Размещено на http://www.allbest.ru/

NATIONAL RESEARCH UNIVERSITY HIGHER SCHOOL OF ECONOMICS

DEPARTMENT OF ECONOMICS

Bachelor's thesis

Gender (and other) diversity in corporate boards and firm performance

Ten Anastasiya, BEC131 group

Scientific Supervisor:

Anastasia Stepanova



• Contents
• Abstract
• Introduction
• 1. Literature review and hypotheses developed
• 1.1 Board characteristics and firm performance
• 1.2 Gender diversity on board and financial performance
• 1.3 Industry matters
• 1.4 Are female directors more risk-averse
• 1.5 Female directors and independence
• 2. Methodology and data
• 2.1 Methodology
• 2.2 Data
• 3. Results
• 4. Results interpretation and discussion
• Conclusion
• References
• Appendix
• 
• Abstract
• The study examines the relationship between gender diversity on boards and firm performance. Providing complex analysis, we identify the main firm characteristics, to which gender diversity on board has the greatest impact, and test them with regression analysis. Thus, we highlight the following aspects: firm risk-taking and board independence. The initial sample includes US-listed firms from Russell 3000 from 2010 to 2015. It was stated that firms with women on boards tend to decrease risk-level. Moreover, female directors improve the independence of the board, acting similarly to independent directors. Finally, we investigate that women on boards could differently affect firm performance. The results are robust to endogeneity issues and provide evidence that the relation between the presence of women on boards and firm's characteristics matters. We contribute to the existing literature providing additional evidence about the role of women on corporate boards.



Introduction

The reasons for debate about gender diversity can be viewed in two ways: gender inequality issue and female contribution in firms. Through the last century, larger half of leadership positions were represented by the male. Over the past decade, women have significantly increased their share in the number of leadership positions. In the US women held 14.8% of Fortune 500 board seats in 2007 (Catalyst, 2007). In Europe, women represent 13.9% in 2010 (European Women on Boards, 2016). For 2016, the percentage of female directors in the United States and Europe is estimated to be 19.9% and 25% respectively (Catalyst, 2017). These changes are caused by the fact that the gender diversity issue is discussed not only at the firm level but also at the state level. Many of European countries (among them Norway, Belgium and Italy) have passed legislation mandating more female board representation for certain firms. The sharp increase of the proportion of female directors in Europe based on the state pressure. For example, in Norway there was an extreme condition, according to the law, all listed firms must increase the proportion of female directors to 40%. In turn, Spain has followed this practice by enacting a law requiring companies to increase the share of female directors to 40% by 2015. It is hard to make conclusions about these practices as it needs to take more time, however, several researches report a positive result both on the financial performance and investing attractiveness (Campbell and Minguez-Vera, 2007; Boubaker et Al., 2014). According to the US data, there is no strong evidence about how female representation on board affects firm's financial outcomes and effectiveness of the board. One part of studies suggests a positive relation between higher gender diversity and firm performance, another part examines a negative relationship or absence of the effect at all.

The purpose of this study is to investigate the type of relationship between gender diversity on corporate boards and firm performance. According to the previous studies, we highlight the following issues:

1. The relationship between gender diversity on board and firm performance;

2. Women - friendly and non-women - friendly economic sectors;

3. Firm risk-taking and gender diversity on board;

4. Independence of board and gender diversity.

These issues represent steps of analysis, which help us to get an understanding of changes in the proportion of female directors on boards. This study also analyzes causal relationship between the level of risk and changes on boards. The next question should be answered: should female directors be more risk-averse and decrease firm's level of risk or female directors choose low-risky firms?

In general, demographic diversity on board becomes an attractive property for investors (by demographic diversity we understand age, tenure group, ethnic group, race, gender etc.) Such a property meet the challenge of society and has a significant influence on firm's corporate governance. Due to several works from the field of psychology, it is claimed that women are more financially risk - averse than men (Charness and Gneezy, 2012; Jianakoplos and Bernasek, 1998). Furthermore, we can suppose the greater the proportion of female on the board is, the lower is the leverage pressure. Another block of studies suggests that women on boards perform as independent directors, affecting such board inputs as CEO turnover and attendance of board's meetings (Adams and Ferreira, 2008, Terjesen et Al., 2016). Thus, women influence the management of the firm, for example, one of the studies suggests that CEO turnover is more sensitive to stock return performance in firms with relatively more women on boards (Weisbach, 1988).

It is difficult to measure the influence of gender diversity on firm's financial outcomes as there is no universal approach solving different problems that arise in the analysis process. Board characteristics are not exogenous and can be influenced by observable and unobservable factors, all these leads to endogeneity problem that misinterprets results. To deal with the endogeneity we use the Hausman's test and dynamic models. Another problem was discussed earlier in this paragraph - causality. This problem is time dependent, so we estimate different specifications of models, including instrumental variables to deal with reverse causality.

The structure of this study is as follows. In the next paragraph, we discuss fundamental theories and recent empirical researches. The third and fourth paragraphs are devoted to data description and methodology. In the fifth Section, we present results of estimating models and finally, in the sixth Section, we conclude all results and give an empirical interpretation, concluding the research.

