The impact of board gender diversity on firm value and corporate social responsibility

Lotte Nijenhuis – 6181848 08.07.2013

The impact of board gender diversity on firm value and corporate

social responsibility

Abstract This study examines the effect of female board members on firm value, as measured by Tobin’s Q. In addition, it is investigated if this effect is stronger for firms who are more corporate social responsible. The empirical analysis, based on firms drawn from the S&P500 over a period of 5 years, shows an insignificant effect of board gender diversity on firm value. Also, women are not more valuable for companies who are highly involved in corporate social responsible. A panel data analysis was used to control for unobserved characteristics.

1. Introduction

The position of women in corporate governance became a topic of discussion in media and politics, especially after the noticeable problems during the financial crisis. As a result, studies on the salary gap between men and women as well as the effectiveness of the board of directors have increased. (Matsa and Miller 2011) Also, this debate has been accompanied by some governmental legislation changes like the in 2003 adopted Norwegian law, which stated that 40 % of the board members have to be female (Ahern and Dittmar 2012).

There are several reasons for initiating this study. To start, as can be perceived from the literature: it is a very recent topic. Carter, Simkins and Simpson (2003) indicate to be the first one covering the subject broadly. Second, the results of foregoing studies are diverging. A positive relation, no relation, as well as a negative relation has been shown between female board members and company performance. A reason for this diversity could be the amount of factors that have to be taken into account or the difference between direct and indirect impact on company performance. This makes it interesting to extend the research and add knowledge to existing literature. Hence, this study could provide additional clarity on board gender diversity.

To narrow this research down the focus will be on companies that are highly involved with corporate social responsibility (CSR). Several studies found that female board members positively influence qualitative board tasks like strategic decisions (Baysinger and Hoskisson 1990) and CSR control (Huse et al. 2009). On the contrary, research conducted in 2009 showed that the effect of women on control tasks, including CSR tasks and strategic tasks, was insignificant due to the lack of experience women have with working in a minority position of corporate boards (Boulouta 2012). It is interesting to investigate if this insignificance effect on efficiency can also be found on firm value. Additionally, they used data on Norwegian firms from one year, 2006. Extending the scope of the research to other countries with a sample on multiple years might give different results. Therefore, the following research question is derived: What is the impact of board gender diversity on firm value and is this effect stronger for companies that are more corporate social responsible?

The sample of this empirical research consists of US companies included in the S&P 500 index, with information gathered from 2007 until 2011. Huse et al. (2009) used a sample gathered from 2006. Since they state that the insignificant effect is a result of inexperience, the following five years after their sample has been used in this study. Therefore this study could contribute to the existing literature by examining if women have gained experience with working in a minority position Also, it extends research on the impact of female board members on CSR when using firm value as a dependent variable.

The remainder of this paper is structured as follows. This study begins with an overview of former studies on the differences between men and women in a business environment; to see how gender could influence company activities. This will be supplemented with literature covering board gender diversity and the impact of women on CSR. Finally, based on the literature review a model will be composed, which will be followed by the results, the discussion of the results and a conclusion.

2. Literature review 2.1 Men vs. women

The scope of the debate on female participation in the business environment and society is broad. There are several aspects related to differences in gender. For example, the role that woman traditionally fulfil in family life. In addition, there is an on-going discussion on the salary gap between men and women with comparable positions in the business (Matsa and Miller 2011). A study on this subject was done by Bertrand et al. (2010). They compared the careers of male and female MBA graduates from US business schools over a period of 10 years. The results show that although their earnings (US dollars) were equal at outset of their careers, male earnings soon deviated from females with 60 log points per decade after completing their MBA. (Bertrand et al. 2010). The research was conducted in 2010, showing that gender wage gap is a very current topic. Difference between professional positions of men and women is another issue considered when exploring gender diversity. Statistics from 2011 show that although women represent 47% of the US labour force, only 6% are corporate CEO’s or have a top executive job (Matsa and Miller 2011). Herewith a term often mentioned is the ‘glass ceiling’, containing that women experience difficulties to reach top corporate level and have to deal with a barrier, when climbing up the ladder of their professional career.

There are several explanations, coming from both the supply and demand side, for the different professional paths of women and men. Career interruption due to childbearing and differences in weekly hours associated with motherhood are factors that influence the salary gap (Bertrand et al. 2010). Also, women may shy away for a competitive environment. In a laboratory experiment where both men and women could choose their environment, without difference in performance, men selected a competitive environment twice as often as women. This gap arises

because of differences in preference. Here women avoid competition and men are overconfident, preferring competition (Niederle and Vesterlund 2007). As a consequence the pool of applicable women is limited compared to the pool of men, which makes the women that are suited for board membership a scare commodity (Farrell and Hersch 2005). On the demand side there are institutional barriers; the ‘glass ceiling’ preventing women from reaching corporate top level. This may occur for many reasons, for example men can tactically discriminate in favour of their own sex. When searching for a successor, they often prefer people with similarities to themselves. Additionally, individuals are better at interpreting signals about competence from people of the same sex. As such, sex similarities make it easier to identify with. Finally, powerful male dominated networks are less accessible for women, which are called ‘the old boys’ networks (Fitzsimmons 2012).

