ISS Voting Policy: The Effect of Boardroom Gender Diversity on Firm Performance and Innovation

(1)

1

ISS Voting Policy: The Effect of Boardroom Gender Diversity on Firm Performance and Innovation

MSc Finance

Master Thesis

Name: Alex Valma

Student Number: 13332058

Thesis Supervisor: Dr. Florencio Lopez de Silanes Molina Date: 01/07/2021

MSc Finance, specialisation Corporate Finance

Faculty of Economics and Business, Amsterdam Business School

(2)

2

Statement of Originality

This document is written by Student Alex Valma who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

Abstract

This empirical study investigates if there is a relationship between board gender diversity and firm financial performance and innovation. A sample of 790 public listed US companies, that belong to the S&P 1500 and Russell 3000 indexes, for the time period 2014-2020 is employed, so that there are observations before and after the implementation of the ISS Voting Policy. Gender diversity is measured by the percentage of females in the board, the Blau Index and the Shannon Index for the first time, as prior literature uses one or two of the measures. As anticipated, the percentage of females on the board increased after the policy. Firm performance is measured by the return on assets (ROA) and the return on equity (ROE). Innovation is measured by the amount of research and development expenses (R&D) and the intensity of innovation. Blau Index was negative and Shannon Index was positive, and both were significant for firm performance for 2019-2020. The hypothesis that gender diversity affects positively firm performance and innovation cannot be accepted or rejected. Prior literature also shows contradictory or no effects while examining board gender diversity. This study contributes to the literature since no research has been done on US regulation on boardroom gender diversity, and the literature combining gender diversity and corporate innovation is extremely limited.

Keywords: Board gender diversity, Firm Performance, Innovation, ISS Voting Policy

(4)

4

Nowadays, gender diversity has concerned many researchers, and in particular gender diversity on corporate boards or in top management positions has attracted a lot of interest (O'Rourke, 2003; Sangle, 2010). It is argued that gender diversity on the board will increase the talent pool from which companies select their board members, and that can potentially increase firm performance and competitiveness (Doldor et al, 2012). Regarding the corporate governance literature, one of the most important issues concerning directors and managers of the firm is the diversity, so gender, racial, and cultural composition of the boardroom. Further, they define this diversity on the board of directors as the percentage of women, Asians, African Americans and Hispanics. According to the EIGE Gender Equality Index for Europe, the percentage of women as board members is only 34% for all the members of the board for all the companies that were analysed, and in fact boards have the highest participation of women, comparing to other layers of corporate governance (EWOB, 2020). Bilimoria (1995), supported that the one reason that women have such low percentage of participation in the boards is that they are less experienced and they do not qualify for leadership skills. The second reason is that in case women do have qualifications same as their counterparts, they come across the “glass ceiling” because of their gender.

Consequently, higher expectations and more qualifications are expected from them, accompanied by them getting less support and lower rewards. This research will focus more on gender diversity among the board member of the firm.

There has been an extended literature in the role of females in the boardroom, both in Europe and the US, that has proved that women are better monitors and when added to the board, but also reduce the firm value in firms that already have a good governance (Adams & Ferreira, 2009). In the Netherlands for instance there is a target quota of 30% (European Commission, 2012a) and companies that are located in Norway, France, the UK, Finland, and Sweden are the ones that are reaching a gender-balanced governance, but this is almost stable since 2019.

According to the European Commission the improvement in board gender diversity is evolving in a very low rate. Therefore, they initiated an EU Gender Equality Strategy that presents new policies, objectives and actions in order to lead to a gender-equal Europe by 2025.

(6)

6

Table 1: Region Percentage of Women in Senior Management (Catalyst, 2020) Africa

Eastern Europe Latin America European Union North America Asia Pacific (APAC)

38%

35%

33%

30%

29%

27%

But opinions differ, for instance some researchers support the opinion that women have the tendency to be less overconfident, more risk-averse, and more averse to competition (Bertrand, 2011), while others support that, women are relatively riskier in financial firms (Adams & Ragunathan, 2018). Furthermore, after examination of a gender quota in Norway the conclusion of the research stated that the consequences of the quota included younger and less experienced female board additions, increase of the number of acquisitions, and deterioration in firm operating performance (Ahern &

Dittmar, 2012). Mandatory quotas were also instituted in Belgium, France, Italy, Germany, and the state of California in the United States. As stated by the European Commission, improvements in the percent of women on the boards of listed companies have been most significant in countries that have imposed mandatory quotas or similar measures that are binding. In the same line of thought as for the quotas, my research will be based on the ISS Voting policy. Additionally, evidence shows that female board representation has a positive relation to the performance of companies that are quite innovative and creative, and these characteristics are of great importance. In general, companies that already have female directors in their boards before an implementation of a quota perform better than firms that do not have any female directors ex ante. Therefore, it should raise the question how does a policy from the Institutional Shareholder Services affect US firms in the percentage of females they will have in conclusion. In combination with the selection of less experienced and more ineffective female directors, this paper will investigate what is the overall effect of the ISS¹ voting policy that applies on companies listed in Russell

(7)

7

3000 and S&P 1500, and will inspect how US firms will increase their female directors’ percentage in the boardroom and examine the effect of this policy on the innovation activity of these firms. Further, while existing papers examine female board policies/quotas across countries, this research will investigate specifically for the US market, in which, besides the state of California, there is no similar regulation that affects firms in their decisions to hire female board members. This evidence is presented in the following Table.