Such characteristics as age, race or group tenure are included in the definition «demographic diversity» and gender is also part of it. First studies were devoted to gender diversity among working groups, in other words among ordinary workers. With the lapse of time women began to claim leadership positions. Social and psychological researchers began to explore differences in behavior by gender and they were found. For instance, women are more risk-averse than male and ready to maintain more complex and difficult issues to solve (Graham et Al., 2002; Sarin and Wieland, 2016; Lam, 2015). In the frame of corporate governance, gender diversity includes both external influences such as monitoring by shareholders and internal mechanisms, such as risk-taking or making financial decisions. Internal governance mechanism believed to have a particularly large impact on financial outcomes of a firm.

A considerable amount of valuable work on gender diversification of boards in the field of social and economic sciences has been done to accomplish the revision of the role of women on leadership positions. In the last decade gender diversity has attracted special attention among top-management and board of directors. Generally, the following components are studied by researches:

1. Corporate governance (decision making process, monitoring, compensations);

2. Dynamics of stock price;

3. Firm risk taking;

4. Firm performance (organizational or financial).

For a long time, the research on gender diversity on leadership positions was concerned with ethic and cultural issues. Recent investigations put forward new definitions, studying the effect by gender diversity on financial outcomes. It is impossible to overestimate F. Fama's agency theory which sheds a new light on the nature of relationships between shareholders and management. Managers are risk averse due to concerns about their own undiversified human capital (Fama and Jensen, 1983). Thus, managers can influence financial results of the firm negatively. To avoid this situation, it is necessary to maintain stronger monitoring function and generally effectively make decisions.

The most important works have been made by Eugene F. Fama, Vathunyoo Sila, Renee B. Adams, Siri Terjesen. We contribute to the literature by introducing the complex analysis of gender diversity. We highlight the main directions which are intensively influenced by female directors, then we will test it on our sample. Thus, we present the literature review devoted to the relationship between gender diversity on boards and firms' performance.

More generally, our research contributes to the existing literature, providing a complex analysis about gender diversified board, which has been studied primarily by researchers in organization theory and increasingly in economics and corporate finance. Empirical studies in this tradition have looked at the effects of gender diversified boards on firm risk-taking, board independence and firm performance separately. We will highlight these issues altogether, analyzing observable property (gender) and unobservable (independence). We put the following research question: does firm performance vary with gender diversified boards?

In the following parts, we review the literature devoted to the studies, that examine females' role on board in decision-making process, monitoring functioning. Then, we will examine the evidence that gender diversification of boards affects firm performance. And the last two parts of literature review will highlight risk behavior of female directors and evidence, suggesting the similarity between female and independent directors.



1. Literature review and hypotheses developed

1.1 Board characteristics and firm performance

Board structure influence manager behavior by such function as tough or weak monitoring and changes in compensation. These, in turn, result in high or low CEO turnover, their effectiveness and, hence financial sustainability of a firm. Research on the relationship between the board of directors with the proportion of female directors answers the following questions: gender director female

· Is there an effect by gender diversity on boards on attendance behavior?

· Is there an effect by gender diversity on boards on CEO compensation and CEO turnover?

· If gender diversity on board influence corporate governance, is there an effect on financial outcomes?

Agency theory discussed issues about reducing agency costs that result from the separation of ownership and control (Fama and Jensen, 1983). To decrease agency costs, board controls management on behalf of shareholders, so the following properties should be taken into consideration: firstly, the board size matters as it affects performance of the firm and the larger the board is, the greater are operational costs, decisions are more difficult to make. Large board size negatively affects firm value, especially the effect is stronger in small firms (Nguyen et Al., 2015; Cheng, 2008). Secondly, the character of directors is an important point. Recent studies suggest that independent directors are better monitors (Weisbach, 1988; Guo and Masulis, 2015) and disproportionately high number of inside directors has a negative effect on firm performance (Jermias and Gani, 2014). Concerning the issue of gender diversity, it may have a positive effect on the decision - making process, as in general, a higher level of demographic diversity help to resolve conflicts through easier way and find consensus more cooperatively (Pelled, 1996).

1.2 Gender diversity on board and financial performance

Most of existing literature suggests a significant relationship between gender diversity on board and firm financial performance. Proxies for checking financial outcomes are often represented by Tobin's Q and ROA as these variables represent market performance and operating efficiency respectively. Multi-country studies represent a positive association between the fraction of female directors on board and firm performance (Erhardt et Al, 2003; Campbell and Minguez-Vera, 2007; Terjesen et Al., 2016). Most of studies face the problem of omitted factors that could bias results, to deal with this problem, most of authors use econometrical instruments.

Adams and Ferreira (2009) analyzed gender diversified boards and governance in US publicly listed firms and found that the coefficient on diversity is statistically significant and positive. They further found that gender diversity does not add value on average, but the contribution of gender diversified boards is more valuable in weak governance firms. To avoid the problem of omitted factors, Adams and Ferreira found an instrument that is correlated with the proportion of female directors, but uncorrelated with firm performance. The variable was a sum of 24 indicators variables and represented by itself the measure of governance (takeover level). Adoption of this variable represented an increase in firm value with otherwise weak governance.