The discussion in politics and the media about the position of women in corporate governance has strongly increased over the last decade. This is indirectly caused by the financial crisis. Research on the effectiveness of board of directors has expanded especially after the noticeable problems during the crisis. Companies were overconfident and focused too much on short-term profit. As a result, risky management led to severe damage to the global economy (Colander et al. 2009). There is a lot of literature available on the reactions of men and women towards overconfidence and risk preferences. Multiple experiments show that women are substantially less overconfident and more risk averse than men (Croson and Gneezy 2009). According to Croson and Gneezy (2009) one reason is that women more often experience fear when a negative outcome is expected, whereas men tend to feel anger. As a consequence this anger causes men to evaluate risk, as less risky than it is in reality. Next, they also showed an experiment where subjects were asked what they thought about their investment decision. Women again were less overconfident than their male opponents. Considering the problems that caused the current financial crisis, this argument favours female representation in board composition. Furthermore, certain experiments explain that women regularly are more long-term oriented than men and more altruistic (Matsa and Miller 2013). This can lead to a different approach of tasks and problems. Therefore it can be possible that although men and women aim for the same goals, they have different abilities to achieve them.

2.2 Board of directors

Corporate governance can be described as ‘‘the system by which companies are directed and controlled’’ (Cadbury Report 1992). It comprises a mechanism to align the interests of different shareholders, executives, board of directors and other stakeholders. To achieve this goal the board of directors concentrate on specific tasks. These contain the monitoring and motivation of executives (Francoeur et al. 2008) (Adams and Ferreira 2009.), maintaining control over managerial decisions and the implementation of strategic actions. (Baysinger and Hoskisson 1990). Control is required when account is taken of the agency theory. Fama and Jensen developed this theory in 1983 (Fama

and Jensen 1983). The agency theory says that principles and agents can have different incentives. When board members control managerial decision, this can create incentives for executives to act in the best interest of the company and aim for achieving objectives set by the board of directors.

Many papers were written on the influence of board gender diversity on organizational performance, with as a result a range of diverse conclusions. Positive, negative as well as neutral relations were found to be significant. An explanation for this could be the use of different variables. Meaning that authors have taken different factors to measure the influences of female board members. Some examples of variables that are often utilized are firm value, stock prices, return on investment, return on assets or strategic decisions. Another clarification could be the difficulty to measure characteristics. Not all men or all women are comparable. Part of being a good board member depends on specific characteristics or strategic capabilities. These factors are not gender related.

Carter, Simkins and Simpson (2003) claim to be the first who present empirical evidence of a positive relation between board diversity and firm value. They measure diversity as a percentage of the minority, gender or ethnical, on the board. Their sample consisted of Fortune 1000 firms and they measured firm value by Tobin’s Q. Adams and Ferreira followed this approach and found comparable results (Farrell and Hersch 2005). Furthermore, Erhardt (2003) found evidence on the significant positive impact of gender diversity, measured as a percentage, on ROA and investment.

Various explanations for these results can be found. To start, research was done to examine if women are able to enhance decision-making. Nielson and Huse (2010) contributed to the literature by revealing a significant impact on strategic decision making. In line with that, research Boulouta (2012) show that women differ from men in making strategic decision and handling mergers and acquisition. (Matsa and Miller 2011) (Matsa and Miller 2013). Secondly, a variety of studies found that diversity enhances creativity, innovation and expands the knowledge base at hand (Erhardt et al. 2003), (Carter et al. 2003). Research by Kevin Campbell also states that; “creativity and innovation are not random distributed over the populations but vary with demographic variables like gender”. Hence, these factors can contribute to the development of competitive advantages.

One of the most common arguments explaining the positive impact of board gender diversity is the input of a different perspective (Campbell et al. 2007) (Carter et al. 2003). Women can maintain an additional view on situations through which they can be value enhancing. Also, through a diversified composition of the board a wider part of society is represented. These representatives create a high awareness of the factors that are important to customers. This will result in a better understanding of the marketplace, which increases the ability to penetrate new markets and reach potential new customers (Carter et al. 2003). This is in accordance with a study done by (Brammer et al. 2007). They state that the level of proximity to the final customers influences the formation of board gender diversity. As such, an above average prevalence of women in Utilities, Retail, Banking and Media was found. All of these sectors require a good understanding of customer needs. When referring to additional perspective, board gender diversity can additionally contribute to solving

problems. A complementary inside in the complexity of the environment and other relevant matters can consequently lead to an increase in effective leadership. Likewise, it can advance the analyses of different alternatives. On average this is more time consuming however it can eventually lead to more effective and well-considered solutions. Naturally this can provide a reduction of risk. As such, women are more valuable in complex environments. Francoeur et al. (2005) supports this argument in favour of female officers active in a complex environment. They found an excess return for companies with a higher percentage of women, using the Fama and French valuation framework. Contrary to others, their estimates reveal no effect of female directors on financial performance in complex environments (Francoeur et al. 2008).

Most existent research examining the impact of female board members suggests that women allocate more effort to monitoring. They reach a high level of attendance and cause the attendance problems for male directors to decline. Consequently, they increase transparency through a better oversight of managers. Other results, due to the extension of monitoring activities, are that CEO turnovers are more sensitive to stock performance and remuneration is more equity-based (Adams and Ferreira 2009.). This is supported by Gul et al. (2011). Their research revealed that since monitoring activities are higher for companies with female representation, stock prices reflect more firm-specific information when the board of directors contain women. Nevertheless, in spite of the progression Adams and Ferreira (2009) found on monitoring, on the overall firm performance they discovered a reduction when the fraction of female members grew. This is in conformity with former research revealing that too much monitoring can worsen shareholder value. As such, it can be possible that firm value will only increase, with a rise in female board members, when shareholder rights are low and additional monitoring can enhance firm value. Faleye and Hoitash (2011) also investigated the consequences of an improvement in monitoring quality. They conclude that intensive monitoring comes at a cost. These are weaker strategic advising, managerial short-term focus, worse acquisition performance and diminished corporate innovation (Faleye et al. 2011).