Table 2: Women’s Global Representation on Corporate Boards

Country

% Women Directorships 2019

% Women Directorships 2016

% ≥ 3 WOB 2019

% =1-2 WOB 2019

% =0 WOB 2019

Quota &

Year Introduced

Australia 31.2% 26.0% 58.2% 40.3% 1.5% No

Canada 29.1% 22.8% 63.0% 35.9% 1.1% Pending

France 44.3% 37.6% 98.6% 1.4% 0.0% Yes, 2010

Germany 33.3% 19.5% 81.0% 17.2% 1.7% Yes, 2015

India 15.9% 12.8% 21.3% 78.8% 0.0% Yes, 2013

Japan 8.4% 4.8% 3.4% 63.2% 33.4% No

Netherlands 34.0% 18.9% 65.2% 34.8% 0.0% Yes, 2013

Sweden 39.6% 35.6% 96.6% 3.4% 0.0% Yes, 2016

Switzerland 24.9% 17.5% 48.8% 51.2% 0.0% Pending

United Kingdom 31.7% 25.3% 82.2% 17.8% 0.0% No

United States 26.1% 20.3% 56.2% 42.8% 1.0% CA Only,

2018 Notes: WOB stands for women on board. The table presents the percentage of the directorships for women in 2019 and 2016. For 2019 the table presents the countries that had a percentage of ≥ 3, between 1 and 2 and 0 WOB. The table also shows the countries that have a gender quota and the year this quota was introduced. Source: Catalyst (2020)

The methodology follows an ordinary least squares (OLS) method to examine first the effect of women on the firm’s performance and then to examine the effect of women on the innovation of the firm. For both these regressions the first time-frame is from 2014, the beginning of the sample data, to 2019 that the ISS policy is implemented.

1 ISS is announced a voting policy with respect to U.S. companies with no female directors serving on their boards, with a year’s grace period before implementation. The new policy will be effective for meetings on or after Feb. 1, 2020, and will be applicable for companies in either the Russell 3000 or S&P 1500 indices. After the grace period, which will allow boards who wish to do so time to recruit qualified female candidates, adverse voting recommendations may be issued against nominating committee chairs at boards with no gender diversity. Under the policy, ISS will generally issue recommendations against the election of the chair of the nominating committee, but on a case-by-case basis, the election of other directors who are responsible for the board nomination process may be impacted.

(8)

8

And the second time-frame is from 2019 to the end of the sample, 2020. The data was collected from Thomson Reuters- Eikon database, and they consist of initially 793 firms, which drop to 790 after dropping those that had missing values for all the years.

The number of firms provides us with a sample large enough to portray accurate results. In case some firms had missing values on some of the years they are still retained on the sample. Furthermore, in order to control for possible outliers in the sample, the variables are winsorized at the 1% level. Additionally, the initial sample contained two R&D variables, R&D filed and R&D actual, but the first variable had fewer missing observations and consequently it was selected as the basis for the construction of the dependent variable. More specifically, the natural logarithm of 1 plus the value of R&D is constructed. All 21 duplicate observations were reported and then dropped, and further the variables Blau Index, Shannon Index, Ln (Total Assets), Ln (Market Cap), Risk had to be constructed. Because of the endogeneity issue that surrounds the gender diverse corporate boards’ literature, a two stage least squares (TSLS) will be implemented, using the lagged percentage of females on the board as the instrumental variable, and then examining if there are any changes on the coefficients after the use of the IV. Endogeneity could potentially have many forms;

for instance, it could be that the best firms will have the advantage to take the best women directors, and therefore leaving some of the other companies with worse director characteristics that do not suit the firm and lower the prospects of the management. Moreover, another form of endogeneity would be that firms which already have female directors at their boards, have ensured to take the best female directors, and therefore all the female directors left now are of lower capabilities.

Further, there might be endogeneity since firms may have a smaller pool of female directors to choose from and also because the adjusting year is from 2019 to 2020 it is possible that the time frame may be limited for a selection of the best fit. Ahern &

Dittmar (2012) in their paper propose that endogeneity should be extinguished by using the exogenous impact of the quota, so respectively the voting policy in my research, and also controlling for CEO characteristics that might affect the results.

The second stage will use these coefficients and will then regress again the initial model and observe changes in the coefficients, as well as the residuals and the t- statistics to measure the importance of the instrumental variable. This is the manual way of performing TSLS, but since the t-statistics and the R-squared values are higher, the Stata command ivregress is going to be used to avoid these problems. The

(9)

9

research questions that are going to be answered with the methodology described above are firstly; what is the effect of the percentage of females on the board on the firm’s performance? Secondly, what is the effect of the percentage of females on the board on the firm’s innovation? Both these questions hypothesize that after the implementation of the ISS voting policy, the percentage of women in the corporate board is going to be augmented and consequently this will affect the company’s performance and innovation.

The results overall show that the gender diversity as measured by the three independent variables is not significant for either firm performance and innovation, yet in the robustness section Blau Index shows a negative and significant effect for the time period 2019-2020 on firm performance and Shannon Index has a positive and significant effect for the same period of time. Furthermore, the percentage of women indeed increased after the implementation of the voting policy by 5 percentage points on average. Also, total assets and market capitalization affect significantly the innovation of the firm, and more specifically the intensity of the innovation. For firm performance, board characteristics also have an impact on firm financial performance;

for instance, the longer the duration of the board the lower the performance. This research cannot conclude on the effect of gender diversity on firm performance and corporate innovation; nevertheless, the significance appeared in Panel B of the tables, which could potentially indicate that there is indeed an effect, positive or negative, that depends on the gender diversity index that is used. Concerning the robustness tests of the research, the independent variables were replaced by other proxies for both firm performance and innovation. More specifically, for the firm performance the lagged value of ROE was used, and the results were overall similar, but in certain cases, such as the gender diversity indexes the values were higher and they were also more significant. For the innovation regression, the scaled value of R&D expenses by the total assets was used; this measure is usually manipulated for innovation intensity.

The robustness test for the dependent variable used the lagged percentage of females on the board as a proxy for gender diversity. The lagged value has the purpose to deal with endogeneity issues, since it contains a deviation of the normal observation.