However, the recent study by Sila et Al. (2016) found no significant relationship between gender diversity on boards and firm financial performance, analyzing the US publicly listed firms. The authors investigate the association through risk-taking, applying time-series regression models. Another evidence is highlighted by Galbreath (2011), the author studied the influence of gender diverse boards on corporate sustainability and firm performance in Australia. The results suggested a positive link between an economic growth of the firm and social responsiveness.

Another field of studies is devoted to involuntary gender diversification. European firms are the best representatives of it. In 2003, Norway introduced the law, according to which all listed firms must increase the proportion of women on boards to 40%. At that moment, the fraction of women on board was in average 9% in Europe. This experience touched most European countries. Figure 1 represents the changes on boards since the introduction of law and there is an obvious significant increase for most European countries. The effect of an exogenous change to corporate boards is studied by Ahern and Dittmar (2012). The study was based on Norwegian public listed firms in the period from 2001 to 2009, also the study provides placebo tests for Denmark, Finland, Sweden and US firms to compare the effects. The results showed a negative link between firms under gender quota and Tobin's Q. As Scandinavian countries Sweden and Denmark has a similar law system, however, placebo test showed that the instrumental variables estimation is insignificant. The same result presented placebo-test with the US data. Overall, the results suggested strong negative effect - value losses were persistent across time. The opposite result was received by Cambell and Minguez-Vera (2008). The authors investigated Spanish companies for the period 1995 - 2000 and showed that gender diversity positively influences firm value. The investigation proved the fact that increased gender diversity can be achieved without destroying shareholder value. Overall, these results suggest that governance pressure matters and, as an exogenous factor, significantly affects firm performance. As the results has no unified solution we would test whether the presence of female directors on boards is tokenism or they can really affect firm organizational or financial processes. Thus, we build the following hypothesis:

Hypothesis 1: There is no performance effect of gender diversity of board of directors.

1.3 Industry matters

All studies use dummy variables in regression analysis, the authors mention that macroeconomic factors matter and can significantly affect our results. Intuitively, it is possible to assume that, firms from technological sector would be less committed, increasing the fraction of female directors on boards than firms from «traditional» sector. On the other hand, modern studies suggest that higher gender diversity may enhance investments in innovative developments (Bernile et Al., 2016), so technological firms can be interested in increasing the proportion of female directors on the board.

Hypothesis 2: Non-traditional firms are less «friendly» to female directors in contradictions to traditional sector.

At the beginning of the 2000s, the statistics indicate that women account for only about 20% of technology workers in the United States (Ahuja, 2002). We cannot ignore the fact that technological progress had a great impact on gender proportion in the workforce. Li et Al. (2008) mentioned the study by Shuttleworth (1992) about the position of women in the computer industry in the US and the UK, and the impact of new technology on women employed. According to this research, only 3% of the females had achieved senior managerial positions. A later study by Farrell and Hersch (2005) suggested that industry factors do influence the representation of women on corporate boards. The authors tested the effects by internal characteristics of a firm (the type of ownership, firm size, board size and so on). The result was achieved by sensitivity test.

We use the distribution from previous researches on women entrepreneur, assuming female directors are not selected as «grey», and the incentives of female directors on boards are similar to those of women entrepreneurs. To test this hypothesis, we use the distribution of sectors by Anna et Al. (1999). For today, we assume that traditionalism of sector does not have a strong significance. In the case of supporting of Hypothesis 2, we will apply sectoral dummy to highlight differences of effects.

1.4 Are female directors more risk-averse

To clearly understand the effect by women on firm performance, researchers began looking through this issue in more detail. Risk-taking of the firm is closely related to board's decisions and financial outcomes. Uncertainty and risk could both help to get higher earnings and incur large losses. Such decisions are taken by the board of directors and we have already mentioned that women are not only more risk-averse (Byrnes et Al., 1999), but also help to find consensus (Moran, 1992). The following studies provide evidence of how gender diversified boards take the risk.

Originally, the issue about risk - taking is discussed in fields of social and psychology. There are many factors that can influence the risk-taking behavior of women or men. Generally, it is argued that women are more risk averse than men, such property as social status matters and single women become more risk averse than single men (Jianakoplos and Bernasek, 1998). However, we are interested in organization performance and effectiveness of making decisions. Gender diversity in groups smooth conflicts and make a decision-making process easier (Byrnes et Al., 1999). These results take place not only in the economic sphere but also in fields of law and social (Abay and Mannering, 2016; Gorzen-Mitka, 2015; DiBerardinis et Al., 1984). An essential point is that risk taking behavior is one of the main indicators of firm effectiveness. Risk taking behavior of directors on board influence the decision-making process that is reflected in financial outcomes and firm performance.