There is also an ethical aspect to the debate that resulted in several governments implementing legislation to mandate a fraction of female board members. The Norwegian government legislation, which stated that 40% of board members has to be female is an accurate example Ahern and Dittmar (2012). These quotas were carried out to ensure the advancement of women. Norway, France and Spain are early adopters of mandatory quotas, but recently other EU countries also make use of such legislation (Faleye et al. 2011). For many the introduction of the quotas caused an opportunity with respect to research on the impact of female board members. Meaning, that the implementation could be used as an exogenous variable. Since it is hard to measure unobservable characteristics when referring to the performance of the board of directors, many of the current literature is affected by endogeneity bias 1(Faleye et al. 2011). Consequently, a few years after the implementation in Norway                                                                                                                          

Ahern and Dittmar (2012) examined the effect of the quota on firm value. As a result they proposed that the mandatory quotas generated a decline in Tobin’s Q and caused a significant drop in share prices at the point of announcement. Causalities were increased leverage and acquisition, followed by deterioration in operating performances (Faleye et al. 2011). Although revenues and non-labour cost were similar, fewer people were laid off, therefore decreasing short-term profit. This was a direct effect of the female contribution (Ahern and Dittmar 2012). Matsa and Miller (2013) quoted a corporate female board members saying “When you make a decision, whatever that decision is whether it’s about an acquisition, whether it’s about anything, [being a woman] just makes you more sensitive to everyone that’s involved; their health care, their retirement, all their benefits”, referring to the differences between men and women. In addition; the quotas led to younger and less developed women. They had obtained less CEO experiences and although they were highly educated, on average they were eight years younger than the already seated male directors (Faleye et al. 2011). Differences in preferences between older male and younger new board members were according to existing literature possibly a reason why the quotas did not succeed (Ahern and Dittmar 2012). Heterogeneity can be valuable in a complex environment where an extended knowledge base is valuable, however in these cases the positive effect of a homogenous board outweighed the benefits of a different perspective.  

Accordingly, mandated quotas may lead to tokenism (Fitzsimmons 2012). When a single woman is appointed because of her gender, negative effects can be influencing her performance. For instance, being a minority makes it easier for a group of men to ignore or dismissed certain ideas. Also, esteem has been argued to be crucial for the performance of board tasks. Esteem refers to “ how individual board members are perceived and included by others (Huse et al. 2009). Newcomers with different backgrounds often have lower esteem. The group may take their voice for granted, leading to a deterioration of the input of a new perspective, skills and knowledge. (Huse et al. 2009). According to Farrell and Hersch (2005) benefits of female representation starts when there are at least three women seated on the board. Only then gender diversity can make a difference through improvement of governance. This is in accordance with the debate on homogeneity amongst board members. Although women can increase understanding of a complex environment, this will be at the cost of unanimity. (Francoeur et al. 2008) Furthermore, the process of making decisions in a situation of heterogeneity can be delayed (Erhardt et al. 2003) and eventually more difficult (Carter et al. 2003), resulting in a deterioration of flexibility. Therefore women can be valuable within a creative or innovative society, where a wider perspective and judgement can be more valuable than smooth communication and coordination.

When examining former literature the following hypothesis can be formulated: Hypothesis 1: Female representation has a positive impact on firm value.

2.3 Board gender diversity and CSR

Due to the large consequences of the financial crisis, corporations now more than ever pay attention to the ethical aspect of their business (Colander et al. 2009). Several shareholders in corporations are less confident about their investment because of the risk that was taken preliminary to the crisis. In an attempt to restore that confidence companies investigate in Corporate Social Responsibility (CSR). By definition CSR is “the continuing commitment by businesses to behave ethically and contribute to economic development while improving the quality of life of the workforce and their families as well as of the local community and society at large” (Chaud 2006). Mercer Investment Consulting supports the relevance of aiming for improvement of CSR activities in their study. They state that 46% of institutional investors consider environmental and social involvement as well as corporate governance when framing an investment decision. As a result they will pay a premium of 12-14% (Boulouta 2012).

The board of directors make decisions concerning strategic implementations, like the investment in CSR. As examined before, female representation can influence these decisions through the input of a different perspective. Referring to the literature, an important argument in favour of female board members is that they exert a positive impact on qualitative board tasks. For example, strategic decisions (Baysinger and Hoskisson 1990) and CSR control (Huse et al. 2009). As such, when women are more concerned about CSR they can affect decisions of the board and so the involvement of the company in CSR.

Former studies learn that women have a positive effect on CSR (Boulouta 2012). Multiple explanations are suggested in various articles. The first is that women are more concerned about human and social aspects (Huse et al. 2009). Second, they are more long-term orientated whereas men are generally more short-term oriented. (Matsa and Miller 2013) Other differences between men and women that could alter corporate social awareness through female perspective are trustworthiness (Croson and Gneezy 2009) and altruism (Matsa and Miller 2013). Also, research by Boulouta (2012) argued that board gender diversity significantly affects CSR. Through empathic and caring characteristics they are more concerned about negative business practices, which they can try to improve.

These suggestions lead to the following hypothesis:

Hypothesis 2: The effect of female representation is stronger for companies that focus on CSR

3. Data

The sample of firms is obtained from the S&P 500 index. The index consists of companies who account for 70% of all US publicly trade firms (Boulouta 2012). This results in a broad range of industries. To acquire the complete dataset, three individual databases will be utilized. Compustat, Riskmetrics and Kinder Lydenberg Domini (KLD)2. Merging these databases, some limitations arose concerning the years included. As a result the data as collected from 2007 until 2011 to match all variables involved. The first database is Compustat. Compustat provides financial, statistical and market information of companies among which those admitted to the S&P 500. Data on board gender diversity and board characteristics was collected from Riskmetrics. Finally, by gathering data from the KLD from the database, the level of corporate social responsibility is calculated. KLD rates companies’ social performance and generates data once a year at the end of the calendar year. As a result from combining three databases a number of firms were missing data on one or more of the 5 years in the sample. To avoid the deletion of valuable information, these companies were included in the sample nevertheless. Hence, the panel is unbalanced.

