Do gender board quotas in high-tech matter? Explaining variance of gender board quota effects on gender board equality and performance of high-tech firms across EU member states

(1)

Thesis Public Administration

Do gender board quotas in high-tech matter?

Explaining variance of gender board quota effects on gender board equality

and performance of high-tech firms across EU member states.

Student: J.J. Marinussen (s1270567)

Universiteit Leiden – Campus Den Haag

Public Management - Linking Politics and Policy

Supervisor: Dr. B.S. Kuipers

Second reader: Prof. dr. S.M. Groeneveld

Words: 19,209

Date: 10/08/2018

(2)

J.J. Marinussen - s1270567 - Public Administration: Public Management

Preface

Finally the time has come to hand in my curriculum with my thesis. With great interest and the quote in mind of A. Fuentes “Gender parity is not only good for women, it’s good for societies” I started writing my thesis for the master in Public Administration at Leiden University. Along the way I became more and more intrigued by the views of different people with regard to this topic, the initiatives firms promote and the measures governments take. This triggered the idea that the intention to achieve gender board equality is a complex objective challenged by different contexts. By this multilevel analysis based on longitudinal data I hope to provide relevant insights with regard to sector that is known for its lack of gender diversity: the Information and Communications Technology (ICT) industry. I would be great to see that an increase in women in board will trigger more to follow and benefit society.

I would like to take the opportunity to thank several people who supported me during this journey. First of all I would like to thank my thesis supervisor Dr. B.S. Kuipers who guided me throughout the process for his patience, relevant feedback and elaboration on multilevel analysis. In addition, I would like to thank Dr. P.E.A. van den Bekerom for her help creating the right multilevel models and equations. Your help has been much appreciated and really gave a boost to my analysis and statistically significance to my results. Finally, but not least important, I would also like to express my gratitude to my friends and family for their mental support and patience throughout the master.

In addition, I really enjoyed my master and writing my thesis at Leiden University. Thank you for the great experience! I hope you will enjoy reading it too.

(3)

Abstract

The lack of women in boards is a worldwide problem, which is claimed to hinder economic development and growth. Especially the high-tech sector is known for its lack of gender diversity. In order to unlock the potential of women in high-tech boards, it is believed that appointing more women to board will boost and will be beneficial to our economy and society. This thesis aims to contribute to extant literature and provide a deeper understanding regarding the effects of gender board quota on gender board diversity and performance of high-tech firms across EU member states. In order to explain variance across countries, a multilevel analysis was conducted based on data of 199,600 board members, active in 12,925 firms, nested in 11 EU member states. Data was retrieved from database Orbis – Bureau van Dijk. For the purpose of this thesis, solely large ICT firms active between 2006-2016 were selected. During this study it was argued whether the introduction of a gender board quota will lead to increased levels of women within boards (BODs) and firm performance (ROA, ROE, Tobin’s Q, Patents). Although various studies suggest that the introduction of gender board quotas will improve firm performance, it is hypothesized that in an innovation-driven sector this might hurt firm performance (Patents). This study supported that more women in the boardroom led to partial improved firm performance. Empirical analysis revealed that variance in both accounting (ROA, ROE) and market-based (Tobin’s Q) performance can be explained by differences between firms. In addition, a large part of the variance in innovation performance (Patents) could be explained by differences between countries. The multilevel regression found significant proof that and increase in women on boards (WOB) will positively affect accounting measures (ROA, ROE). This supported the assumption that observations are level dependent. Comparison between

self-governance and gender board quotas showed that WOB has a stronger effect on ROA and ROE when their firms are exposed to a gender board quota. This makes it important for (high-tech) firms, national governments and the EU to develop and participate in context appropriate policy development.

Keywords: gender diversity; gender board quota; corporate governance; firm performance;

Boardrooms, multilevel analysis

Files?