Bernile et Al. (2016) find that a higher diversity on the boards reduces stock market volatility, and US firms take on less financial risk that leads to higher profitability and firm value. It is important to highlight the fact that authors consider the efficiency of risk - taking. According to this research, it was stated that higher level of gender diversity on boards lead to increase of investments in research & development (R&D) and other innovation investments. Also, women on boards represent risk avoidance behavior in developing markets. Moreover, it was found that large boards tend to invest in risky projects, for example, R&D (Loukil and Yousfi, 2016). Board size effect can bias results when we investigate the effect of women on R&D investments, that is why we must look through such problems as reverse causality and endogeneity. Overall, these results allow us to assume that technological and innovation firms are interested in attracting female directors.

To get more detail understanding of risk taking behavior by women, we investigate studies, focusing on different type of risk only, as previous works did not investigate different problems in models and analysis. Sila et Al. (2016) research on the relationship between three types of risk (total, systematic and idiosyncratic) and female directors on boards. In diversity - risk model we meet two main problems: omitted factors and reverse causality. Omitted factors are represented by sector (for example, innovative firms are riskier than firms from retail sector) and inner politics such as high level of corporate social responsibility. Reverse causality is a result of incentives, we cannot really know whether women choose less risky firms because of their risk aversion, or it was the inner decision by management due to risk-averse behavior of female directors. However, the authors found no evidence that female boardroom representation influences equity risk. Thus, it is a good investigation of how the results may change if the study fails to supply enough control for models. Addition model was investigated for the financial sector as it is important to divide firms by «average» debt ratio.

What happened if firm risk level was controlled by SOEs or government organizations? Khaw et Al. (2016) investigated risk-taking of Chinese firms where after Non - Tradable Share (NTS) reform (the reform was launched in 2005) the level of risk was no longer under the control of the state. The data consists Chinese firms in the period from 1999 to 2010 and the results suggested that women prefer low risk - taking, also it was stated that female directors were more active in monitoring activities and more cautious in decision making. Firm performance was negatively associated with risk-taking, that proved effectiveness of high gender diverse boards.

However, firms differ by initial debt ratio, for example, financial sector would always have higher debt ratio that retailing. We do not consider financial, insurance and real estate sector, but we can use the logic of differences in debt ratios for technological firms and traditional sector. Arun et Al. (2015) investigated the difference in the effect of female board representation on firm financial outcomes in high and low debt firms, using the UK data. The authors investigated the relationship between the quality of earnings management and board of directors. The results suggested that low-debt firms are more conservative if there are a high number of females and independent female directors on the board, however, the same representation of female on corporate boards of a high-debt firm has no effect on earnings management.

Finally, we build the following hypothesis:

Hypothesis 3: Higher proportion of women on board decrease risk - taking level of firm.

1.5 Female directors and independence

In the United States between 1950 and 2005, there was a significant increase in a number of independent directors on boards, it was not provided by legislation, but numbers were dramatically changed from 20% to 75% (Gordon, 2007). Only the period from 2003 to 2005 was affected by the introduction of Sarbanes-Oxley Act. This phenomenon initiated a great number of researches that tried to find a link between higher proportion of independent directors and firm performance. Later, when there was another trend of increasing female directors, few studies found similar results by the presence of independent or female directors. Understanding whether female directors increase the independence of the board, it is important to highlight the factors which are more closely related to board dependent and independent decisions. The following block of studies highlights main functions of independent directors and their influence in comparison with female directors.

Independence of board may have a strong correlation with firm performance and the level of board independence is affected not only by the number of independent directors, but also the effectiveness of their decisions. The following block of literature highlights the main functions of independent directors and outcomes obtained by the firm. Klein (2002) investigated the interaction between the audit committee and board of director characteristics with abnormal accruals on US large-cap firms. The study suggested that more independent board of the CEO is more effective in monitoring the financial accounting process. Another evidence suggested that independent directors can influence managers' behavior by enhanced monitoring, that, consequently, highlight the association between ownership structure and firm value (Byrd and Hickman, 1992). Moreover, tough monitoring can increase shareholder value (Rosenstein and Wyatt, 1990), making them more informed about different frauds by managers, this statement can be a solution to the agency problem.

Studies in both psychology and economic literature suggest that female behavior as monitor differ from the male. Recent studies suggest that relatively high levels of board diversity would lead to higher organizational performance and financial outcomes (Erhard et Al, 2003; Adams and Ferreira, 2009). We mean «diversity» not just as demographic (age, education, gender etc.), but as the composition of the board in general, especially, inside or outside directors.

Upadhyay et Al. (2017) investigated the association between insider ratio and firm risk, CEO compensation and firm performance, using the US data. This study represents another side of boards' independence issue. The findings suggest a positive association between the insider ratio and firm performance (Tobin's Q or ROA), however boards with higher insider ratio tend to invest less in intangible assets and more in tangible assets. Though, the results suggested that greater insider ratio manages their firms more conservatively. Nowadays, there is a significant increase in the number of technological firms, so conservative approach can be not relevant.

