a lower firm performance. Another important consideration is that the impact of female directors on firm performance can be moderated by institutional aspects that are not related to the firm such as cultural aspects. The results indicate that in countries characterized by lower gender gaps the effect of women on firm performance is positive. The main reason is that in these countries, society tends to have a more positive and supportive attitude towards women in managerial roles and board positions. Also, gender stereotypes related to women’s low managerial capabilities are less pronounced. Moreover, females are less likely to be perceived as tokens on boards or in other words they are less likely to have a low management authority as they are not appointed only to meet the legislation requirements. Both tokenism and gender stereotypes may lead to a strong performance pressure that does not allow females to express their full potential and in consequence they may detract from firm performance.
3 main characteristics of the sample. Section four describes the research methodology. Section five presents the obtained results. Finally, section six presents the concluding remarks, the main limitations of the study and further research implications.
2. Literature Review and Hypotheses
In recent years, board dynamism became a topic of interest both in theory and practice as in the long term the company’s success and profitability depends on the board’s decisions. The more diverse the board is, the better the organization is integrated in the global, political and economic environment. The concept of board diversity indicates a heterogeneous board composition in terms of individual characteristics such as age, gender, nationality, education, lifestyle and experience4. Board diversity can be divided into two main categories: surface level and deep-level diversity5. Surface level diversity is classified into two categories: observable and unobservable diversity. Observable diversity refers mainly on demographic characteristics such as age, gender and race while non-observable diversity includes aspects less visible such as education and experience of the board members.
Much of the diversity research is based on demographic characteristics. According to Rosenzweig (1998), the main explanation relates to the existence of reliable databases which make objective measurement with respect to both observable and non-observable diversity possible. Also, many empirical studies assumed a systematic correlation between cognitive variables and demographic variables (Peterson and Philpot 2007; Smith, 2007; Rose, 2007). As a consequence, gender diversity contributed to a vast collection of research with regard to demographic diversity. As gender inequalities at work still continue to persist all over the world, the main aim of this research study is to shed more light in terms of the effect of board gender diversity on firm performance.
In many actual studies, the theoretical framework assumes that a good corporate governance is essential when examining the relationship between firm performance and board diversity (Carrasco and Laffarga, 2006; Ruigrok et al. 2006). Thus, lower board gender diversity may be a cause of poor corporate governance. Furthermore, the link between good corporate governance and performance is emphasized in the financial literature and a lot of attention is given to the board characteristics. With regard to the dominant theories in studies of corporate governance, Kiel and Nicholson (2003) state that there is no theory that offers an extensive framework to explain the strength of the relationship between board diversity-firm performance. The following may be due to the versatile nature of the topic. According to Carter et al. (2003), until a stronger theoretical framework that gives a clear prediction of the role of board diversity in firm value is developed, more empirical examination is required.
4
Mishra, R.K., and Jhunjhunwala, S., 2013. Diversity and the Effective Corporate Board. NY: MacMillan.
5
4 This section first provides an overview of four dominant theories that take a positive approach towards gender diversity: agency theory, stakeholder theory, resource dependence theory and stewardship theory. Then, it discusses relevant studies on board gender diversity and firm performance that lead to the development of the research methodology and provide a basis for the study’s empirical tests. Finally, the relation between board gender diversity and firm performance is discussed in terms of national cultural differences.
2.1. Theoretical Framework Agency Theory
The most important theory in the literature that tries to explain the relationship between board gender diversity and firm performance is the agency theory. Its main message is that weak corporate governance results in agency costs and consequently, in a poorer firm performance (Core et al., 2006). According to Finegold et al. (2007), the agency theory perspective’s main purpose is to increase financial performance by reducing agency costs, caused by a misalignment of the interests of managers with those of board directors. Fama and Jensen (1983) state that boards of directors are responsible to monitor and control managers in order to ensure that interests of both managers and shareholders are aligned. They also consider that the board of directors should be independent for a better performance. A more diverse board leads to more independence as outside directors that differ in both observable and unobservable characteristics are more likely to see the problem from a different perspective (Carter et al., 2003). By choosing the right board characteristics, the interests of both managers and board members become better aligned and as a consequence the agency problems are reduced (Hillman and Dalziel, 2003). For instance, empirical evidence suggests that women are stricter monitors and thus, having women on boards may lead to greater audit efforts and managerial accountability (Adams and Ferreira, 2009). Also, Singh and Vinnicombe (2004) state that women tend to take their roles on the board very seriously, which may result in better governance. Therefore, agency theory implies that gender diversity on boards may be seen as a tool for decreasing agency costs and, hence, improving firm performance. This argument applies only when heterogeneous boards are considered as a means for a better control. The higher the level of board heterogeneity, the better is the company access to the multitude of skills, knowledge, resources and innovations that are critical for the company’s long-term stability and profitability.
Stakeholder theory
5 and Rindova, 2001). According to Burke and Mattis (2000), women understand the market more, so they could be more willing to listen to the customer’s needs in terms of products and services. Also, the transformational leadership styles of women classify them as good listeners when it comes to the main needs of their employees (Eagly et al., 2003). As the importance of women as a global consumer continues to rise at a very rapid pace, women directors can establish better relationships with these groups of customers, fact that can be immediately observed in specific industry sectors. For example, women directors are considered to be more receptive when dealing with social and environmental issues (Nielsen et al., 2010; Williams, 2003). Moreover, Bear et al. (2010) provides evidence that a higher number of women board members is associated with better Corporate Social Responsibility (CSR) ratings. Social and ethical responsibility issues may have an important effect on corporate reputation as they are related to stakeholder’s perceptions with respect to the company’s core activities (Roberts and Dowling, 2002). According to Bernardi et al. (2006), Branco et al. (2008) and Bramer et al. (2009), having women directors on the board contributes to reputation enhancement and respectively to firm prosperity in the long term. In sum, stakeholder theory argues that board gender diversity indicate more transparent government processes that ultimately support the interests of different stakeholders (Hillman et al., 2002).
Resource dependence theory
6 following classification with respect to four types of directors: insiders, support specialists, business experts, and community influentials. Different types of directors bring different valuable resources to the firm. Hence, a more gender diverse board will provide beneficial resources that ultimately will lead to an overall increase in the firm performance.