(10)

10 2. Literature review

2.1 Gender diverse corporate boards and directors

Evidence from Adams & Ferreira (2009) showed that women are better monitors since they attend more meetings and therefore, they have a higher chance of being assigned to monitoring-related committees comparing men. Consequently, if they have a more active presence at the board and monitoring committee meetings, they could lead to more intense monitoring by the board. Gender diversified corporate boards have overall better monitoring activities. They also propose that gender quotas for directors can reduce firm value for firms that already are well-governed. In cases of weak-governed firms, over-monitoring from female directors could be beneficial.

Also, Guldiken and Darendeli (2016) show in their article that a relationship between board monitoring and R&D investment suggests that there is a threshold for board monitoring, after which it can create more agency problems through less-than-optimal R&D investments.

Female directors are less hostile in investment policies, they make better decisions concerning acquisition activity, and also enhance the financial performance of firms that operate in overconfident industries. Also, around periods of economic crisis, firms that have a low number of females on the board had a higher drop in performance compared to other firms that had greater female board representation.

Chen et al (2019)

Ahern and Dittmar (2012) after examining the quota on female boards in Norway, they found evidence that it caused a large negative effect on the firm value that occurred from the extreme reorganization that the boards had to make due to the gender quota. At the same time, they find more evidence that all female directors added to the boards were substantially different from their male counterparts.

Furthermore, don’t find any evidence about increased likelihood of women CEOs being appointed following the board quota.

To add to these results, we should also consider not only personal characteristics of the directors but also board characteristics. More specifically, board size is an important variable that determines the effect that the board of directors has on firm value and decisions. Larger boards have coordination problems and additionally agency problems due to the need for communication between more directors Cheng

(11)

11

(2008). Therefore, both board and personal characteristics can affect the decision- making process of a firm. Matsa and Miller (2013) also examined Norway’s quota on female board representation, and showed a decline in corporate profitability in the short-run. Furthermore, there was also a decline in profits since female directors had a tendency to perform fewer layoffs. Norway and the Nordic countries that they study, have a high gender equality mentality, so we can expect that in countries, such as the US, there should be a stronger effect of such quotas and policies.

2.2 Boards and innovation

Huang and Kingsen (2013) also study the differences between male and female executives on corporate decisions, and they find that female executives are less likely to issue debt, and also the abnormal returns around the announcement for debt offerings are higher compared to their male counterparts. Furthermore, R&D expenses need resources to back them, which means that the firm needs to pick debt and equity financing methods, and as a result the lower probability of women issuing debt may affect the intensity with which R&D expenses take place.

Among the board’s activities is monitoring and advising, and consequently these two actions can affect a company’s strategic decisions. An et al (2020) examine the effects of board diversity on a firm’s innovation, and they find a positive relationship between board diversity and corporate innovation by the increased number of patents and also the greater number of citations these patents receive. Furthermore, diversified boards tend to engage in development of new technologies in new areas, which involve great risk since the new areas are not associated in certainty with success or failure. But over-diversified boards can potentially create more costs than benefits, which could be different depending on the firm. For instance, the positive relation of board diversity and innovation is stronger when the firm is subject to greater external governance, and this is a result of the advising function of the board.

After establishing that innovation is associated with more risk, we can also relate to Adams and Funk (2012) research that shows that female directors are less risk averse than their male counterparts. They perform their investigation on Sweden, but they compare later the results to the United States, where the gender gap is greater, and conclude that in both countries female directors usually are of younger age, have less

(12)

12

experience compared to male directors and also have the same educational background.

Prior literature that has investigated the relation of female boards and innovative success has shown that firms who have female directors do in fact have greater success in innovation. To be more specific, given certain R&D expenditures, women invest more in innovation and in patent filing. But this positive relation is sturdier when the product market competition is not high and the management is more embedded, which is in agreement with past research showing that women are better monitors. Furthermore, these observations show a larger effect for firms that belong in innovation-intensive industries Chen, Leung and Evans (2018).

It is noted that the effort that goes into a company’s innovation activity and venturing is actually determined by the board’s decision making. To be more specific, Zahra et al (2000) show that corporate entrepreneurship is stronger when the board of executive directors receives higher equity-based remuneration, as well as outside directors, when there is no duality of CEO-Chair and lastly when the size of the board is medium. Investments in innovation are long-term projects and require from the board to have patience, which is a characteristic of female directors.

Boards with gender diversity also encourage a long-term explorative behaviour, meaning that they provide room for innovation. These behaviours seem to be specific based on the gender of a board, and include the advanced monitoring and also a more risk averse short-term character for R&D investments, which is the opposite for long- term. The main motive behind the encouragement for long-term risk is the structure of senior managements’ remuneration packages, which is mostly equity based. It is known that stocks are tied to a firm’s performance, therefore the higher the performance the higher the salary. Equity-based compensation has a stronger effect in promoting R&D investments when in combination with a gender diverse board Almor et al (2019).

Even though the literature on diverse boards is extended and specifically on gender diversity, the opinions on the effect of addition of females in the boardroom are controversial. There is no specific positive or negative outcome, mainly because of the endogenous nature. So far, plenty of European countries have had implementation of gender quotas, but Greene et al (2020) study the only quota in the US for mandated

(13)

13

female board representation that was established in California in 2018. From their evidence, it is visible that US companies are male dominated and therefore there was a great need for female directors after the bill. Griffin et al (2021) supported that investment in innovation as measured by research and development (R&D) expenditures is highly risky, characterized by a prolonged period of resource commitment and a high degree of uncertainty. Patents, a common marker for innovation output, take a number of years to develop, and there is no guarantee that granted patents will turn out to be novel and impactful. In the same line of thought, Schumbert et al (1999) say that when there are controlled economic conditions, then females do not have a tendency to avoid risk and make safer financial choices.

Corporate innovation is measured by the number of patents and the citations they have and most commonly by the expenditures on research and development (R&D).

Chen et. al (2021), show that firms with additional female directors on their boards, have a more positive association between R&D and future firm performance, which they measure by earnings and operating cash flow, and therefore females have a positive impact on R&D expenses, higher innovation output and higher R&D productivity. In addition, female directors enhance the R&D outcomes by attending more board meetings and also the better R&D outcomes brought by female directors are mainly driven by their monitoring role, which is one of the main characteristics that prior literature stresses about female directors.