Looking precisely through the issue of gender diversity most investigations conclude that women operate as independent directors. Female CFOs positively affect the quality of financial reporting and earnings management (only in low-debt firms), this effect is explained by risk aversion of women and their approaches and methods to solve financial problems (Arun et Al., 2015). Adams and Ferreira (2009) found that women on boards have a significant impact on board inputs: decreasing in attendance problem, higher turnover of CEO, significant changes in compensations. Usually independent directors force CEO turnover after poor stock performance (Weisbach, 1988), as CEO turnover is sensitive to changes in the level of gender diversity on board, we can assume representation of women on boards as independent directors. This evidence is applicable not only for the US firms. Higher efficiency and independence were obtained by multi - country study, especially in Europe (Terjesen et Al., 2016; Baixauli-Soler et Al., 2015). Another study is devoted to ASEN-5 and banking industry. The significant relation between financial efficiency and board was found when women are appointed as independent directors, separately, women on board or independent directors did not have significant effect on financial outcomes (Ramly et Al., 2017). Thus, the main problem is identifying independence, as there is no strong definition of their functions. The independence and its effectiveness can be influenced by country, industry (Kang et Al, 2007) or even family connections (Abdullah, 2013; Choi et Al, 2007).

We look through the next question: can female directors on board play a role of independent directors? Existing literature suggests that independent directors has a different effect depending on industry, so the following hypothesis are assumed:

Hypothesis 4: The positive effect of independent directors on firm performance is greater when the board is comprised of a greater proportion of female directors.

Taking paragraphs of literature review all together, we will test the influence of female directors on risk and independence, trying to explain the effect on firm performance.



2. Methodology and data

2.1 Methodology

It is generally accepted that board composition is determined by internal factors. Fama and Jensen (1983) wrote that board characteristics are driven by the scope and complexity of the firm. To test the effect by representation of women on board we need to take into consideration the following barriers:

1. Omitted variables;

2. Endogeneity;

3. Reverse causality

4. Missing values.

To deal with omitted unobservable factors and endogeneity issue, we will use general least squares (GLS) model, instrumental variables (IV) model and two-stage least squares (2SLS) model. We will use an appropriate regression for each model specification and discuss the economic logic behind our instrument and its limitations in the section «Results». The problem of missing values can bias the results of regression, therefore we will winsorise such variables (command «winsor» in STATA13).

Gender diversity and firm performance

Hypothesis 1: There is no performance effect of gender diversity of board of directors.

(1)

(2)

Where the dependent variable is Q (Tobin's Q) in the first model and ROA in the second model.

PWOM - proportion of women on corporate board;

PIND - proportion of independent directors on corporate board;

BSize - size of board (the number of directors);

FSize - size of firm (logarithm of book value of total assets)

SALES - logarithm of sales revenue turnover.

We use market-based measure of performance, a proxy for Tobin's Q, and return on assets (ROA) as an accounting measure. Also, we control board size, sales and firm size. Firm size influence the level of independence of board and financial outcomes (see paragraph 2.1).

Control of board size matters as in literature review we highlight a few evidences suggested that large board size can negatively affect firm value in small firms. We also control proportion and number of independent directors on board as with higher independence of board the firm can be riskier (Fama and Jensen, 1983).

Traditional and non-traditional industries

We use the following distribution of industries to test hypothesis 2. We create the dummy variable, where 1 - traditional industries, 0 - non-traditional sectors. This distribution is applied while testing hypothesis 2 - 4 to highlight the difference in effects by female directors.

Traditional Business Type (Anna et Al.)	Traditional Industries	Non-traditional Business Type (Anna et Al.)	Non-Traditional Industries
Traditional Retail (apparel, crafts, hobby, etc.)	Consumer Durables & Apparel	Non-traditional Retail (cars, auto parts etc.)	Automobiles & Components
	Food & staples	Manufacturing	Capital Goods
	Retailing	Energy	Energy
	Food Beverage & Tobacco	Materials	Materials
	Household & Personal Products	Information Technology	Semiconductors & Semiconductor Equipment
Traditional Services	Commercial & Professional Services		Software & Services
	Consumer Services		Technology Hardware & Equipment
	Media	Telecommunications	Telecommunication Services
Health Care	Health Care Equipment & Services	Transportation (trucking)	Transportation
	Pharmaceuticals, Biotechnology & Life Sciences	Utilities	Utilities

We will apply the specification from hypothesis one and add the sectoral dummy to test the sectoral effect, using poisson regression analysis (Farrell and Hersch, 2005). We generate a new dummy variable to indicate firms from traditional and technology sectors.

Hypothesis 2: Non-traditional firms are less «friendly» to female directors in contradictions to traditional sector.

if TRAD = 0 or 1,

where TRAD - dummy variable (1 - non-traditional firms, 0 - traditional firms). We will represent the results for each sector separately (traditional and non-traditional) and compare outcomes. In case of significant difference, we would apply the segmentation to Hypotheses 3 and 4 to highlight the difference of effects by female directors on risk-taking and by independent directors on financial outcomes.

Risk-taking by female directors

Hypothesis 3: Higher proportion of women on board decrease risk - taking level of firm.

To test Hypothesis 3, we will estimate the following model:

,

where dependent variable (RISK) refers to the volatility of a firm's return on assets (ROA) over one-year overlapping periods (Khaw et Al., 2016), i.e. the difference between ROA in 2011 and ROA in 2010. A few variables were already represented in previous models, the following description concerns only new variables:

Duality - dummy variable for CEO duality (0 - CEO is not chairman, 1 - CEO is chairman);

R&D Expenditure - the value of R&D expenditures divided by book value of total assets;

CAPEX - the value of capital expenditures divided by book value of total assets;

g(Sales) - sales growth;

Leverage - total debt to total assets ratio;

FAge - firm age, that calculated as present year (2017) minus date of incorporation.