 Data set: https://drive.google.com/open?id=1nI6o3vQD-1m2dnKRVetcSlqybY_QRUIC

 References:

(4)

1. Introduction

1.1 Problem indication

Diversity is important for an economy to grow and develop. Especially the lack of gender diversity in boards, in some countries and sectors more than others, has been a problem for years. Gender diversity is not only considered the most discussed diversity topic with regard to participation in boards, economy and society (Campbell & Mínguez-Vera, 2008, p.437). Gender equality is an important value of the EU, where unfortunately large publicly listed companies show a rate of just around 23% of women in boards (European Commission, 2016; Deloitte, 2017). However, gender equality has become an important issue on political agendas and the percentage of women occupying a board position has been increasing, this rate is still considered to be too low. Still much effort is needed to understand the origins of gender inequality to achieve a gender balance within boards.

The improvement of board seats held by women is important for firms. Various studies claim that a higher female representation in boards will lead to more diversity and a higher impact on financial firm performance (Smith et al., 2006; Campbell & Mínguez-Vera, 2008; Lückerath-Rovers, 2013). When firms will avail themselves more women in boards, less talent would be wasted and new opportunities for economic growth could be unlocked (Smith et al., 2006; Campbell & Mínguez-Vera, 2008; Lückerath-Rovers, 2013; Byron & Post, 2015; Bennouri et al., 2018). However, there are quite some studies that prove otherwise (Jianakoplos & Bernasek, 1998; Lau & Murnighan, 1998; Rose, 2007; Bøhren & Strøm, 2010; Nielsen & Huse, 2010; Ahern & Dittmar, 2012; Pletzer at al., 2015). Nevertheless, there are a variety of initiatives that claim that a positive gender board diversity - financial firm performance relationship exists.

One of these initiatives is the female board quota, where targets are set by the EU or national authorities. After in 2011, the European Commission gave firms first the opportunity to self-regulate gender diversity within boards, not much improvement has taken place. In 2012, the European Commission initiated a legislation that set a target of 40% for large publicly listed companies to balance out the underrepresented sex within boards by 2020. Since women mostly have a minority within boards, this pushes most firms to increase their female board representation rate. However, this target has not been adopted by every country within the EU and some countries set their own legislation.

(6)

Nationally, a variety of countries adopted soft or binding quotas to get more women in board positions and on the long term achieve the EU target (Kanter, 1977; Terjesen & Singh, 2008; Lückerath-Rovers, 2015; Koch, 2015; Terjesen, 2015; Ferrari et al., 2016; Palá-Laguna & Esteban-Salvador, 2016). Between 2003 and 2017, a variety of EU countries did set female quotas for firms to achieve gender equality within boards. However 11 countries did develop female board quota policy, still a majority of 17 countries did not (Appendix VII). From the EU countries that follow a female quota, 4 countries follow the EU target of 40% and 6 countries have set a different national target (<40%). Norway is considered as a frontrunner, because during the late 2003s they were the first country to introduce a female board quota of 40%. What initially seemed to fail, turned in the long-term into a success (Ahern and Dittmar, 2012; Bertrand et al., 2014). Currently, Norway shows the high degree of 42% of seats held by women in boards (Deloitte, 2017). It is debatable whether positive gender diversity in boards and financial firm performance exists in across countries or firms that follow a female board quota based on the EU (40%) or national (<40%) target, especially in industries with a gender gap. This makes re-examination of the relationship both relevant for both economic and societal reasons. For the purpose of this study one industry was highlighted in specific.

1.2 Economic relevance: high-tech

Worldwide, firms in the high-tech sector are known for their lack of gender diversity and inclusion. Some take the lack of women in this sector for granted due to being less suitable candidates, where others see this as unused potential (European Commission, 2013). Nonetheless, in this sector gender inequality exists, which makes it important to attract more women to this sector in order to benefit from and trigger economic growth (Kroes, 2013; European Commission, 2016). Companies not taking action would not only miss the opportunity to avail themselves, but will also hurt their industry and competitive position to the rest of the world. Many tech-companies benefit from their national reputation. Such as in Germany, where companies benefit from their global domination in innovation (Breznitz, 2014). Especially in high-tech it is important to develop, exploit and market new high-technologies to remain competitive (Eurostat, 2015). For high-tech firms to perform better in one country than others, this requires innovation policies focusing on the entire innovation cycle to create growth and benefits for society. Thus, introduction of limited gender equality policies for high-tech firms might hurt their performance.