2.3 Female corporate boards and firm value

Assigning women positions in the boardroom has positive valuation effects, and more specifically, they improve the financial performance of the firm and the ESG compliance which overall enhance the firm value Isidro and Sobral (2014). The European Commission supports gender diversity and introduced regulation that would secure positions as non-executive directors for females by 2020 (European Commission 2012a). Because women have different and also uniquely valuable skills to the board, board performance is in conclusion better and that affects positively the firm value. Campbell and Mínguez-Vera (2008).

As mentioned above, Ahern and Dittmar found a negative relation between females in the boardroom after the implementation of the quota and firm performance. Bøhren and Strøm also find a negative correlation between the firm performance and the

(14)

14

number of female board members, more specifically when the number increases (2007). Contrary to these results, another study by Carter et al (2003) shows that between firm value and gender diverse boards, meaning women and minorities, there is a positive relationship. Furthermore, they show that as the board size and the firm size increase, the percentage of women and minorities also increases.

2.4 Theories on gender diversity 2.4.1 Resource dependence theory

According to Pfeffer (1972) boards allow firms to minimize dependence or to gain resources and board size and structure are influenced by the external environment of the firm. Pfeffer and Salancik (1978) organize the contribution of the directors to organizations to four parts: (a) advising and counselling contribution (b) providing access to new means of information related to the firm (c) provide entry to resources from external organizations, and (d) establishment of legitimacy. Kor and Misangyi (2008), find a negative relationship between the CEO’s levels of experience in the industry and the board’s levels, which implies that the board adds to the characteristics of the top management vital advice and counsel. Hillman et al (2007) show that firms with certain forms of environmental dependencies have a higher likelihood of including female directors in the board. Hillman et al (2000) also organize the types of directors based on the RDT benefits that they provide, in order to understand how each type can be more valuable or less while the external influences change. More specifically, they categorize them into three types; “business experts,” “support specialists,” and “community influentials”, which is based on the variety of types of resources that they add when appointed to a board. Diversity of the directors can bring different perspectives and progressive approaches to problems since they are not either insiders or business experts.

2.4.2 Agency Theory

The agency theory analyses the board’s responsibility of monitoring and controlling the management (Jensen & Meckling, 1976). More specifically, the theory explains the conflict between the agent and the principal. In the corporate governance literature that is translated as the conflict between the managers, which are the agents, and the board of directors, which is the principal. Therefore, increasing the fraction of independent directors in the board could reduce this conflict, and by extent improve

(15)

15

the firm performance. Onetto says that the agent works in order to satisfy the needs of the principal, and the principal in turn has to be committed to the agent under the correct remuneration. At the same time though, the principal, meaning the board, has to satisfy the needs of the shareholders as well and then the board becomes the agent and the shareholders the principal (2007).

2.4.3 Human Capital Theory

Human capital theory (Becker, 1962) examines the role of a person’s proportion of experience, education, and skills in improving this person and his/her organization.

Tharenou et al (1994) indicate in a comparison between men and women, that the latter usually invest less in education and work experience and consequently they have lower salaries and fewer promotions. But, Singh et al (2008) show in their research that for new directors, women are actually more likely to hold an MBA degree and also have working experience in an international level, compared to male directors, but they have more experience in smaller firm boards. Westphal and Milton (2000) find that women are significantly less likely to have influence on the majority of the board based on their previous experience as directors. They also suggest that diverse group members contribute to differentiated reasoning in the decision-making process. However, creating a controversy, Campbell and Minguez-Vera (2008) argue that more gender diversified boards can be affected by more diverse sentiments and critical judgement that makes the decision-making process more time intensive and less effective.

3. Data

This chapter describes the sample data and also presents the descriptive statistics for the sample. Further an analysis of the variables pictured on the tables is provided.

3.1 Sample Data

This paper investigates if there is a causal relationship between the number/

percentage of women on the board of directors and the firm performance initially and then if there a causal relationship between the female directors and the innovation of the firm. To begin with, the context under which these relationships will be tested is before and after the Institutional Shareholders Services (ISS) announced a benchmark policy in 2019. More specifically, the new voting policy with regard to U.S.

(16)

16

companies that have no female directors serving on their boards, with a year’s grace period before implementation, was applicable for companies in either the Russell 3000 or S&P 1500 indices. After the year-long grace period passed, which allowed boards time to recruit qualified female candidates, adverse voting recommendations would be issued against nominating committee chairs at boards with no gender diversity. Under the policy that would be implemented in 2020, ISS would generally issue recommendations against the election of the chair of the nominating committee.

The speculation of this paper is that firms will add women to their board of directors in order to avoid the adverse recommendations of the ISS. Therefore, the purpose of this paper is to study how the addition of women directors on the board of the firm affected the firms’ performance and the firms’ innovation, before and after the implementation of the benchmark voting policy.

The data collected are from the Thomson Reuters Eikon database, and the sample starts with the year 2014 in order to gather enough accounting information on the companies to correctly measure the differences in the performance and innovation.

Initially 793 firms that belong to the S&P 1500 index and the Russell 3000 index, which drop to 790 after dropping those that had missing values for all the years and 5,544 observations. The firms had missing values on some of the years and not consecutively for all the years, are still retained on the sample. Additionally, in order to control for possible outliers in the sample, the variables are winsorized at the 1%

level. Moreover, the initial sample contained two R&D variables, R&D filed and R&D actual, but the first variable had fewer missing observations and consequently it was selected as the basis for the construction of the dependent variable. More specifically, the natural logarithm of 1 plus the value of R&D is constructed. The 21 duplicate observations that were found were then dropped. Furthermore, the variables Blau Index, Shannon Index, Ln (Total Assets), Ln (Market Cap), Risk had to be constructed. The sample has now 5,551 observations. Data is collected until the year 2020, so that we can observe the changes after the voting policy. The year 2021 is not selected, because for many companies there was missing information.