In our analysis, we will use the methodology by Khaw et Al. (2016) and Lenard et Al. (2017): we will provide OLS regression and then, compare the results with IV regression, using intercept of board size and presence of women on board. In the case of different results, we will apply additional estimations.

We include CEO duality as control variable as CEO's dual role can negatively affects firm performance through risk indicators (Chen et Al., 2015). R&D and capital expenditures, leverage, growth rate of sales and firm age are widely used as control variables in this field of analysis (Sila et Al., 2016; Khaw et Al., 2016; Lenard et Al., 2014). R&D and capital expenditures represents proxy for investment and growth opportunities. Leverage of firm is a determinant for firm risk.

Gender Diversity and Independence of Board

Hypothesis 4: The positive effect of independent directors on firm performance is greater when the board is comprised of a greater proportion of female directors.

We use different specifications of model to test Hypothesis 4 following Terjesen et Al. (2016), identifying dependent variable as logarithm of Tobin's Q and gross ROA. Such specification of dependent variables helps to get more sustainable results. We apply GMM regression with endogeneity test. According to limitations about absence of correlation with error term, we determine instrumental variables as lagged proportions of women and independent directors on board and lagged board size. Endogeneity test show whether instrumental variables are endogenous.

We control CEO Duality dummy, leverage, logarithm of ROA (in case where Tobin's Q is dependent variable), Firm Size, proportion of free float shares, fraction of insider and institutional ownership. CEO duality can negatively affect independence of board and reduce firm performance through managerial entrenchment (Duru et Al., 2016). Impact of institutional investors is ambiguous, though we include this factor (Schmidt and Fahlenbrach, 2017).

2.2 Data

The initial sample consists of data firms for Russell 3000 collected by Bloomberg and Capital IQ in period from 2010 to 2015 except financial, insurance and real estate industry. The Bloomberg data contain financial and directorship data and according to this study, we highlight female directors and independent directors on board. Directors are classified as independent if they have no business relation with the firm. The numbers of firms and observations are 1,809 and 10,859 respectively. However, we exclude firms which has a minimum information about directorship structure (zero value for period 2010 - 2014), and final sample includes 10,859 observations for 1,809 firms. According to Bloomberg data we have zero values in number of directors on board, to eliminate it, we fill the values with averaged indicators.

Descriptive statistics for firm and board characteristics is presented in Table.

Descriptive statistics

The table represents summary statistics of a panel data and consists of 10,859 observations from 2,108 firms for the period 2010 - 2015, which were in Bloomberg database. Tobin's Q is the ratio of the firm's market value to its book value of assets. ROA is net income before extraordinary items and discounted operations divided by book assets. Log(Sales) is logarithm of sales revenue turnover. The variable g(Sales) is from Bloomberg database. Firm Size is logarithm of book value of total assets. Firm age is difference between 2017 and year of incorporation. Leverage is the ratio of total debt to total assets. Both capital and R&D expenditures are divided by book value of total assets. Board characteristics: board size, number and % of Female Directors, number and % of Independent Directors, % insider ownership, % free float and % institutional ownership are from Bloomberg database. «Firm Has Female Directors» is a dummy variable where 1 - 1 or more female directors, 0 - no female directors. «Firm Has Only One Female Director» is a dummy variable, where 1 - existence at least 1 female director on board, 0 - no female directors. CEO Duality is dummy variable with 1 - CEO is also Chairman on board, 0 - otherwise.

Variable	Number of Observations	Mean	Standard Deviation	Min	Max
Firm characteristics
Tobin's Q	10,859	1.972	1.682	0	8.938
ROA	10,859	2.183	9.793	-23.267	15.051
WACC	10,148	9.343	3.656	-80.06	98.885
Log(Sales)	10,859	6.861	2.189	-4.605	13.093
g(Sales)	10,859	0.125	0.288	-0.392	1.255
Firm Size	10,859	7.152	1.94	2.805	11.349
Firm Age	10,859	27.448	25.561	1	201
Leverage	10,859	2.854	1.461	-8.517	7.729
R&D Expenditure	10,859	0.089	2.014	0	201.053
Capital Expenditure	10,859	0.05	0.66	0	1.375
Board characteristics
Board Size	10,859	8.735	2.165	1	19
Number of Women	10,859	0.954	1.076	0	9
% of Female Directors	10,859	10.606	10.588	0	71.429
Firm Has Female Directors	10,859	0.567	0.495	0	1
Firm Has Only One Female Director	10,859	0.302	0.459	0	1
Number of Female Directors Added	10,859	0.175	0.51	0	9
Female Departures	10,859	0.175	0.558	0	8
Number of Independent Directors	10,859	6.961	2.345	0	17
% of Independent Directors	10,859	66.281	30.272	0	91.475
CEO duality	10,859	0.391	0.488	0	1
% Insider ownership	8,585	6.253	10.482	0	100
% Free float	9,597	84.711	20.083	3.044	96.708
% Institutional Ownership	8,585	81.173	25.437	0	91.348

Also, we might be sure that there is a right metric of gender diversity on board, as proportion of women can increase in two ways: increase in number of female directors on board or decrease in number of directors on board, while remaining number of female directors will increase their proportion, remaining the same number. Table represents changes on board of directors for the firms from Russell 3000. The graph shows that the number of female directors grew proportionally to the growth of the number of directors on board, so we can be sure, that increase in proportion of women was not caused by board reduction.