3.2 Variables

To investigate both hypotheses the dependent variable will be Tobin’s Q. This ratio is frequently used in comparable studies to measure firm value. (Carter et al. 2003) The data that was collected from Compustat supplies information to calculate Tobin’s Q. When dividing the market value of assets by the book value of assets, a simple approximation of the ratio is given. This method is less complex then some alternative measurements for Tobin’s Q and often used with comparable studies (Carter et al. 2003). Also, available data on revenue and assets are obtained from Compustat. Both log revenue and log assets are included in the model to control for firm size. Finally, to control for industry differences the 3-digit North American Industry Classification System (NAICS) code was added to each of the 500 companies.

Riskmetrics delivers data on factors related to board size and board gender diversity. In this paper female representation is calculated as the percentage of female members relative to the size of the board. The last database from with data is collected is KLD. It will provide information to calculate the level of corporate social responsibility. KLD is worldwide one of the largest institutions that measure social performance. As such they are the leading authority on providing social research for institutional investors. By focussing on different areas they try to collect an overall view on the level of corporate social responsibility. In this study six different issue areas are incorporated to obtain a CSR score. These areas are: community, corporate governance, employee relations, environment, human rights and product. Diversity is excluded as an issue area to avoid correlation with female

representation. For each area the concerns are subtracted from the strengths. This will lead to a score per issue area. The sum of the scores on all different areas provides a total net CSR score.

4. Methodology

The following methodology is formulated to measure the impact of female board members on firm value (H1). A panel data analysis was used to control for omitted or unobservable characteristics. These are factors that can not be measured for example corporate culture, management ability or quality. Omitted variables could also be factors that change over time but not across companies like national policies and federal regulation. Regarding CSR and board gender diversity the list of characteristics that might be omitted is large. These omitted variables could both influence the dependent, as well as the independent variable, which will threaten the causal relationship between them. (Boulouta 2012)(Stock and Watson 2012) Hence, when not controlling for unobserved characteristic a significant positive effect could be found when actually the causal relation comes from the variable that can’t be observed. This is quite a serious problem. Panel data analysis uses fixed effects (fe). Fixed effects control for time-invariant, company specific characteristics that are included in the companies error term. When firm specific characteristics are included in the error term this could bias the dependent variable. Correlation between variables and the error term this is called endogeneity. (Stock and Watson 2012). Taken into account certain assumptions, panel data analysis is an appropriate treatment to account for endogeneity (Boulouta 2012).

A second econometric model will provide insights on the effect of changes in the percentage of female board members on the changes in Tobin’s Q. Account has to be taken when measuring causality of the time that is needed to see results after implementing new management. Therefore, to estimate causality the following econometric model is used:

∆!!!" =   ∆!"!!"!!+   ∆!!!"+  !"#$!!"+   !!"      ! = 2007, … , 2011. where i indexes firms and t indexes time periods. The change in the percentage of female board members is measured one year earlier than the other variables. This way the time it takes to see results after a change are included in the model.

To see if women are more valuable for companies with a high degree of CSR (H2) again panel data analyses is utilized to control for unobserved characteristics. In addition to the former model, a dummy variable for a high CSR score is included. The distinction between high and low scores is made using the median. Table 1 shows that the median CSR score is 0. Also, an interaction term of the percentage of board gender diversity and the CSR score of a firm is incorporated into the model:

where i indexes firms and t indexes time periods. When using fixed effects, there is no need to add industry as a control variable to the mode. Since panel data analyses controls for firm specific characteristic, industry is automaticity included. The fifth and final model is comparable to model 2. It measures the impact of a change in board gender diversity on a change in Tobin’s Q, while including CSR. All regressions will include a robust option for estimating the standard errors. The coefficients, when using a robust regression method, are exactly the same as in ordinary OLS. Only the standard error estimates will be adjusted for some flaws in the data. It is often hard to meet all assumptions underlying multiple regression. Examples of problems for which the robust method adjusts are minor problems about normality and heteroscedasticity (Stock and Watson 2012).

In this study, the dependent variable is Tobin’s Q, as mentioned before. The independent variables are as follows:

!"!

!!!! ! !!!!!"#$"%&'("!!"!!"#$%"!!"#$%!!"!#"$%!

!!

!!!!!!! ! !"#$%!!"#$!

!"#$!

! ! !"#$%&'!!!!"!!"!#$!!""#$"!

!"!

!!!! ! !"#$%&'(!

!"!

!!!! ! !"#$"#%&'!!"#$%&!!"#$%&#'(')'*+!!"#$!

Descriptive statistics for all variables included in the model are displayed in table 1. Table 2 shows the correlation coefficients.

Table 1

Descriptive statistics for sample

Variable Obs Mean Std. Dev Min Max

Tobin's Q 1692 1.82084 1.083795 .6296661 10.14768

PRF 1692 .1513023 .090155 0 .5

BS 1692 10.74409 2.509378 5 34

logAT 1692 95710.95 13218.55 64264.02 146334.3

CSR 1692 -.2411348 3.192393 -11 12

Table 1 shows that the mean of the percentage of female board members is 15.13%. When comparing this number to the study of Carter at el. (2003), the female presence has increased by 5.53%. The comparison with this particular study is made, because they claim to be to first who show a significant and positive relation between women on the board and Tobin’s Q. While the percentage of board gender diversity has increased, the board size has stayed the same. The study by Carter et al.