Stewardship theory
Last but not least, stewardship theory assumes that managers act in the best interests of the shareholders and they allow the boards to concentrate on long-term strategic planning rather than on methods of reducing agency costs (Huse, 2005). Contrary to the agency theory, stewardship theory emphasizes the positive aspects of human behaviour by assuming that the behaviour of the agents in the firm is not an opportunistic one (Davis et al, 1997). The main priority of agents is firm performance rather than personal interests (Nordberg, 2008). The loyalty of managers towards the company is attributed to the need to achieve, to gain intrinsic satisfaction from successfully performed tasks and to achieve recognition from the superiors. Davis et al. (1997) states that the behaviour of stewards tends to be collectivistic, indicating a larger room for cooperation between board members and managers. The collectivistic behaviour develops trust, valuable competences, experiences and ease of communication, characteristics that ultimately lead to business success (Muth et al., 1998). There is evidence in the literature that women are more focused on improving communication channels and networks within the organization (Claes 1999). Also, Eagly et al. (2003) describes women as being more likely to transform their leadership styles due to their aligned communal behaviour. Therefore, gender diversity in the boardroom may influence the firm performance through the assumption that women may act as guardians of the firm’s assets, build stronger networks and communicate more closely with senior management.
7 H1: There is a positive relationship between board gender diversity and company performance.
2.2. Empirical Evidence
The existing literature regarding board diversity is expanding rapidly. However, the majority of the research is focusing on US data and provides mixed results regarding gender diversity and firm performance. A summary of well-known studies on board gender diversity and firm performance is provided in Table A.1 (See Appendix). Generally, the existing studies can be classified by the results they find: positive, negative or neutral relationships. According to Smith et al. (2006), boards with more women achieved higher performance among the 2500 largest firms in Denmark in the period 1993-2001. A positive relationship between the presence of women or minorities on the board and firm value is also detected by Carter et al. (2003) in a study of 1000 Fortune firms. Likewise, Luckerath-Rovers (2013) examines the relationship on 99 listed companies in the Dutch Female Board Index by using a time frame of two years (2005-2007) and finds significant positive results. Outside US and Europe, evidence is still scarce, however positive results are found by Liu et al. (2013) in China and Vafaei et al. (2015) in Australia. Among the main reasons found in the literature regarding the positive relationship between board gender diversity and firm performance are: a better understanding of the market environment, a better public image of the company and an increase in the external talent pool (Smith et al., 2006).
Contrary to the evidence provided above, Adams and Ferreira (2009) detected in their research study on 1500 US firms from 1996 to 2003 a negative relationship between gender diversity and firm performance. Similar results are found by Carter (2010) and Levi et al. (2014) who also conducted their studies on US companies and used ROA and Tobin’s Q as performance measures. Al-Shammari and Al-Saidi (2014) used a sample of 121 firms listed on Kuwait Stock Exchange from 2009 to 2011 and supported again that a greater number of female on board significantly reduces the performance of the firm. The negative relationship can be explained by the fact that more diverse boards require more time to take the right decisions (Ancona and Caldwell 1992).
Some other studies (Bianco et al., 2011; Wachudi, 2005) concluded that there is no impact of gender diversity measured as percentage of female directors on the performance of the firm. One of the main reasons why no significant relationship was found is the small number of females included in the study sample (Wang et al., 2009).
8 As it can be seen from the above arguments, there is a vast collection of studies in the literature that account for the effect of gender diversity on firm performance but unfortunately the mix evidence of results makes them to be very equivocal. Especially, the mixed results tend to prevail in Europe if compared to the existing results from US studies. Researchers believe that this happens due to methodology limitations, diversity of control variables, short time frame and small sample size (Wang et al., 2009). According to Campbel and Minguez-Vera (2008), the mixed results could also be caused by the fact that endogeneity issues between gender diversity and firm value are not taken in account properly. According to Wintoki et al. (2010), endogeneity may arise due to unobserved heterogeneity, simultaneity and dynamic endogeneity. Unobserved heterogeneity occurs when an unobservable factor may affect the relation between variables. For example, the CEO’s level of risk aversion may affect the firm value (Haubrich, 1998). The main reason of simultaneity is that two variables are co-determined implying that each variable may influence the other simultaneously. For instance, simultaneity occurs when a company selects the governance structure based on previous performance. Dynamic endogeneity appears when the current value of a variable may be influenced by its past value. For example, past poor firm performance may lead to stricter governance controls that ultimately will affect the current firm performance and its governance structure (Hermalin and Weisbach, 1998). The increasing evidence that confirms the existence of the three potential sources of endogeneity mentioned above indicates that if endogeneity issues are not taken in consideration, the estimates may lead to spurious results (Hartzell et al., 2006; Lilling, 2006).
2.3. A multi-country study: attitudes towards women’s involvement in civic life
9 is difficult to define and operationalize (Taras et al., 2009). Its origins date centuries ago and until now many researchers tried to measure different aspects of culture but only a few succeeded to obtain models of culture that gained a worldwide acceptance and popularity (Hofstede, 1980; Schwartz, 1994; Trompenaars, 1993). Some scholars associate culture with personality (Hofstede et al., 2004; Steel et al., 2002). In this study the term of culture is represented by society’ attitude towards females. The same approach is taken by Low et al. (2015) on a sample of Asian companies form Hong Kong, Malaysia, Singapore and South Korea. Among the most important constructs that try to measure the society’s attitude towards females are: World Economic Forum Global Gender Gap Index, United Nation’s Development Programme’s Gender-Related Development Index and Gender Inequality Index, Economist Intelligence Unit Women’s Economic Opportunity Index and Social Watch’s Gender Equity Index. Basically, it is assumed that the importance of board gender diversity depends on cultural and social environment. For example, in countries where gender stereotypes are more pronounced and where social networks and promotion systems tend to be biased towards men the extent of women on board is significantly low (Brooks, 2006; Terjesen and Singh, 2008). In terms of corporate culture, in a survey conducted by McKinsey and Company it was found that men tend to believe that gender diversity measures are unfair and combining parenting with top careers is seen more difficult for women6. Clearly, these are barriers that make it harder for women to find their way to the top. According to Hofstede (2001), the attitude of society towards females tends to be more discriminatory in more masculine countries. In such countries females tend to have lower wages and less important authority in management. Also, specific characteristics for such an environment are a high degree of tokenism and gender stereotype.