Changes on board of directors

The table represents changes in number of female directors, proportion of women and the whole number of directors on board. The green line (represents average proportion of female directors), the blue columns (represent average number of directors on board) and the red columns (represents average number of female directors) has the same trend.

Correlation matrix is represented in Table. According to correlation matrix, we expect positive influence by proportion of women on board with firm performance.

Correlation matrix

	Tobin's Q	ROA	% of Female Directors	Female Dummy	% of Independent Directors	Board Size	Firm Age	Firm Size	R&D Expenditures
Tobin's Q	1
ROA	-0.1082	1
% of Female Directors	0.0569	0.1344	1
Female Dummy	-0.0043	0.0793	0.2196	1
% of Independent Directors	0.0172	-0.0335	-0.0366	0.0048	1
Board Size	-0.0316	0.2664	0.3842	0.2875	-0.0799	1
Firm Age	-0.1042	0.2189	0.1989	0.2024	-0.0899	0.3302	1
Firm Size	-0.0267	0.3902	0.3939	0.2845	-0.0964	0.6871	0.2955	1
R&D Expenditures	0.1361	0.2352	0.3401	0.2896	-0.0717	0.5116	0.2070	0.6479	1
Capital Expenditures	-0.0245	0.3897	0.3586	0.2520	-0.0907	0.6367	0.2875	0.7174	0.6580
Leverage	-0.0890	0.0233	0.0707	0.0712	-0.0263	0.1727	0.0843	0.1744	0.0889
g(Sales)	0.1361	-0.0592	-0.1400	-0.1373	0.0251	-0.1567	-0.1338	-0.1990	-0.1008
log(Sales)	-0.0508	0.4340	0.3743	0.2684	-0.0849	0.6449	0.3117	0.7280	0.6446
CEO duality	0.0042	0.1655	0.1495	0.1718	-0.0622	0.1513	0.1761	0.2233	0.1211
% Free Float	0.0092	0.1477	0.1841	0.4004	0.0111	0.2464	0.1735	0.3518	0.2991
% Inside Ownership	0.0369	-0.0954	-0.1249	-0.3208	0.0126	-0.2441	-0.1330	-0.3028	-0.2744
% Institutional Ownership	0.1079	0.0971	0.0681	0.2838	0.0217	0.1288	0.0253	0.2812	0.2166

3. Results

· Gender diversity and firm performance

Tobin's Q and gender diversity.

The table consists of an unbalanced panel of firm-level data from 1,809 firms for the period 2010 - 2015, which were in the Bloomberg database. Tobin's Q is the ratio of the firm's market value to its book value of assets. The specification in Column 1 includes industry dummy and represented by ordinary least squares regression. Model in Column 2 is represented by GLS regression with firm fixed effects. Column 3 reports test for robustness. Standard errors are adjusted for group correlation at the firm level in all specifications. Values of t-statistics or z-statistics are in brackets. All specifications include year dummy. Number of observations, and type of regression are represented in the last three rows.

Independent variable	Dependent variable
	Tobin's Q
	(1)	(2)	(3)
% of Female Directors	0.006*** [3.77]	0.011*** [6.05]	0.005*** [5.49]
% of Independent Directors	0.005*** [7.49]	0.007*** [14.05]	0.005*** [13.74]
Board Size	-0.007 [-1.60]	-0.023** [-2.35]	0.002 [0.31]
Firm Size	-0.05** [-2.54]	-0.026 [-0.98]	0.018*** [1.55]
log(Sales)	0.017 [1.04]	0.013 [0.56]	0.018* [1.86]
Constant	0.847*** [7.05]	1.69*** [14.24]	0.244** [3.46]
Observations	10,152	10,152	9,976
	0.11	0.01	0.11
Industry dummy	Yes	No	Yes
Regression type	OLS	Firm Fixed Effects	Robust Regression

***, **, * shows significance at the 1%, 5%, 10% levels respectively.

Table represents association of female directors on board with Tobin's Q. Firstly, we provide OLS regression with year and industry dummies (Column 1). Consistent with the positive relation between gender diversity and firm performance shown in previous studies, the coefficient on gender diversity is positive and significant at the 1% level. We add firm fixed effects in Column 2 to check for effect by omitted variables, and the results does not change significantly. The proportion of women is still positive at the 1% level, moreover board size become significant at the 5% level. The sign of women proportion did not change, so we could not assume the effect by omitted firm specific factors (Adams and Ferreira, 2009). The association of board size with performance is negative, i.e. increasing of board size decrease performance. This conclusion is also consistent with previous studies.