(2003) has a mean board size of 10.98 and table 1 shows a mean board size of 10.74. Therefore it is fair to say that the amount of women on corporate boards has increase in the last decade.

Table 2

Correlation coefficients

Cusip year TQ CSR BS PRF logAT logREV

Cusip 1.0000 year 0.0132 1.0000 TQ -0.0376 -0.0541 1.0000 CSR -0.0115 0.3551 0.1134 1.0000 BS 0.0627 0.0359 -0.3014 0.0446 1.0000 PRF 0.0490 0.0574 -0.0920 0.1595 0.1947 1.0000 logAT -0.0532 0.1000 0.4759 -0.0094 0.4495 0.2216 1.0000 logREV -0.0622 0.0365 -0.2484 -0.0360 0.3281 0.2955 0.7174 1.0000

Since total assets (logat) and revenue (logrevt) are correlated by a factor of 0.7174, total revenue will be excluded from the model. Total assets will therefore be the variable used to control for firm size. This variable is preferred before revenue since its correlation with the percentage of female board members is lower.  

5. Results

5.1 Summary statistics

Graph 1 shows the average percentage of board gender diversity within different industries. Account has to be taken on the number of observations per industry. Some industries have a significant lower amount of observations. From graph 1 can be concluded that the industries ‘Professional, scientific and technical services’ and ‘Wholesale trade’ contain the highest percentage of board gender diversity. These results are in line this the theory. Former research found that there are more women on the board of directors at companies that require a good understanding of the marketplace and focus on services.

Table 3 presents the differences between the mean values of board gender diversity over different years. As derived from the table, the mean is increasing per year. Starting from 9,28% in 2007, rising to 15.81% in 2011. An argument to explain these results could be the attention that has been given to board gender diversity in politics and the media. A growth in the mean of CSR is shown in table 4. It seems that corporations are reacting to the preferences of stakeholders, who are getting more and more concerned about the impact of corporate operations on the environment and the companies social involvement. One example is that investors increasingly rate investment opportunities according to the firms CSR activities when considering investment decisions (Boulouta 2012). In addition, the number of consumers that are purchasing eco-friendly products is growing (Chen and Delmas 2010).

Table 3

Year

Obs

Mean

Std. Dev.

Min

Max

2007

243

.1395948

.0927967

0

.4545455

2008

328

.1509027

.0897567

0

.4166667

2009

380

.1504922

.0875339

0

.4615385

2010

383

.1544773

.0909316

0

.5

2011

362

.1581509

.0904212

0

.5

n

Percentage of female board members per industry

Agriculture (n=2) Mining, Oil/Gas Extraction (n=8) Utilities (n=3) Construction (n=5)

Manufacturing (n=89) Wholesale Trade (n=6)

Retail Trade (n=14) Transportation, Warehousing (n=7) Information (n=8) Finance and Insurance (n=13) Real Estate (n=8) Professional, Scient., Tech. Serv. (n=5) Administrative, Support (n=4) Health Care, Social Assistance (n=3) Accommodation, Food Service. (n=4)

Table 4

Year

Obs

Mean

Std. Dev.

Min

Max

2007

243

-1.213.992

2.730.961

-10

10

2008

328

-1.381.098

2.717.799

-11

9

2009

380

-1.418.421

2.745.335

-11

9

2010

383

.7206266

3.153.618

-7

10

2011

362

1.654.696

3.125.198

-4

12

5.2 Results from the analyses

The results of the panel data analysis of model 1 are presented in table 5. These results show that when controlling for firm specific characters, the effect of board gender diversity on firm value is not significant. The coefficient shows a small negative effect, but is insignificant (b = -0.096, p > 0.10). Model 2 measures the effect of the change in the percentage of female board member at t-1on Tobin’s Q. Although the coefficient on PRF is positive it is again not significant (b = 0.286, p > 0.10). As such, hypothesis 1 is not confirmed. Additionally, a regression of PRF against TQ was run. The reason for adding this regression is to measure simultaneous causality. When using panel data analysis the coefficient of TQ is not significant. Still these results thus not mean that there is no simultaneous causality. It could be that the panel data analysis does not capture all unobserved variables. Furthermore, consistent with the literature is the negative effect of board size on Tobin’s Q. Several studies have found that an increase in board size could deteriorate flexibility in the process of making decisions (Erhardt et al. 2003), (Carter et al. 2003). Also, it could be more time consuming and costly to achieve unanimity (Francoeur et al. 2008).

Model 3 tests the impact of CSR and board gender diversity, both as independent variables, on firm value. With this model the follow could have been argued. If the coefficient on board gender diversity in model 1 would have been significant, than for hypothesis 2 to hold this effect would become less. The effect of female representation would have been absorbed by the CSR variable. Since the effect of board gender diversity is not significant this argument does not hold. Furthermore, from table 5 can be seen that the coefficient of the CSR dummy that indicates if firms are highly involved in corporate social responsibility is positive. The effect on firm value is significant with 5% (see table 5). Additionally, board gender diversity again has an insignificant relation to firm value.

To test hypothesis 2 an interaction variable is added to the model. The results of this analysis can be seen in table 5. The coefficient is 0.120 and insignificant (p > 0.10). Interesting to see is that the CSR dummy became insignificant, were it had a significantly positive effect on firm value in model 3. Correlation between the interactive term and the CSR dummy variable is the cause of this change.

Model 5 is an analysis of model 2 with as an additional variable the change in CSR at t-1. Again, would the effect of board gender diversity been significant, then adding CSR could measure the relationship between the two. If adding CSR had made board gender diversity less significant, than that would be a sign of correlation. Since this relationship is not found, hypothesis 2 is not confirmed. Finally, some additional tests were executed to measure the effect of board gender diversity on CSR. The coefficient was found positive, but not significant (b = 0.8370, p > 0.10). In line with the literature is the positive effect found of CSR on board gender diversity. As diversity is often seen as part of corporate social responsible behaviour, this result was expected.