Tokenism
The concept of tokenism was first proposed by Kantner (1977), who described women on boards as ‘tokens’ or in other words, individuals who represent a particular group based on such characteristics as race and gender. In her study, tokenism leads to three main behavioural consequences: visibility, polarization and assimilation. Visibility threat happens when tokens feel that they are monitored all the time and due to this they experience a performance pressure. Polarization refers to a situation when the dominant group, usually represented by males feels threatened by tokens and try to set high barriers between them or even exclude the tokens from the group. The last behavioural characteristic- assimilation causes misconceptions of the real personality of the token because he/she is judged by the dominant group on a stereotype basis. As tokens represent the minority group on the board, they can be both males and females. All these three aspects can have severe consequences for the company as they can lead to a reduced firm performance.
6
10 In more general terms, women are appointed on board of directors in order to meet the legislation requirements but their authority in decision-making process is highly limited (Zimmer, 1988). In the prevalent literature, tokenism is frequently associated with the notion of ‘critical mass’. The theory behind it stresses that women have a higher power in terms of management only when their extent is sufficiently high (Low et al., 2015). Moreover, Kristie (2011) states that ‘one female is a token, two is a presence and three is a voice’. Only in this case, women may have a word to say and respectively the society’s attitude towards them may become more favourable. If there are three or more women on board their influence in terms of management decisions may increase significantly (Kramer et al., 2007). It is interesting to examine this effect on European countries, especially if it is accounted for cultural differences between East and West Europe. According to Terjesen and Singh (2008), the degree of tokenism is different for each culture. To sum up, understanding the concept of tokenism can provide important reasons why women still tend to cope with difficulties at the workplace and especially if they hold positions that are more specific to males.
Gender stereotype and stereotype threat
Nowadays, stereotypes against women still tend to prevail in jobs that previously were considered typical male jobs (Bergeron et al., 2006). According to Eagly et al., (2001) this is due to the wrong perceptions of the society that women do not have good management skills and that they are not associated with good leaders. Evidence shows that females and males may be different in their core values and risk attitudes but this may not necessarily lead to different priorities or more risk-averse decision making between genders (Adams and Funk, 2012). The following aspect can lead to a collapse in terms of gender stereotypes. Moreover, Chia et al., (1994) states that gender stereotypes are shaped by the culture of a particular country and that they can be totally different across countries. The term of stereotype threat is observed in the majority of cases among minority groups that are responsible for certain activities at which society believes they fail most of the times (O’Brien et al., 2003). According to Roberson and Kulik (2007), stereotype threat is defined as ‘the fear of being seen and judged according to a negative stereotype…’. These threats can have a huge impact on the firm performance as they lead the minority group to think too much on personal performance and at the beliefs of others (Cadinu et al., 2005). In the end, stereotype threat may cause a lot of distress among the minority groups (Steele and Aronson, 1995).
11 degree of tokenism and stereotype threat the more likely that females will lose from their concentration and confidence level in the decision-making process. Taking in consideration the impact of society’s attitude towards women in the relationship between board gender diversity and firm value, the next hypothesis is proposed:
H2: The influence of board gender diversity is greater in environments where a positive society’s
attitude towards women prevails.
3. Research Methodology
The main model and estimation method The next is the main regression model:Performanceit = β0+β1Femalesit+ β2Controlsit + αi+ λt + εit (1)
where, i is an index for the company, t is an index for the time, Performance is a measure for firm performance, β0 is a constant, β1 and β2 are coefficient vectors for Females and Controls, αi is a
firm-fixed effect, λt is a time-fixed effect and εit is the disturbance term. Performance is measured by two
indicators: return on assets (ROA) and Tobin’s Q. Females is measured by three indicators: the total number of female directors on the board, the percentage of females and the female dummy variable. Controls are divided into two sets: board characteristics and firm characteristics.
According to Liu et al. (2014), the two main estimation methods that prevail in the literature regarding the relationship between board diversity and firm performance are: pooled ordinary least square regression (OLS) and panel regression with fixed effects. The main regression model (1) will be estimated using both methods. Then, F-test will be used in order to determine which method is more suitable for the following research. Considering that the null hypothesis for the F-test is the absence of both firm fixed effect and unobserved heterogeneity, the decision to reject the null hypothesis will be made at a significance level of 5 %. An F-test statistic greater than the respective critical value at the 5 % significance level will indicate that the pooled OLS method should be rejected in support of the panel regression method with fixed effects. The main advantages of the panel approach are that αi
contributes in excluding important variables that were omitted by the researchers and λt accounts for
yearly variations in the economy of the particular country.
12 expected that the influence of the extent of females on performance to be stronger in countries characterized by lower gender inequalities.
According to Low et al. (2015), due to the fact that the extent of females and the interaction term are highly correlated, dummy variables for Global Gender Gap Index (GGI) are created and used in the interaction term instead of the original GGI scores. For each year during the period 2008-2015, the individual country GGI scores are compared to the median for all the countries available in the World Economic Forum list. Therefore, two dummy variables are created: High GGI indicating an above-median score and Low GGI indicating a below-above-median score. An above-above-median score indicates that a particular country is more likely to fully close the existing gender gaps in the nearest future. The performance models with the inclusion of the high interaction term (Females*High GGI) and low interaction term (Females*Low GGI) are presented below:
Performanceit = β0+β1Femalesit+ β2Femalesit*High GGI+ β3Controls+ αi+ λt + εit (2)
where, β1, β2 and β3 are coefficient vectors for Females, Females*High GGI and Controls.
Performanceit = β0+β1Femalesit+ β2Femalesit*Low GGI+ β3Controls+ αi+ λt + εit (3)
where, β1, β2 and β3 are coefficient vectors for Females, Females*Low GGI and Controls.