We also conclude that independent directors and female directors has the same impact on performance. There is a positive relation between fraction of independent directors on board and performance for both OLS and firm fixed effects regressions. This result play a role of background to Hypothesis 4.

The last column (Column 3) represents robust regression with Huber iterations. We compare results from Column 3 with Column 1 and 2 to check for robustness and make sure that the results are similar. The coefficient on the female proportion is 0.005 with a p-value lower than 1%.

Concerning control variables, all models show negative relationship between Firm Size and Tobin's Q at the 1% (robust regression) and 5% (OLS regression) levels. Sales are positively related with Tobin's Q, however, the variable shows significance at the 10% level only in robust regression.

In Table, we rerun our performance regression with the other dependent variable. Columns 1 and 2 represents the same methodology as for Tobin's Q, however the results differ. OLS regression in Column 1 shows that fraction of women has no effect on ROA. However, independent directors have positive and significant influence in all specifications (Columns 1 - 5) and this result consists with Terjesen et Al. (2016). In Column 2 we represent firm fixed effects regression, the percentage of women on board is still not significant and the value is closer to zero. In Column 3 we report robustness check, providing OLS regression with the restriction of non-winsorized ROA in the interval [-3;3]. The proportion of women becomes positive and significant at the 10% level, but the variable is still close to zero.

To deal with reverse causality, we estimate two models with 2SLS regressions. We determine instrumented variable as interaction of independent and female directors. In Column 4 the impact of the fraction of women on ROA is insignificant as well as the effect of ROA on the fraction of women on corporate board. This result suggests a lack of inside pressure to add women to the board during the time frame of the study.

Return on assets and gender diversity.

The table consists of an unbalanced panel of firm-level data from 1,809 firms for the period 2010 - 2015, which were in the Bloomberg database. ROA is the net income before extraordinary items and discounted operations divided by book value of assets (winsorized). Full description of variables is represented in Table. The specification in Column 1 includes industry dummy and represented by ordinary least squares regression. Model in Column 2 is represented by GLS regression with firm fixed effects. Column 3 reports test for robustness. Column 4 reports the results of 2SLS regression. Column 5 also reports the results of 2SLS regression, but dependent variable is fraction of women on corporate board. Standard errors are adjusted for group correlation at the firm level in all specifications. Values of t-statistics or z-statistics are in brackets. All specifications include year dummy. Number of observations, and type of regression are represented in the last three rows.

Independent variable	Dependent variable
	ROA	% of Female Directors
	(1)	(2)	(3)	(4)	(5)
ROA					-0.012 [-1.16]
% of Female Directors	0.001 [1.23]	0.0003 [0.25]	0.002* [1.81]	0.011 [-1.16]
% of Independent Directors	0.002*** [4.59]	0.002*** [3.80]	0.002*** [4.95]	0.007* [1.83]	0.053*** [13.32]
Board Size	0.001 [0.12]	-0.007 [-0.91]	0.001 [0.15]	0.148*** [2.76]	0.979*** [17.54]
Firm Size	-0.069*** [-3.63]	-0.244*** [-7.46]	-0.08*** [-4.30]	-0.354*** [-3.17]	0.164 [1.39]
log(Sales)	0.115*** [7.09]	0.069*** [2.62]	0.128*** [8.12]	1.684*** [17.85]	0.661*** [6.53]
Constant	1.068*** [11.94]	2.89*** [13.48]	1.111*** [13.81]	Omitted	Omitted
Observations	8,158	8,158	8,148	10,152	10,152
Industry dummy	Yes	No	Yes	No	No
	0.22	0.11	0.05
Regression type	OLS	Firm Fixed Effects	OLS (robustness check)	2SLS	2SLS

***, **, * shows significance at the 1%, 5%, 10% levels respectively.

In contrast with the results of regressions on Tobin's Q, Board Size has a positive effect on ROA at the 5% level with firm fixed effects.

The results from Table represents different effects. It can be possible as Tobin's Q and ROA represents external and internal outcomes respectively. Tobin's Q responds to market's changes, though the fraction of women on boards can influence partnership of firms and investments. Return on assets provides efficiency of firm's assets and it closely connected to such aspect as making decisions.

Thus, positive effect by fraction of women on Tobin's Q can be explained as reaction of market on gender diversity, it was mentioned that social pressure has a significant effect and it may be the reason why the proportion of female directors on corporate boards matters. These results do not reject Hypothesis 1.

In case with ROA, we find no effect by proportion of women in all specifications except those with firm fixed effects. Absence of effect consists with the results in the study by Sila et Al. (2016). The results in Table suggest that there is no reason to expect gender diverse boards always improve firm value. The specification with firm fixed effects shows negative relation between proportion of women and ROA, so on average female directors as well as independent directors could decrease firm value. The reason of it depends on decision making process, firm's specific and other internal characteristics of board structure.

· Sectoral test

Table reports the results of adding women to the board with sectoral effect. Comparing the results, we find that a lower percentage of women on board in the previous period significantly increases the likelihood that a woman will be added to the board, however the difference of effects is not significant, but the change is higher for technological sector.