3.2 Summary Descriptive Statistics

Table 5 of the summary descriptive statistics section shows the number of observations, the mean, the standard deviation, the minimum and the maximum for all

(17)

17

variables that portray firm characteristics. All variables are measured in millions of dollars, but ROA, ROE and Risk are all measured in percentages. It seems that the mean observations for return on assets is negative and equal to 34.8% and similarly for return on equity the sign is also negative and the mean value is equal to 5.1%.

Risk that is measured as the debt of the firm divided by the total assets has a maximum value of 81% and a mean value of 26%. Furthermore, we can see that the mean observation has R&D expenditures of 750 thousand. The natural logarithm of total assets and market capitalization is used as a proxy for firm size.

Table 5: Descriptive Statistics Firm Characteristics

Variables Observations Mean Median Std. Dev. Min Max

Total Debt 4490 1.123 52 3.333 0 24.14

Ln (Market Cap) 3215 20.879 20.851 1.446 17.295 24.587

Ln (Total Assets) 4582 20.321 20.482 2.054 13.976 25.44

Market Cap 3215 3.387 1.136 7.020 3.243 47.63

Capex 2440 0.091 0.017 229 7000 1.541

Shareholders’ Equity 2808 1.285 332 3.249 -707 21.24

ROA 3976 -.179 -.011 .438 -2.846 .329

ROE 2252 -.035 .072 .693 -3.301 2.749

Risk 4115 .233 .139 .277 0 1.332

R&D

Ln (1+ R&D)

3505 3505

62 12.158

18 16.677

148

8.035 0 0

1.070 22.373 Notes: All variables are extracted from Thomson Reuters Eikon Database. Total Debt is the sum of all short- and long-term debt. Ln (Market Cap) is the natural logarithm of market capitalization, Ln (Total Assets) is the natural logarithm of total assets; the logarithm is used to normalize the values of the variables to have as valid data as possible. Market cap is the market capitalization, Capex is the capital expenditure, ROA is the return on assets and equals the net income divided by total assets, ROE is the return on equity and equals the net income divided by shareholder’s equity. Risk is constructed as the firm’s long-term debt divided by the firm’s total assets. R&D is the amount of research and development expenses for each firm, and Ln(1+R&D) is the natural logarithm of 1 plus R&D, again used to normalize the values of the variable for more consistent results.

(18)

18

Table 6 of the summary descriptive statistics section again shows number of observations, mean, standard deviation, minimum and maximum. All variables are measured either as numbers or as percentages. The number of observations describes how many observations the variables that have values for the 790 firms actually have;

Therefore, since the sample contains firms that might have missing values for some of the years, the number of observations varies. Moreover, the size of the board has an average number of 8 board members. The percentage of females on the board shows that on average it is around 16%, and the median observation is almost 2% lower. The standard deviation is quite high, which signifies that the percentage varies a great deal among the firms. Managers have a tenure value of almost 5 years and the duration that the board keeps the same directors is about 2.2 years on average.

Table 6 (I): Descriptive Statistics Board Characteristics

Board Size 2105 8.22 8 2.089 3 14

% Females 2105 16.013 14.286 11.671 0 50

% Independent Directors 2105 73.343 77.778 16.904 0 92.308

% Non-Executive Directors 2105 78.002 77.778 10.125 50 100

% Executive Females 2104 14.509 14.286 13.421 0 50

Board Duration 2101 2.194 3 .979 1 3

Tenure 2037 4.658 4.15 2.746 .279 14.083

Blau Index Shannon Index

2105 1666

.241 0.478

.245 0.474

.15 0.111

0 0.287

.5 0.693 Notes: All variables are extracted from Thomson Reuters Eikon Database. Board size represents the number of members in the board. % Females 2014-2018 represents the percentage of females on the board for the period 2014 to 2018; the same goes for % Females 2019-2020, for the period 2019 to 2020. % Independent Directors shows the percentage of independent directors on the board. % Non-Executive Directors is the percentage of outside/ non-executive directors on the board. % Executive Females is the percentage of female executive directors on the board. Board duration is the period of time that the board remains unchanged.

Tenure is the period that the CEO stays on the firm. Blau Index is calculated as 1 − 𝛴_𝑖=1^𝑛 𝑃_𝑖², where Pi is the percentage of board members for each gender and n is the total number of board members. Shannon index, calculated as −𝛴_𝑖=1^𝑛 𝑃_𝑖𝑙𝑛𝑃_𝑖, where the components have the same meaning as for Blau Index.

(19)

19

Table 6(II): Descriptive Statistics Gender Diversity

% Females 2014-2018 907 13.442 12.5 11.405 0 50

% Females 2019-2020 1198 17.96 16.67 11.497 0 50

Notes: This Table measures the changes in the percentage of females on the board for the period 2014- 2018, before the implementation of the ISS Voting Policy, and for 2019-2020 after.

The percentage of females on the board in 2013-2018 is around 13% in the mean and there is a large variation observed by the standard deviation, compared to the percentage in 2019-2020 the variation is almost the same but the mean now is 18%, therefore it is evident that after the implementation of the ISS voting policy the number of women on the board increased as hypothesized.

4. Methodology

This chapter firstly describes all the dependent and independent variables that will be used in the regressions. Secondly all the control variables are analysed. Lastly, the regressions and all related components are discussed.

4.1 Variables

4.1.1 Dependent Variables

For the first regression the firm performance is going to be tested, and more specifically how the addition of women directors on the board has affected it. The dependent variable therefore will be a financial performance measure. According to prior literature, several measures have been used to count firm performance, but there is still little agreement about which one should be used. The two most used types of measures are the following; First accounting-based measures that include ROA and ROE, and second market-based measures that include Tobin’s Q and stock returns.