Both pooled OLS estimates and panel regression with firm-fixed and time-fixed effects are undertaken. As it was previously pointed out, endogeneity issues tend to prevail in studies that examine the governance-performance relation. According to empirical evidence, simultaneity is one of the most encountered source for endogeneity with respect to board gender diversity and firm performance (Bianco et al, 2011; Marinova et al., 2010). However, fixed-effect panel specifications provide consistent parameter estimates only under the assumption of strict exogeneity. This implies that corporate governance variables are orthogonal to past, present and future changes with respect to firm performance (Schultz et al., 2010). Thus, the fixed effect method may not be an adequate control for sources of endogeneity such as simultaneity and dynamic endogeneity. One way to address simultaneity is to use instrumental variables (IV) in a two-stage least square regression (Carter et al., 2003).
Culture as a moderating effect and Endogeneity
In the next part of this study, the moderating cultural effect represented by society’s attitude towards women at work on the relationship between board gender diversity and firm performance will be tested using the two-stage least square method (2SLS).
13 reverse causality is addressed using an instrumental approach. First, instrumental variables are selected and then the main model is estimated using a two-stage least square method (2SLS). The main concern with the instrumental approach is in choosing the instruments that meet the instrument exogeneity and relevance conditions (Adams and Ferreira, 2009). The Durwin-Wu Hausman instrument test will be used to check if the chosen instruments satisfy the exogeneity requirement (Davidson and MacKinnon, 1993). The first stage model related to the two-stage least square method can be seen below:
Femalesit = β0+β1Instrumentsit+ β2Controlsi+ λt + εit (4)
where, i is an index for the company, t is an index for the time, β0 is a constant, β1 and β2 are
coefficient vectors for Instruments and Controls, λt is a time-fixed effect and εit is the disturbance term.
Model (4) will be estimated in order to obtain the fitted values of Females measured by the total number of female directors on the board, percentage of females, a female dummy variable and those of two sets of control variables: board characteristics and firm characteristics. The second stage regressions are estimated using the fitted values of model (4). Culture is accounted in the model through the inclusion of an interaction term between the extent of females directors and Gender Gap Index (GGI). Model (5) is undertaken with the high interaction term and model (6) with the low interaction term, respectively.
Performanceit = β0+β1Femalesit+ β2Femalesit*High GGI+ β3Controls+ λt + εit (5)
Performanceit = β0+β1Femalesit+ β2Femalesit*Low GGI+ β3Controls+ λt + εit (6)
As it can be seen from both models (5) and (6), the firm-fixed effects were not accounted for due to the fact that the cultural (GGI) predictor changes over time but slowly and this may lead to statistically insignificant results.
4. Data and Descriptive Statistics
The following study distinguishes itself from existing scientific literature on board gender diversity, as it takes a multi-country approach and concentrates on firms located in 28 countries from both Eastern and Western Europe. Additionally, the data set used can be considered an improvement when compared to previous research as it captures a time frame of eight years (2008-2015). According to Campbell and Minguez-Vera (2008), panel data is preferred over cross-sectional data because it leads to a more reliable analysis and it reduces the biases regarding unobservable heterogeneity and neglected variables.
14 European area was included in the dataset. Second, firm characteristics data was retrieved from Bureau van Dijk’s Orbis database- a world of company information. From the initial sample, firms with missing specific board or financial data for the span 2008-2015 were eliminated. This lead to a final sample of 989 companies and 4348 observations. Additionally, country scores on Global Gender Gap Index (GGI) are manually collected for each country and year from eight Global Gender Gap Reports for the time period between 2008-2015. The Global Gender Gap Reports date since 2006 and are published yearly by the World Economic Forum. A more detailed description of variables used in the following research study is provided below, in Table 1.
Table 1
Definition of board, firm and country characteristics for 28 European countries in the period 2008-2015.
Variable Operationalization Source
Board Characteristics
Number of females Total number of female directors on the board at the end of the year BoardEx Percentage of females
Total number of female directors on the board divided by the total number of
board directors at the end of the year BoardEx Female dummy
Dichotomous variable that equals 1 when the board has one or more female
directors at the end of the year and 0 otherwise BoardEx Board size Total number of directors on the board at the end of the year BoardEx Average time in role The length of time, stated in years, that the director has been in the current role BoardEx Average years on other
quoted boards
Total number of years a director has sat at board level on quoted boards, divided
by the number of them BoardEx
Average age The sum of the directors ages being measured divided by the number of directors BoardEx Average education Total qualifications gained by the directors divided by the number of directors BoardEx Nationality mix
Proportion of directors from different countries. A high number depicts a more
diverse board. BoardEx
Firm Characteristics
Return on assets (ln) Annual net income divided by the book value of total assets at the end of the year Orbis Tobin’s Q Total market value of the firm divided by total asset value of the firm Orbis Firm age (ln) Natural logarithm of the number of years since incorporation Orbis Total assets (ln) Natural logarithm of the book value of total assets at the end of the year Orbis Number of employees (ln) Natural logarithm of the total number of employees at the end of the year Orbis Industry Dummy Dummy variable that equals one for production firms or zero for service firms Orbis Gearing Long-term liabilities divided by the capital employed Orbis
Country Characteristics
Culture (GGI) Gender Gap Index that takes values between 0 and 1 GGGR7 Culture (High GGI)
Dummy variable that equals 1 for countries with GGI higher than the world
median GGGR Culture (Low GGI)
Dummy variable that equals 1 for countries with GGI lower than the world
median GGGR
7 Global Gender Gap Report (GGGR) is published annually since 2006 by the World Economic Forum. In order for the
15 4.1. Firm Performance
In the following research study, the dependent variable is represented by the firm performance. The existing research studies distinguish between two main categories of firm performance measurement: accounting-based and market-based performance measurement. The most used accounting-based measure are return on asset (ROA) and return on equity (ROE) while the most used market-based measure is Tobin’s Q. The majority of studies try to combine both types of measure in their study and most of the times both ROA and Tobin’s Q are chosen (Adams and Ferreira, 2009; Carter et al., 2010; Darmadi, 2013; Levi et al., 2014). The main advantage of Tobins’Q as a performance measure is that it uses the right risk-adjusted discount rate and diminishes the impact of different tax regulations and accounting practices among countries (Wernerfelt and Montgomery, 1988). In this study, both ROA and Tobin’s Q are used as measures of firm performance. A more detailed description regarding the computation of the firm performance measures is provided above, in Table 1.