Table 5

Panel data analyses

(1) (3) (4) Variables TQ TQ TQ PRF -0.0960 -0.107 -0.174 (0.337) (0.337) (0.397) BS -0.0265* -0.0255* -0.0252* (0.0148) (0.0148) (0.0148)

logAT -5.55e-05*** -5.86e-05*** -5.86e-05***

(7.14e-06) (7.23e-06) (7.23e-06)

CSRdum 0.0921** 0.0726 (0.0358) (0.0714) PRF*CSRdum 0.120 (0.382) Constant 7.428*** 7.670*** 7.679*** (0.684) (0.689) (0.690) Observations 1,692 1,692 1,692 R-squared 0.051 0.056 0.056 Number of Cusip 425 425 425 N 1692 1692 1692 df_m 427 428 429 F 22.65 18.71 14.98 rss 307.5 305.9 305.9

Standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Table 6 (2) (5) Variables Δ TQ Δ TQ Δ PRF t-1 0.286 0.275 (0.265) (0.260) Δ BS -0.0262** -0.0262** (0.0114) (0.0113)

logAT -5.70e-06*** -5.28e-06***

(2.02e-06) (2.03e-06) Δ CSR t-1 -0.0245*** (0.00735) Constant 0.584*** 0.562*** (0.211) (0.211) Observations 824 824 R-squared 0.020 0.029 N 824 824 df_m 3 4 F 3.938 5.477 rss 237.4 235.3

Robust standard errors in parentheses

*** p<0.01, ** p<0.05, * p<0.1

Table 6

6. Discussion

In the light of these results, neither a positive nor a negative effect of board gender diversity on firm value was found. Hence, the gender of corporate board members does not have a direct effect on firm value. This doesn’t mean that women can’t be value enhancing. As was shown from former literature, women differ from men on many aspects. There might be a possibility that female characteristics affect companies on other levels. Furthermore, since several studies show that gender mixed teams out perform homogenous teams a different model could have different results. (Hoogendoorn et al. 2013)

A panel data analysis was used to control for unobserved firm specific characteristics. Although fixed effects analysis can by valuable to avoid endogeneity, certain assumptions have to hold. When these assumptions are invalidated, the panel data analysis is not appropriate. Also, a strong argument why the model, after controlling for fixed effects, has endogeneity are the reasons to assume that simultaneous causality might also be present. For example, a better performing firm could sooner add female members to their board then the ones that are performing less. They might include female representation as a signal of social responsibility to their stakeholders. In the case of simultaneous causality, panel data analysis is not the right econometric analysis to use. Under these circumstances, with a type of endogeneity that is called ‘simultaneous bias”, instrumental variables is the appropriate treatment.

Furthermore, although KLD has been found to be the most widely used database for estimating the level of corporate social responsibility (Boulouta 2012), there are some critical arguments about their ability to accurately rate CSR. Currently there is not a universal agreed upon definition of CSR. This causes for some insecurities about the preciseness of the data. In addition, including more different companies from other countries could extend the sample. This will provide a better representation of population. Since the S&P 500 only contains U.S. corporations there could be a limitation to the sample. Finally, including more years in another method to acquire a bigger sample.

As mentioned in the literature review, the results of various studies on board gender diversity are diverse. Maybe it is therefore necessary to change the method of measuring the impact of women relative to men. Future research should look beyond the standard path that is currently used to measure corporate governance. Although it might be more time consuming, perhaps research on the subject will produce different results when focussing on the examination of actual board behaviour.

7. Conclusion

From two angles the effect of female board members on film value has been presented in this article: one where a CSR score was included and one without the CSR variable. As such, two hypotheses were tested. The first measures the effect of board gender diversity on firm value, while using a panel data analysis to control for unobservable characteristics. The second tests if female presence on corporate boards is more valuable for companies that are more corporate social responsible. The findings reveal that for both hypotheses an insignificant relation is found. Additionally, a significant effect is found of CSR on firm value, when excluding the interaction variable of board gender diversity and CSR. When interpreting these results the possibility of endogeneity and simultaneous causality have to be taken into account. There are reasons to believe that both are included in the model and as such the results could be biased. This study contributes to the literature by extending the research on board gender diversity. In addition, the effect of female presence on CSR is investigated which is not often measured in a similar manner. The results show that there is no difference in including a male or a female in the board of directors.

When comparing these results to existing literature similarities are found. As mentioned before, former studies on board gender diversity show diverging results. Some say that women can be more valuable in complex environments. Through contributing a different perspective a better understanding on the marketplace is obtained, which can increase the ability to penetrate new markets and reach potential new customers. On the contrary, homogenous boards are found to be more flexible and better at making quick decisions. Since findings on this subject have been inconclusive another approach to measure board gender diversity is recommended. There is a need to gain a better understanding on board behaviour under different circumstances. Therefore studies should go beyond

measuring demographics and try to explore board working style or other comparable characteristics on which additional qualitative data is needed.

Bibliography

Adams, R., Ferreira, D., 2009. “Women in the Boardroom and Their Impact on Governance and Performance.” Journal of Financial Economics, 94 (2): 291–309.

Ahern, K., Dittmar, A., 2012. “The changing of the Boards: The impact on Firm Valuation of Mandated Female Board Representation.” The Quarterly Journal of Economics (2012) 127(1): 137-197.