ROA as a financial measurement has been used by previous researchers that focus on board (gender) diversity and financial performance (Berman et al., 1999; Kang et al., 2010; Tang, et al., 2012). Tobin’s Q portrays the market’s beliefs of future earnings

(20)

20

and constitutes a good proxy for a firm’s competitive advantage (Montgomery and Wernerfelt, 1988) and is a long-term measure comparing to ROA that is short-termed.

For the second regression the firm’s innovation is going to be tested, and as mentioned before, the effect that the addition of women directors to the board has. In more detail, innovation can be measured in two ways; The first one is through the R&D expenses of the company and the second one is by counting the number of patents each company has.Following Aghion et al. (2013) I will measure the firm’s expenditure for innovation activities by the natural logarithm of R&D expenditures.

For firm–years with missing R&D observations, the missing values are replaced with zero.

4.1.2 Independent Variables

The independent variables have to be proxies for gender diversity. More specifically, the first proxy is the percentage of women on the board, that is measured as the number of women divided by the total number of board members. The second proxy is the dummy variable that takes the value of 1 if there is at least a woman on the board and the value of zero otherwise. Moreover, two further gender diversity measures are calculated, that take into account both the number of gender categories, male and female, and the evenness of the distribution of the firm’s board members.

Stirling describes them as ‘dual concept’ measures of diversity, since they combine

‘variety’ and ‘balance’ (1998). The first measure is Blau index, measured as 1 − 𝛴_𝑖=1^𝑛 𝑃_𝑖², where Pi is the percentage of board members for each gender and n is the total number of board members. The values that the Blau index for gender diversity takes, range from 0 to a maximum of 0.5, which is the case when the board preserves an equal number of men and women. The second measure is the Shannon index, calculated as −𝛴_𝑖=1^𝑛 𝑃_𝑖𝑙𝑛𝑃_𝑖, where Pi and n have the same meaning as in the Blau index. Accordingly, the minimum value of the index is zero and diversity reaches the maximum when both genders are present in equal proportions, which is represented by a value of 0.69. The properties of the Blau index are qualitatively similar to those of the Shannon index although it will always take a smaller value than the Shannon index and has a greater sensitivity to small disparities in the gender composition of boards because of its logarithmic nature.

(21)

21 4.1.3 Control Variables

Board size: Is defined as the total number of members, both men and women, in the firm’s board. Larger firms tend to also have larger boards compared to smaller firms, and because of that Guest (2009) found that there is a negative relationship between the size of boards and firm performance for large firms. Additionally, Eisenberg et al find a negative relationship between board size and small to medium size firms (1998). At the same time, the larger size of the board also increases the chance of having more women directors as noted by Carter et al (2010). By measuring gender diversity as a percentage, it controls for this effect. Further, Bantel and Jackson (1989), suggest that a larger board will increase the chance of observing heterogeneity, and that signifies also higher representation of women, which will increase the innovation activity of the firms. Lastly, the view that the pool of knowledge increases with the number of board members is supported, and this leads to a more effective decision-making process that benefits firm performance (Jackling and Johl, 2009).

Firm size: Is measured as the natural logarithm of market capitalization. The natural logarithm is used in order to deal with potential economies of scale that occur in generating patents, since there are fixed costs because of the need of the firm to maintain a legal department that handles Intellectual Property (IP) issues (Hall and Ziedonis, 2001). Firm size can also be measured by the natural logarithm of total assets. Previous literature has shown that larger firms can take advantage of these economies of scale that might occur and therefore have higher operating efficiency, which results in higher firm performance (Penrose, 1959).

Risk: Is measured as the firm’s long-term debt divided by the firm’s total assets. Risk has been found to affect firm financial performance after having controlled for many factors. More specifically, risk is an extremely important factor that needs to be taken into account, since the amount of money that investors will grant for a company will be determined by the amount of leverage that this company has. Consequently, for a firm that has a relatively lower fraction of risk, it should have greater chances of possible growth and success of the firm.

Capex and Revenue: These two variables affect the investment of a company, so the growth and investment opportunities. Literature has showed that certain firms have

(22)

22

the tendency to invest less in higher risk R&D projects comparing to lower risk capex projects. (Groci et al, 2011)

Shareholder’s Equity: The cost of shareholders' equity is important because it constitutes as a comparison for investment opportunities (Rose & Hudgins, 2008).

The returns that investors expect, are affected by size, risk and growth. (Babadi and Salehi, 2017)

Tenure: Robinson and Dechant (1997), support the notion that demographic characteristics such as education, tenure, experience and age affect values and beliefs of people. These beliefs and values are not randomly distributed over people but instead they are developed through various demographic characteristics. Therefore, the board of the firm, that is basically human capital, can be a competitive advantage and also can allow for higher innovation, that can also turn the firm to be more competitive due to the board’s demographic background.

CEO-Chairman Duality: Evidence has shown that if the CEO is not a member of the board that positively affects the firm performance. Dogan et al (2013) show in their research that the performance of the companies that there was no duality of the CEO was higher when compared to firms that the CEO was a member of the board. That can be explained by the agency theory; More specifically, when it comes to monitoring and entrenchment, CEO duality can cause the manager to be more entrenched because of the undivided formal authority that he/she receives, and as a result to limit the firm’s monitoring and consequently the firm’s financial performance (Finklestein, D’Aveni, 1994). Lastly, based on the agency theory, the identification of the CEO as a chairman at the same time actually has negative effects on monitoring and consequently on firm performance.

Independent and Non-executive directors: Firms that have a vigilant board- a board that consists of a large group of independent, outside directors, that are not affiliated with the firm, whose board they are going to sit (Johnson et al, 1993). According to Fama and Jensen (1983), independent directors are more vigilant comparing to their non-independent counterparts, because their intense monitoring activities focus on financial performance, furthermore they have a higher likelihood of dismissing the CEO after poor performance and at the same time they have to protect their reputations as directors (Coughlan and Schmidt, 1985; Weisbach, 1988).

(23)

23

Officer Age: The age as a variable usually shows the experience that the CEO has, and this experience can signify the executive’s future performance. Instead of using tenure, the age is a better proxy for the executive experience (Smithey, 2009).