4.2. Board Gender Diversity
In this study, board gender diversity is represented by three measures: total number of females on board of directors, percentage of females and a female dummy variable that equals one when there is at least one female on the board and zero otherwise. The same approach was taken by Adams and Ferreira (2009), Rose (2007), Vafaei et al. (2015) and Wachudi (2009). Information regarding gender diversity was directly retrieved from BoardEx database as it provides both general information such as percentage of males on board and more detailed information such as the name of each board member.
4.3. Control variables
16 directors and firm performance, indicating that the director’s years of experience and wisdom are essential for a higher firm performance. Thus, a positive relationship between the age of directors and firm performance is expected. Average education is measured by the total qualifications gained by the directors divided by their total number. Previous research indicates that CEO’s holding degrees from renowned universities are more successful in terms of firm performance (Darmadi, 2011). This may suggest that a higher education is equivalent to a better management of scarce resources. Consistent with the findings of previous studies (Yermack, 2006; Ujunwa, 2012), a positive relationship between the education level of board members and firm performance is hypothesized. Nationality mix is measured by the proportion of directors from different countries. According to the agency and resource dependence theory, nationality diversity brings huge competitive advantages as it provides a better access to worldwide information and strengthens the international networks (Marimuthu and Kolandaisamy, 2009; Oxelhei and Randoy, 2003). Hence, a positive relationship between board nationality and firm performance is expected.
17 thus this leads to a lower firm gearing (Francoeur et al., 2008). A high level of gearing indicates that the firm is more exposed to bankruptcy risk and respectively managers are required to put more effort in order to increase the performance of the company (Jensen and Meckling, 1976). Therefore, it is expected that there is a negative relationship between the gearing level and firm performance.
4.4. Instruments
18 4.5. Culture
According to the Global Gender Gap Report 2013, the competitiveness of a particular nation is also affected by the way how it educates and takes advantage of its female talent pool. Thus, closing the gender gaps all over the world is not only about ethics and respecting human rights but also about efficiency.
In the following study, culture is defined as the country’s attitude towards women at work and is measured by the World Economic Forum Global Gender Gap Index (GGI)8. A more detailed structure of the Global Gender Gap Index is given in Table A.2 (See Appendix). Some other alternatives for GGI found in the existing literature are: Social Watch’s Gender Equity Index, United Nations’ Development Programme’s Gender-Related Development Index and Gender Inequality Index (Low et al. 2015). The scores for each country in the dataset are retrieved from the Global Gender Gap Report published by the World Economic Forum for each year separately. Detailed information about individual country scores during 2008-2015 is provided in Table A.3 (See Appendix). In order to determine whether the influence of board gender diversity on firm performance is affected by society’s attitudes toward women in managerial roles, an interaction term between a high/low dummy GGI and board gender diversity measures is used. The high GGI and low GGI are constructed using the world median GGI score. Basically, countries that have a higher score than the world median are characterized by a high GGI and those that a have a lower score than the world median by a low GGI, respectively. The GGI scores for each country included in the report seem to change over time but very slowly. The main aim of the index is to incorporate changes in gender gaps between continents, countries and time periods.
4.6. Descriptive Statistics
In terms of descriptive statistics of all variables used in this study, relevant information is provided below, in Table 2. The total number of observations for all variables, besides nationality mix and the natural logarithm of total assets is 4348. On average, the total number of females on boards is 1.042 and the total percentage of females is 8.777 (%). The maximum number of females on boards in Europe is 9.019 and if expressed as a percentage this is equivalent with half of the board members. The mean of the board size is 11.437. In terms of tenure, the mean of the average time in role is 3.941 years, the minimum is 1.5 years and the maximum is 15.45 years. These results are close in values with the average time spent on other quoted boards (mean of 2.807 years, maximum of 12.350 and minimum of 0 years). The average age of board members is 56.087 years. According to this sample, the youngest board member is 39.750 years while the oldest is 77.750 years. In terms of education, the average number of qualifications held by board directors is 1.645. The nationality mix indicator of
19 0.164 shows that the proportion of directors from different countries is quite low. At the country level, it can be seen that no country fully closed the gender gap as no country reached a score of 1 yet. The maximum GGI achieved is 0.850.
An interesting perspective is to see how board characteristics change over time. Descriptive statistics for all variables during each year from 2008-2015 is provided in Table A.4 (See Appendix). Generally speaking, the majority of board variables tend to increase over time. During 2008 until 2015, the average number of females on board increased from 0.689 to 1.493 (116.63%) and the average percentage of females from 6.074 to 12.481 (105.48%). The average time in role and years on other quoted boards increased from 3.782 to 4.175 (10.391%) and from 2.499 to 3.136 (25.490%) respectively. The average age of board members increased from 55.417 to 57.229 years (3.270%). The education level of board members also shows a positive trend, but the increase occurred very slowly from 1.625 to 1.634 (0.554%). A similar trend is followed by nationality mix from 0.153 to 0.172 (12.418%) and GGI from 0.733 to 0.764 (4.087%).
Table 2
Descriptive statistics of board, firm and country characteristics for 28 European countries in the period 2008-2015.
Mean Std. Dev. Min Max Obs.
Board Characteristics
Number of females 1.041 1.154 0.000 9.019 4,348 Percentage of females 8.777 8.854 0.000 50.000 4,348 Female dummy 0.683 0.465 0.000 1.000 4,348
Board size 11.437 4.930 2.000 30.000 4,348
Average time in role 3.941 1.295 1.500 15.450 4,348
Average time on other quoted boards 2.807 1.948 0.000 12.350 4,348
Average age 56.087 4.564 39.750 77.750 4,348 Average education 1.645 0.613 0.000 4.650 4,348 Nationality mix 0.164 0.174 0.000 0.800 4,331 Firm Characteristics Return on assets (ln) 1.516 1.148 -6.215 4.535 4,348 Tobin’s Q 34.233 148.810 0.000 2629.318 4,348 Firm age (ln) 3.624 1.081 0.000 7.151 4,350 Total assets (ln) 14.838 2.088 4.842 21.227 4,346 Number of employees (ln) 8.473 2.063 0.000 13.292 4,348 Industry dummy 0.371 0.483 0.000 1.000 4,348 Gearing 56.482 100.893 0.000 912.387 4,348 Country Characteristics GGI 0.743 0.037 0.583 0.850 4,348
20
Table 3
Pairwise correlation matrix.