Baysinger, B., Hoskisson, R., 1990. “The Composition of Boards of Directors and Strategic Control: Effects on Corporate Strategy”. Academy of Management Review 1990, 15:7, 72-87

Bertrand, M., Goldin, C., Katz, L., 2010. “Dynamics of the Gender Gap for Young Professionals in the Financial and Corporate Sectors”. American Economic Journal: Applied Economics (2010) 2:228-255

Bertrand, M., Hallock K., 2001. “The Gender Gap in Top Corporate Jobs.” Industrial and Labor Relations Review, 55 (1): 3–21.

Boulouta, I., 2012. “Hidden Connections: The Link Between Board Gender Diversity and Corporate Social Performance”. Journal of Business Ethics 2012

Brammer, S., Millington, A., Pavelin, S., 2007. “Gender and Ethnic Diversity Among UK Corporate Boards”. Corporate Governance. 2007, 15:2

Cadbury Report, 1992. Report of the Committee on the Financial Aspects of Corporate Governance (Gee, London).

Campbell, K,. Minguez-Vera, A., 2007. “Gender Diversity in the Boardroom and Firm Financial Performance”. Journal of Business Ethics (2008) 83:435–451

Carter, D., Simkins, B., Simpson, W., 2003. “Corporate governance, board diversity, and firm value.” Financial Review 38, 33–53.

Chaud, M., 2006 “The Relationship between Corporate Social Performance and Corporate Financial Performance: Industry Type as a Boundary Condition”. The Business Review, Cambridge; Sep 2006; 5

Chen, C., Delmas M,. 2010. “Measuring Corporate Social Performance: An Efficiency Perspective”. Production and Operations Management: Forthcoming

Colander, D., Goldberg, M., Haas, A., Juselius, K., Kirman, A., Lux, T., Sloth, B., 2009. “The financial crisis and the systemic failure of the economics profession”. Critical Review: A Journal of Politics and Society, Volume 21, Issue 2-3

Croson, R., Gneezy, U., 2009. “Gender differences in Preferences”. Journal of Economic Literature

2009, 47:2, 448-474

Erhardt, N., Werbel, J., Shrader C., 2003. “Board of Director Diversity and Firm Financial Performance”. Corporate Governance 2003, 11:2

Financial Economics 101 (2011) 160–181

Fama, E., Jensen, M., 1983. “Agency problems and residual claims”. Journal of Law and Economics 1983, 26:2

Farrell, K., Hersch, P., 2005. “Additions to corporate boards: the effect of Gender”. Journal of Corporate Finance 11, 85–106.

Fitzsimmons, S., 2012. “Women on boards of directors: Why skirts in seats aren’t enough”. Kelley

School of Business: Indiana University.

Francoeur, C., Labelle, R., Sinclair-Desgagne, B., 2008. “Gender Diversity in Corporate Governance and Top Management”.Journal of Business Ethics (2008) 81:83–95

Gul, A., Srinidhi, B., Anthony C., 2011.”Does board gender diversity improve the informativeness of stock prices?”. Journal of Accounting and Economics 51 (2011) 314–338

Hoogendoorn, S., Oosterbeek, H., Praag, M., 2013. “The Impact of Gender Diversity on the Performance of Business Teams: Evidence from a Field Experiment” Management Science: Articles in Advance, pp. 1–15

Huse, M., Nielsen, S., Hagen, I., 2009. “Women and Employee-Elected Board Members, and Their Contributions to Board Control Tasks”. Journal of Business Ethics (2009) 89:581-597

Matsa, D., Miller A., 2013 .“A Female Style in Corporate Leadership?

Evidence from Quotas”. American Economic Journal: Applied Economics, Forthcoming

Matsa, D., Miller A., 2011. “Chipping away at the Glass Ceiling: Gender Spillover in Corporate Leadership”. American Economic Review: Papers and Proceedings 2011, 101:3, 635–639

Niederle, M., Vesterlund, L., 2007. “ Do women shy away from competition? Do men compete too much?”. The Quarterly Journal of Economics, August 2007

Nielsen, S., Huse, M., 2010. “The contribution of women on boards of directors: Going beyond the surface”. Corporate Governance: An International Review, 18(2), 136–148.

Stock, J., Watson, M., 2012. “Introduction to Econometrics”. Pearson Education Limited.

Vilanova, M., Lozano, J., Arenas, D., 2009. “Exploring the Nature of the Relationship Between CSR and Competitiveness”. Journal of Business Ethics (2009) 87:57–69

Appendix 1

Calculate TQ: drop _all

use "/Volumes/USB DISK/Compustat database stata.dta" gen MVE=prcc_f*csho

gen TQ=(at+MVE-(ceq+txdb))/at

label variable MVE "Market value equity" label variable TQ "Tobin's Q"

Merger databases: sort cusip fyear

(*de volgende regel zorgt ervoor dat de cusip code overeenkomt tussen beide databases) replace cusip = substr(cusip, 1, length(cusip) -1)

(*de regel past cusip aan, zodat beide databases samengevoegd kunnen worden) joinby cusip fyear using "/Volumes/USB DISK/KLD database stata final.dta", unmatched(master)_merge(_merge1)

joinby cusip fyear using "/Volumes/USB DISK/riskmetrics database.dta", unmatched(master)_merge(_merge2)

save "/Volumes/USB DISK/masterdatabase.dta", replace drop if TQ>=.

drop if _merge1==1 drop if _merge2==1

Appendix 2 (Hypothesis 1) (1) (2) (3) (4) (5) VARIABLES TQ TQ TQ PRF !TQ PRF 0.0879 0.0879 -0.0960 (0.255) (0.263) (0.337) BS -0.0524*** 24*** -0.05 -0.0265* -0.000842 (0.00918) (0.0103) (0.0148) (0.00123)

logAT -3.28e-05*** -3.28e-05*** -5.55e-05*** 2.07e-06*** -5.70e-06*** (2.49e-06) (2.10e-06) (7.14e-06) (6.07e-07) (2.02e-06)