Furthermore, the CEO’s age can affect the risk preferences that the manager has.

Usually, managers of higher age have the tendency to be more risk-averse compared to managers of younger age. Therefore, a young CEO can potentially take more risky projects (Abed et al, 2014).

4.2 Regression Models

This research is going to be conducted following the simple linear model, Ordinary Least Squares (OLS) as a first test and will continue by using Two Stage Least Squares (TSLS) in order to deal with potential endogeneity.

4.2.1 Hypotheses

H1: Gender diversity of the board will increase firm performance H2: Gender diversity of the board will increase firm’s innovation

The first hypothesis of the research is based on several papers; Adams and Ferreira (2009) find that women are better monitors and have higher attendance in board meetings, therefore the research hypothesizes that the effect of gender diversity on the board could increase firm performance. Low, Roberts and Whiting (2015), show that females have a positive effect on firm performance measured by the return on equity.

In contrast Ahern and Dittmar (2012) examine the Norwegian gender quota and show a negative effect of gender diversity on firm performance. Further, based on the resource dependency theory mentioned in the literature review, it is expected that the age of the female member added could affect the firm performance negatively. In order to test the first hypothesis, the ROA will be the measure for firm performance and the variables testing the gender diversity of the board after the ISS policy will be:

Blau Index, Shannon Index and the percentage of women on the board. All of the three variables will be represented in the regression under the variable name

“WOMAN”. For the second hypothesis, the firm’s innovation will be measured using the natural logarithm of 1 plus the R&D expenditures of the firm. Based on the research by Griffin, Li and Xu (2021), gender diverse corporate boards actually have more innovation in their performance. An et al (2021) support the notion that more

(24)

24

experienced boards, therefore with longer tenure, have better advising capabilities and consequently have an improved innovation performance. The variables measuring the gender diversity will be the same for the second regression as with the first. The variable name Board signifies all the control variable that are board characteristics, and the variable Z signifies all other control variables.

Table 3: OLS Regression Models

Model 1 𝑅𝑂𝐴_𝑖𝑡 = 𝛽₀ + 𝛽₁∑ 𝑊𝑂𝑀𝐴𝑁_𝑖𝑡 + 𝛾 ∑𝐵𝑜𝑎𝑟𝑑_𝑖𝑡+𝛿 ∑𝑍_𝑖𝑡 + 𝜀_𝑖𝑡

Model 2 𝐿𝑛(1 + 𝑅&𝐷)_𝑖𝑡 = 𝛽₀ + 𝛽₁∑𝑊𝑂𝑀𝐴𝑁_𝑖𝑡+ 𝛾 ∑𝐵𝑜𝑎𝑟𝑑_𝑖𝑡+ 𝛿∑𝑍_𝑖𝑡+ 𝜀_𝑖𝑡

In order to deal with endogeneity, an instrumental variable will be needed. More specifically, one kind of endogeneity could be that the best firms will also have the advantage to take the best women directors as well. Furthermore, another kind of endogeneity problem would be that firms which already have female directors at their boards, have ensured to take the best female directors, and therefore all the female directors left now are of lower capabilities. Further, there might be endogeneity since firms may have a smaller pool of female directors to choose from and also because the adjusting year is from 2019 to 2020 it is possible that the time frame may be limited for a selection of the best fit. Ahern & Dittmar (2012) in their paper deal with this endogeneity by first of all using the exogenous impact of the quota, so the voting policy in my research, and also controlling for CEO characteristics that might affect the results. Arellano and Bover (1991) suggest the use of lagged values of the predetermined variables as instruments. In time series linear models, it is usually maintained that there is correlation of all the explanatory variables and consequently, the lagged value consists of a deviation of the original observation and it can be used as an estimator. The instrumental variable that is going to be used in this research is the lagged percentage of women directors on the board. Therefore, a TSLS regression method is appropriate to include the IV.

For the TSLS first stage, an OLS regression of the IV will be conducted. For the second stage all regressions in Table 3 will be regressed using OLS, but including the coefficients from the first stage.

(25)

25

Table 4: TSLS Regression Models

Model 1 𝑊𝑂𝑀𝐴𝑁_𝑖𝑡 = 𝛽_𝜊+ 𝛽₁𝑤𝑜𝑚𝑒𝑛𝑝𝑒𝑟𝑐𝑒𝑛𝑡𝑖𝑡−𝑘+ 𝛾∑ 𝐵𝑜𝑎𝑟𝑑_𝑖𝑡+𝛿 ∑𝑍_𝑖𝑡+ 𝑢_𝑖𝑡 Model 2 𝑅𝑂𝐴_𝑖𝑡 = 𝛽̂₀+ 𝛽̂₁∑𝑊𝑂𝑀𝐴𝑁_𝑖𝑡+ 𝛾̂ ∑𝐵𝑜𝑎𝑟𝑑_𝑖𝑡+𝛿̂ ∑𝑍_𝑖𝑡+ 𝑢_𝑖𝑡

Model 3 𝐿𝑛(1 + 𝑅&𝐷)_𝑖𝑡 = 𝛽̂₀+ 𝛽̂₁∑𝑊𝑂𝑀𝐴𝑁_𝑖𝑡+ 𝛾̂ ∑𝐵𝑜𝑎𝑟𝑑_𝑖𝑡+𝛿̂ ∑𝑍_𝑖𝑡+ 𝑢_𝑖𝑡

5. Results

This section presents the results of the Ordinary Least Squares regressions for firm performance and also firm innovation in Tables 7 and 8. In addition the results from the implementation of an instrumental variable for Two Stage Least Squares method is showed in Tables 9 and 10.