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
1 Number of females - 2 Percentage of females 0.84 - 3 Female dummy 0.62 0.68 -
21 One sign of multicollinearity seems to be between total assets and number of employees with a correlation coefficient of 0.68. However, the high correlation coefficient is logical as both variables measure the firm size. Also, total assets and number of employees have a correlation with board size of 0.64 and 0.50 respectively. This is obvious as the board size tends to increase with the firm size. Lastly, a correlation coefficient of 0.46 was detected between board size and total number of females, which again is rational as females are part of the total board members. Overall, it can be said that there is no sign of severe multicollinearity that could affect the efficiency of OLS coefficients. According to Brooks (2008, pp. 174), as long as the data is used with both cross-sectional and time series dimensions the problem of multicollinearity is reduced significantly.
5. Results
As the following study tries to determine the effect of female directors on board performance in Europe during 2008-2015 and see if society’s attitude towards women’s engagement in public life plays the role as a moderator in the relationship between board gender diversity and firm performance, this section is organized as following:
First, the results from OLS estimations without the inclusion of firm-fixed and time-fixed effects will be presented. Then an interaction term between the extent of female directors and culture construct represented by Global Gender Gap Index will be added in the OLS estimations as an additional explanatory variable. This interaction term acts as a moderator in the relationship between board gender diversity and firm performance. As the regression models are estimated first with a high interaction term, positive coefficients will indicate that the extent of females directors in countries with less gender gaps have a positive effect on firm performance.
Second, in order to account for the omitted variables in panel data the OLS estimations are presented with the inclusion of both firm-fixed and time-fixed effects. Then, again the previous procedure is repeated and the interaction term between female directors and Global Gender Gap Index is added to the model.
Third, as literature suggests that there may be an endogeneity problem in the relationship between board gender diversity and firm performance, the results of instrumental variables estimations are presented with the inclusion of the culture construct in the second stage of the estimations.
22 employees, industry dummy and gearing). The OLS regression models are fitted using robust standard errors.9
Table 4
Firm performance and the extent of female directors in European firms in the period 2008-2015.
This table shows the results of OLS estimations without inclusion of firm-fixed and time-fixed effects, using a European unbalanced sample comprising 28 countries during 2008 to 2015. Data regarding board-level characteristics are retrieved from BoardEx database and firm-level characteristics are retrieved from Bureau van Dijk’s Orbis database. Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (Tobin`s Q) regressed on the number of female directors, percentage of female directors and female dummy respectively. Number of females equals the number of female board directors, percentage of females equals the number of female board directors divided by the total number of board directors, and female indicator is a dummy variable that equals 1 when a firm has one or more female board directors and 0 otherwise. Control variables are divided into two sets: board characteristics (board size, average time in role, average time on other quoted boards, average age, average education, and nationality mix) and firm characteristics (firm age, total assets, number of employees, industry dummy and gearing). The operationalization of all variables is the same as in Table 1. The natural logarithm is used for return on assets, firm age, total assets and number of employees. All variables are measured at the end of the year. Asterisks indicate significance at 0.01(***), 0.05(**), and 0.1(*) levels respectively. Robust standard errors are reported in parentheses under the coefficients.
9
In order to ensure that the OLS estimates are BLUE, several diagnostic tests related to the classical linear regression model are performed. According to the results of Jarque-Bera tests (6 in total), the null of normality is rejected at 1 % significance level in all OLS estimations indicating that residuals are not normally distributed. According to Brooks (2008), the violation of normality assumption is inconsequential as long as the sample size is sufficiently large. The results of Breusch Pagan tests (6 in total) reject the null of homoscedasticity at 1 % significance level in all OLS estimations indicating the presence of heteroscedasticity in panel data. Also, Wooldridge tests (6 in total) for autocorrelation in panel data reject the null of no first order autocorrelation at a significance level of 1 % in all estimations indicating that there is a problem of autocorrelation in this specific sample. The Variance Inflation Factors (VIF) show no sign of multicollinearity as in all OLS estimates they reach values below 3.00. All test results are available on request by the author. In order to correct for heteroscedasticity and autocorrelation, robust standard errors are used.
23 According to Column 1, 2 and 3 it can be seen that there is no impact of the extent of female directors on firm performance measured by return on assets (ROA). However, when firm performance is measured by Tobin’s Q (Column 4, 5, 6) the results become negative and significant at (1 %, 1% and 5% significance level) in all three OLS estimations (β4=-6.890; β5=-0.613; β6=-11.045), indicating that
female directors detract from firm value. The negative results are consistent with the work of Adams and Ferreira (2009), who argue that more gender diverse boards lead to over monitoring and as a consequence, this may lead to a decrease in firm performance. The negative relationship could also be explained due to an increase in gender quotas in European countries during 2008-2015. Gender quotas result in younger and less experienced boards, increases in debt and acquisitions and reduction in operating performance (Ahern et al., 2012).
In order to see if there is an effect of females on firm performance in a more supportive environment and a less supportive environment respectively, the same OLS regression models were performed first with inclusion of a High interaction term between the extent of female directors and Gender Gap Index (GGI) as an additional explanatory variable, and then with a Low interaction term (See Table A.5, Appendix). The obtained results with the High interaction term are reported below, in Table 5.
The results from Column 1, 2, 3, 4, 5 and 6 indicate a negative effect of number of females, percentage of females and female dummy on firm performance measured by both return on assets (ROA) at 1%, 5% and 5% significance level (β1=-0.120; β2=-0.012; β3=-0.155) and Tobin’s Q at 1 % in all three
estimations (β4=-20.476; β5=-2.045; β6=-35.544). Again, the findings suggest that board gender
diversity is costly to the firm as it leads to a reduction in firm performance. From a theoretical perspective, among the main reasons that could explain the negative relationship are the lack of communication (Adams and Ferreira, 2007), emotional conflicts (Williams and O´Reilly, 1998) and more time-consuming and less effective decision making (Lau and Murnighan, 1998).