11b.Ind 0 0 (0) (0) 21.Ind -0.794*** -0.794** (0.269) (0.318) 22.Ind -0.864*** -0.864*** (0.267) (0.315) 23.Ind -0.780** -0.780** (0.303) (0.350) 31.Ind -0.236 -0.236 (0.267) (0.307) 42.Ind -0.453 -0.453 (0.279) (0.351) 44.Ind -0.284 -0.284 (0.279) (0.318) 48.Ind -0.201 -0.201 (0.321) (0.331) 51.Ind -0.104 -0.104 (0.281) (0.315) 52.Ind -0.308 -0.308 (0.275) (0.313) 53.Ind -0.675** -0.675** (0.300) (0.326) 54.Ind 0.0796 0.0796 (0.333) (0.344) 56.Ind 0.0970 0.0970 (0.353) (0.362) 61.Ind -1.373*** -1.373*** (0.284) (0.510) 62.Ind -0.810*** -0.810** (0.268) (0.377) 71.Ind -0.412 -0.412 (0.440) (0.510) 72.Ind 0.344 0.344 (0.340) (0.346) TQ -0.000667 (0.00235) !PRF t-1 0.286 (0.265) !BS -0.0262** (0.0114) Constant 5.849*** 5.849*** 7.428*** -0.0365 0.584*** (0.355) (0.353) (0.684) (0.0596) (0.211) Observations 1,692 1,692 1,692 1,692 824 R-squared 0.297 0.297 0.051 0.010 0.020 Number of Cusip 425 425

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Description of Appendix 2 (Hypothesis 1)

1. OLS Regression with robust standard errors 2. OLS Regression without robust standard errors 3. Panel data analysis

4. Panel data analysis with PRF as dependent variable (measuring simultaneous causality) 5. Analysis with differences per year

Appendix 2 (Hypothesis 2(1)) (6) (7) (8) (9) VARIABLES TQ TQ TQ TQ PRF 0.0267 -0.373 -0.107 -0.174 (0.252) (0.366) (0.337) (0.397) BS -0.0539*** -0.0532*** -0.0255* -0.0252* (0.00914) (0.00921) (0.0148) (0.0148) logAT -3.25e-05*** -3.27e-05*** -5.86e-05*** -5.86e-05***

(2.49e-06) (2.51e-06) (7.23e-06) (7.23e-06)

11b.Ind 0 0 (0) (0) 21.Ind -0.834*** -0.844*** (0.275) (0.276) 22.Ind -0.907*** -0.911*** (0.272) (0.274) 23.Ind -0.824*** -0.831*** (0.307) (0.309) 31.Ind -0.317 -0.326 (0.272) (0.274) 42.Ind -0.478* -0.470* (0.284) (0.285) 44.Ind -0.312 -0.313 (0.284) (0.285) 48.Ind -0.271 -0.271 (0.323) (0.325) 51.Ind -0.168 -0.173 (0.286) (0.287) 52.Ind -0.395 -0.397 (0.281) (0.283) 53.Ind -0.728** -0.727** (0.306) (0.307) 54.Ind -0.00691 -0.00406 (0.331) (0.332) 56.Ind 0.0743 0.0537 (0.358) (0.360) 61.Ind -1.525*** -1.540*** (0.291) (0.295) 62.Ind -0.860*** -0.856*** (0.273) (0.276) 71.Ind -0.565 -0.591 (0.446) (0.441) 72.Ind 0.283 0.275 (0.342) (0.344) CSRdummy 0.176*** 0.0636 0.0921** 0.0726 (0.0449) (0.0978) (0.0358) (0.0714) PRF*CSRdummy 0.748 0.120 (0.513) (0.382) Constant 5.826*** 5.900*** 7.670*** 7.679*** (0.359) (0.368) (0.689) (0.690) Observations 1,692 1,692 1,692 1,692 R-squared 0.303 0.304 0.056 0.056 Number of Cusip 425 425

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

Description of Appendix 2 (Hypothesis 2(1))

6. OLS Regression with robust standard errors with dummy variable for CSR

7. OLS Regression with robust standard errors with dummy variable for CSR and interaction variable

8. Panel data analysis with dummy variable for CSR

9. Panel data analysis errors with dummy variable for CSR and interaction variable

Appendix 2 (Hypothesis 2(2))

Description of Appendix 2 (Hypothesis 2(2))

10. Analysis with differences per year with change in CSR at t-1 11. Analysis (logit) using fixed effects

12. OLS Regression with robust standard errors 13. Panel data analysis

(10) (11) (12) (13)

EQUATION VARIABLES !TQ CSRdummy PRF PRF

SINGLE !PRF t-1 0.275

(0.260)

!BS -0.0262**

(0.0113)

logAT -5.28e-06*** 1.15e-06*** 2.06e-06*** (2.03e-06) (1.84e-07) (6.01e-07) !CSR t-1 -0.0245*** (0.00735) (0.211) CSRdummy 0.0125*** 0.00139 (0.00422) (0.00300) Constant 0.562*** -0.00919 -0.0378 CSRdummy PRF 0.807 (1.924) BS -0.0363 0.00412*** -0.000810 (0.0802) (0.00138) (0.00123) logAT 0.000262*** 1.15e-06*** 2.06e-06***

(4.57e-05) (1.84e-07) (6.01e-07)

TQ 0.466** (0.196) (0.0154) (0.0575) Observations 824 892 1,692 1,692 R-squared 0.029 0.065 0.010 Number of Cusip 207 425

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1