Table 7: OLS Regressions and Fixed Effects Firm Performance

(1) (2) (3) (4)

Variables Lagged ROA

Panel A

Lagged ROA Panel A Fixed Effects

Lagged ROA Panel B

Lagged ROA Panel B Fixed Effects

% Females -0.000417 0.000552 0.00719 0.00432

Ln (Market Cap)

(-0.131) 0.00840

(0.166) 0.00837

(1.381) -0.000128

(0.547) -0.00954 Ln (Total Assets)

(0.803) 0.0724***

(0.719) 0.0366**

(-0.00931) 0.100***

(-0.346) 0.0907**

(5.689) (2.302) (5.425) (2.079)

Market Cap -0** -0 -0* 0

(-2.058) (-0.170) (-1.743) (0.422)

Tenure 0.00877*** -0.00274 0.00929*** 0.00952

(3.837) (-0.459) (3.119) (0.436)

Board Duration -0.0145** 0.0233* -0.0109 0.0484

(-2.285) (1.732) (-1.261) (1.279)

% Executive Females 0.000447 0.000570 -0.000028 -0.000127 (0.980) (1.031) (-0.0427) (-0.0937)

% Non-Executive

Directors

0.000322 (0.460)

-0.000612 (-0.778)

0.000901 (0.890)

-0.00119 (-0.598)

CEO Duality -0.00771 0.0131 -0.00140 -0.0311

(-0.633) (0.661) (-0.0859) (-0.559)

% Independent Directors

-0.000527 (-1.342)

0.000489 (0.915)

-0.000588 (-0.943)

0.00165 (1.021)

(26)

26

Table 7 (Continued)

Board Size -0.0111*** 0.00266 -0.00878* -0.00641

(-3.281) (0.603) (-1.905) (-0.485)

Risk 0.0180 0.0221 -0.0270 0.0422

(0.352) (0.575) (-0.371) (0.354)

Shareholders’ Equity -0 -0 -0* -0

(-1.117) (-0.360) (-1.782) (-0.943)

ROE 0.0951*** -0.00251 0.0846* -0.0492***

(3.301) (-0.372) (1.805) (-2.923)

Blau Index -2.578 0.203 -7.410* -6.105

(-1.175) (0.108) (-1.959) (-1.101) Shannon Index 2.127

(1.271)

-0.126 (-0.0892)

5.667**

(1.969)

5.225 (1.218)

Constant -1.894*** -1.083*** -2.730*** -2.649***

(-5.100) (-2.961) (-4.733) (-2.615)

Observations 1,071 941 642 476

R-squared 0.282 0.832 0.287 0.820

Year fixed effects No Yes No Yes

Firm fixed effects No Yes No Yes

AR (2) 0.270 0.755 0.267 0.611

Notes: Robust t-statistics in parentheses *** p<0.01, ** p<0.05, * p<0.1. The dependent variable is the lagged value of ROA, and the purpose of the lag is to reduce omitted variable bias. Risk is calculated as the amount of long-term debt divided by the total assets of the firm. Gender defining independent variables are the percentage of females on the board, the Blau and the Shannon Index. All other variables are control variables. Panel A stands for years 2014 to 2018, therefore before the implementation of the policy and Panel B signifies the years 2019 to 2020, during and after the implementation of the policy. Columns 2 and 4 include firm-level fixed effects and year-level fixed effects.

Table 7 presents the results of the OLS regressions, with dependent variable the lagged value of ROA and independent variables the percentage of females on the board, the Blau Index and the Shannon Index. In Panel A, the independent variables are regressed for the year 2014 to 2018, while in Panel B the regressions are performed for years 2019 to 2020. In the first column, the effect of the percentage of females on the board is negative, but insignificant. In the second column though, the coefficient has almost the same absolute value but the sign is now positive. For columns 3 and 4 the coefficient remains almost the same. None of the coefficients for the percentage of females on the board are significant though. Controlling for heterogeneity in the firm and year level using fixed-effects reduces the estimated magnitude of the percentage of board gender diversity variable for as well as making it statistically insignificant. As expected, the natural logarithm of total assets affects

(27)

27

negatively the return on assets of the firm since it’s the denominator when calculating the ROA, and is also significant in the 1% level, but after using the fixed effects the significance is reduced for both time periods. The same pattern is observed for the tenure of the board; the effect is quite significant for columns 1 and 3, but in columns 2 and 4 the effect becomes insignificant, which shows that controlling for fixed effects is indeed important. Looking at the board size variable, it significantly affects in a negative way the ROA of the firm, and on columns 2 and 4 the significance does not exist. Furthermore, the return on equity affects the return on assets significantly in almost every column. When ROE increases due to the firm taking more debt, then the ROA decreases and this is something we observe on columns 2 and 4 of Table 6. In regards to the Blau Index, the effect of one percentage point increase in the index on the ROA of the firm is negative and quite large when compared to the positive coefficient in the second column of the table. For Panel B the variable is correlated significantly on the 10% level with the dependent variable, but after the fixed effects it turns insignificant. For the Shannon Index, a significance is again observed on column 3, same as the Blau Index, but now on the 5% level which dissolves in column 4. Overall, the hypothesis that the gender diversity of the board, measured by the percentage of females and the two gender indexes, has a positive effect on the firm performance is not verified from the table.

Table 8: OLS Regressions and Fixed Effects Corporate Innovation

(1) (2) (3) (4)

Variables R&D Panel A

R&D Panel A Fixed Effects

R&D Panel B R&D Panel B Fixed Effects

% Females 0.00877 -0.0177 -0.00511 0.0259

(0.177) (-0.182) (-0.100) (0.144)

Ln (Market Cap) -0.0535 0.475 0.0621 0.294

(-0.282) (1.174) (0.460) (0.411)

Ln (Total Assets) -0.259 -1.177** -0.0920 -0.665 (-1.470) (-2.008) (-0.699) (-0.563)

Market Cap 0 -0 -0 0

(1.076) (-0.782) (-0.707) (0.0398)

Tenure -0.0807

(-1.106)

0.179 (0.770)

-0.00291 (-0.0641)

0.294 (0.578)

ISS Voting Policy: The Effect of Boardroom Gender Diversity on Firm Performance and Innovation