With respect to the effect of females directors on firm performance in a more supportive environment for females, the obtained results indicate that women directors in societies with a lower gender gap (high GGI score) tend to have a significant positive impact on both return on assets (β1=0.125;
β2=0.013; β3=0.154) at a significance level of 1%, 5% and 5% and Tobin’s Q (β4=14.322; β5=1.516;
β6=26.344) at a significance level of 5 % in all three estimates. The significant and positive
24
Table 5
Firm performance and the extent of female directors in European firms in the period 2008-2015.
This table shows the results of OLS estimations with inclusion of an interaction term between the extent of female directors and the Gender Gap Index (GGI), using a European unbalanced sample comprising 28 countries for the period 2008 to 2015. High interaction term is only reported. High GGI is a dummy variable that equals 1 for countries with Gender Gap Index higher than the world median and zero otherwise. Data regarding board-level characteristics are retrieved from BoardEx database and firm-level characteristics are retrieved from Bureau van Dijk’s Orbis database. Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (Tobin`s Q) regressed on the number of female directors, percentage of female directors and female dummy respectively. Number of females equals the number of female board directors, percentage of females equals the number of female board directors divided by the total number of board directors, and female indicator is a dummy variable that equals 1 when a firm has one or more female board directors and 0 otherwise. Control variables are divided into two sets: board characteristics (board size, average time in role, average time on other quoted boards, average age, average education, and nationality mix) and firm characteristics (firm age, total assets, number of employees, industry dummy and gearing). The operationalization of all variables is the same as in Table 1.The natural logarithm is used for return on assets, firm age, total assets and number of employees. All variables are measured at the end of the year. Asterisks indicate significance at 0.01(***), 0.05(**), and 0.1(*) levels respectively. Robust standard errors are reported in parentheses under the coefficients.
ROA (ln) Tobin's Q
1 2 3 4 5 6
Number of females -0.120*** -20.476***
(0.047) (2.772)
Number of females*High GGI 0.125*** 14.322***
(0.046) (2.183)
Percentage of females -0.012** -2.045***
(0.005) (0.317)
Percentage of females *High GGI 0.013** 1.516***
(0.005) (0.280)
Female dummy -0.155** -35.544***
(0.072) (5.893)
Female dummy *High GGI 0.154** 26.344***
(0.069) (3.693)
Board size -0.004 -0.004 -0.004 2.486*** 1.832*** 2.029***
(0.005) (0.005) (0.005) (0.761) (0.675) (0.679) Average time in role 0.034** 0.034*** 0.034*** 5.279** 5.341** 5.299** (0.013) (0.013) (0.013) (2.116) (2.118) (2.142) Average time on other quoted
boards -0.018** -0.019** -0.019** 10.097*** 10.028*** 10.138*** (0.009) (0.009) (0.009) (2.282) (2.279) (2.306) Average age -0.003 -0.004 -0.003 -3.314*** -3.303*** -3.261*** (0.004) (0.004) (0.004) (0.859) (0.860) (0.862) Average education 0.079** 0.080** 0.079** -22.897*** -22.738*** -22.376*** (0.032) (0.031) (0.032) (4.392) (4.392) (4.360) Nationality mix 0.471*** 0.478*** 0.473*** -24.602*** -25.517*** -27.281*** (0.010) (0.100) (0.100) (9.422) (9.485) (9.592) Firm age (ln) -0.002 -0.003 -0.003 10.497*** 10.393*** 10.249*** (0.017) (0.017) (0.017) (1.586) (1.600) (1.629) Total assets (ln) -0.202*** -0.202*** -0.201*** 2.108 2.200 2.381* (0.017) (0.017) (0.017) (1.393) (1.388) (1.384) Number of employees (ln) 0.087*** 0.087*** 0.087*** 4.779*** 4.776*** 4.814*** (0.013) (0.013) (0.013) (1.541) (1.541) (1.545) Industry dummy 0.138*** 0.138*** 0.138*** 4.314 4.732 5.004 (0.034) (0.034) (0.034) (5.354) (5.372) (5.408) Gearing -0.002*** -0.002*** -0.002*** -0.222*** -0.223*** -0.220*** (0.000) (0.000) (0.000) (0.020) (0.020) (0.020) Constant 3.808*** 3.806*** 3.800*** 90.636** 94.327** 88.609** (0.242) (0.243) (0.242) (37.130) (37.203) (37.135) Number of observations 4329 4329 4329 4329 4329 4329 F-statistic 47.67*** 47.32*** 47.61*** 17.53*** 17.47*** 18.01*** Adjusted R2 0.149 0.149 0.149 0.080 0.078 0.078
One concern with OLS estimations is related to omitted variables that could lead to biased results10. In order to account for this potential bias, the OLS models are estimated with the inclusion of both
10
25 fixed and time-fixed effects11. According to the obtained results, that are shown in Table 6, the extent of female directors measured by total number of females on the board, percentage of females and female dummy seem to have no impact on firm performance measured by both return on assets (Column 1, 2, 3) and Tobin’s Q (Column 4, 5, 6). The following, suggests that the pooled OLS method may lead to biased results due to omitted variables or unobserved heterogeneity, hence the fixed effects should be chosen.
Next, as it is hypothesized that society’s attitude towards female at the workplace may have a moderating effect in the relationship between board gender diversity and firm performance, the interaction term between the extent of females on board and Gender Gap Index is added to the OLS estimations with inclusion of firm-fixed and time-fixed effects. All estimations are performed with both High interaction and Low interaction term, respectively. The results with High interaction term are reported in Table 7. As it can be seen the extent of female directors on firm performance measured by both return on assets and Tobin’s Q is still insignificant. However, according to Column 1 and 2 the coefficients of the High interaction term are positive and significant at 10 % level (β1=0.114;
β2=0.015), indicating that female directors measured by both number and percentage of females in a
more supportive environment for women have a positive effect on firm performance. Interestingly, when performance is measured by Tobin’s Q, the High interaction term still remains insignificant, indicating that in this case the culture construct does not act as a moderator in the relationship between board gender diversity and firm performance. The results obtained with the Low interaction term are provided in Table A.6 (See Appendix). All columns indicate a positive insignificant relationship between the extent of female directors and firm performance. In terms, of the Low interaction term, coefficients are negative and significant at 10 % level (β1=-0.114; β2=-0.015) only when firm
performance is measured by return on assets (ROA) and the extent of female directors is measured by the total number and percentage of females on board. This again confirms the hypothesis that in a less supportive culture for female participation and empowerment, women may detract from firm performance.
11
26
Table 6
Firm performance and the extent of female directors in European firms in the period 2008-2015.
This table shows the results of OLS estimations with inclusion of firm-fixed and time-fixed effects, using a European unbalanced sample comprising 28 countries for the period 2008 to 2015. Data regarding board-level characteristics are retrieved from BoardEx database and firm-level characteristics are retrieved from Bureau van Dijk’s Orbis database. Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (Tobin`s Q) regressed on the number of female directors, percentage of female directors and female dummy respectively. Number of females equals the number of female board directors, percentage of females equals the number of female board directors divided by the total number of board directors, and female indicator is a dummy variable that equals 1 when a firm has one or more female board directors and 0 otherwise. Control variables are divided into two sets: board characteristics (board size, average time in role, average time on other quoted boards, average age, average education, and nationality mix) and firm characteristics (firm age, total assets, number of employees and gearing). Industry dummy is omitted due to collinearity. The operationalization of all variables is the same as in Table 1. The natural logarithm is used for return on assets, firm age, total assets and number of employees. All variables are measured at the end of the year. Asterisks indicate significance at 0.01(***), 0.05(**), and 0.1(*) levels respectively. Robust standard errors12 are reported in parentheses under the coefficients.
ROA (ln) Tobin's Q 1 2 3 4 5 6 Number of females 0.025 0.774 (0.026) (1.457) Percentage of females 0.001 0.024 (0.003) (0.160) Female dummy 0.072 2.025 (0.052) (2.737) Board size 0.023 0.025* 0.023 0.199 0.268 0.209 (0.015) (0.014) (0.014) (0.915) (0.883) (0.904)
Average time in role 0.003 0.001 0.004 2.779 2.722 2.810
(0.023) (0.023) (0.023) (2.483) (2.502) (2.508) Average time on other quoted
boards 0.054** 0.054** 0.054** 1.719 1.710 1.711 (0.024) (0.024) (0.024) (1.736) (1.733) (1.739) Average age 0.026*** 0.025*** 0.026*** -0.513 -0.535 -0.516 (0.009) (0.009) (0.009) (0.415) (0.412) (0.412) Average education 0.023 0.022 0.019 -6.982 -6.983 -7.083 (0.077) (0.077) (0.078) (5.737) (5.752) (5.722) Nationality mix 0.122 0.134 0.127 7.030 7.432 7.216 (0.207) (0.207) (0.204) (16.967) (16.904) (16.820) Firm age (ln) 0.005 0.002 -0.001 -4.457 -4.544 -4.630 (0.158) (0.159) (0.159) (4.078) (4.046) (4.052) Total assets (ln) -0.178*** -0.179*** -0.180*** 5.749 5.717 5.686 (0.063) (0.063) (0.063) (4.860) (4.862) (4.854) Number of employees (ln) 0.052* 0.051* 0.051* 9.439 9.427 9.429 (0.030) (0.029) (0.029) (7.542) (7.543) (7.534) Gearing -0.002*** -0.002*** -0.002*** -0.009 -0.009 -0.009 (0.000) (0.000) (0.000) (0.007) (0.007) (0.007) Constant 2.186** 2.238** 2.219** -100.792** -98.994** -99.681** (0.974) (0.973) (0.977) (47.413) (47.260) (46.290) Number of observations 4329 4329 4329 4329 4329 4329 F-statistic 12.71*** 12.630*** 12.740*** 1.710*** 1.710*** 1.730*** Within R2 0.076 0.076 0.076 0.030 0.030 0.030 12
27
Table 7
Firm performance and the extent of female directors in European firms in the period 2008-2015.
This table shows the results of OLS estimations with inclusion of firm-fixed and time-fixed effects, and an interaction term between the extent of female directors and the Gender Gap Index (GGI), using a European unbalanced sample comprising 28 countries for the period 2008 to 2015. High interaction term is only reported. High GGI is a dummy variable that equals 1 for countries with Gender Gap Index higher than the world median and zero otherwise. Data regarding board-level characteristics are retrieved from BoardEx database and firm-level characteristics are retrieved from Bureau van Dijk’s Orbis database. Columns 1, 2 and 3 (Columns 4, 5, and 6) show results from return on assets (Tobin`s Q) regressed on the number of female directors, percentage of female directors and female dummy respectively. Number of females equals the number of female board directors, percentage of females equals the number of female board directors divided by the total number of board directors, and female indicator is a dummy variable that equals 1 when a firm has one or more female board directors and 0 otherwise. Control variables are divided into two sets: board characteristics (board size, average time in role, average time on other quoted boards, average age, average education, and nationality mix) and firm characteristics (firm age, total assets, number of employees and gearing). Industry dummy is omitted due to collinearity. The operationalization of all variables is the same as in Table 1. The natural logarithm is used for return on assets, firm age, total assets and number of employees. All variables are measured at the end of the year. Asterisks indicate significance at 0.01(***), 0.05(**), and 0.1(*) levels respectively. Robust standard errors are reported in parentheses under the coefficients.
ROA (ln) Tobin's Q
1 2 3 4 5 6
Number of females -0.080 -0.227
(0.066) (1.202)
Number of females*High GGI 0.114* 1.076
(0.064) (1.107)
Percentage of females -0.013 -0.027
(0.009) (0.147)
Percentage of females *High GGI 0.015* 0.054
(0.009) (0.119)
Female dummy -0.033 0.633
(0.092) (2.096)
Female dummy *High GGI 0.116 1.543
(0.090) (2.038)
Board size 0.023 0.025* 0.023 0.203 0.270 0.217