The European Commission (2013) has set various actions that should take place on European, member state, industry and 3rd sector level to enable change. In the Europe 2020 Strategy as special focal point

(7)

was included regarding the empowerment of women for technical jobs (European Commission, 2010). Neelie Kroes (2013), as Vice-President from the European Commission, highlighted the priority of the EU to attracting and maintaining more women in the digital workforce. Initiatives are especially focussed on decision making positions, such as in boards, to boost the EU its GDP with €9 billion on an annual base and create healthier businesses (Kroes, 2013; European Commission, 2016). In 2015, in Europe approximately 7.7 million people were working in the ICT sector, where just around 16% of them were women (Eurostat, 2017). Even less board seats within this sector are occupied by women. Although, this sector has been showing massive growth, it is still requiring the fulfilment of 120 000 vacancies a year. The European Commission (2016) expects that by 2020 the lack of diversity in the ICT sector will increase and a gap of 900 000 skilled employees will occur. In response this will lead to less women available to occupy board positions, signal and monitor other women. This makes it relevant and urgent to understand the gender board diversity - firm performance relationship thoroughly within the context of the European ICT sector.

1.3 Societal relevance

In addition, unravelling these phenomenon’s is practically relevant to the public domain. How public organisations act in the future is not only bound to its individual decisions in the past, but depending on the behaviour of others. It is expected that how countries and firm behave is subject to institutionalisation (March & Olsen, 1989; Powell & Dimaggio, 1991; Finnemore & Sikkink, 1998; Mahoney & Schensul, 2006; Allemand et al., 2014; Terjesen et al., 2015). In addition, since already several countries imposed a gender board quota it is likely that more policy will follow (Simmons et al, 2008). However, various studies aim to unravel the relationship between gender board quota, women on board and performance, still not much consensus could be found. Especially evidence in the context of technology driven firms lacks. With the uncertainty it is not only relevant for firms, but predominantly important for policy development to gain a deeper understanding of this relationship.

1.4 Problem statement

This leads to the following problem statement:

To what extent do gender board quotas lead to more women on boards and higher performance in high-tech?

(8)

During this study the following research questions were answered: Q1) How does extant literature define the relationship between gender diversity within boards, financial firm performance and female board quotas?, Q2) How do gender board quotas affect gender diversity within boards and firm performance in the context of high-tech firms?.

Chapter 1 ‘Introduction’ provided an indication of the main problem of interest, problem statement and research questions. Chapter 2 ‘Theory & hypothesis’ provides a theoretical background that build support for the hypotheses. Chapter 3 ‘Methodology’ elaborates on multilevel modelling and analysis. Chapter 4 ‘Results’ presents the most important findings and interpretations of the multilevel analysis. Chapter 5 ‘Conclusion’ elaborates on the most important contributions to extant literature,

(9)

2. Theory & hypotheses

This chapter provides theoretical background information regarding gender board quotas, female board representation and firm performance in the European high-technology domain. First theory regarding the non-existence of a relationship is explained, secondly the existence of a negative and positive relationship between gender board quota and women on boards, thirdly the negative, positive and U-shaped relationship between women on boards and firm performance, and finally the possible relationship between all three variables. High-technology firms are later on referred to as ‘high-tech’ firms. For the overview of all literature see Appendix VIII.

2.1 No relationship

Various studies suggest that focusing on diversity will limit firms to select the most capable applicant for the board position, which will have no affect or even hurt their financial decision-making and performance (Rose, 2007; Pletzer at al., 2015). Pletzer at al. (2015, p.17), who conducted a meta-analysis based on studies regarding female board representation on performance of corporates, found that either increased or decreased level of women on corporate boards did not affect financial firm performance (ROA, ROE, Tobin’s Q). This study suggests that investments in gender board diversity will not lead to financial returns, however, it may improve fairness. Thus, in case of economic reasons an increase in female board representation should not be aimed. Rose (2007, p.411), who conducted a cross sectional analysis across Danish firms, found that no significant link exist between increased levels of female board representation and firm performance (Tobin’s Q). In addition, education and foreignness of board members did not play a significant roles. Rose (2007, p.412) tends to explain this by a process of socialisation, where new board members will take over existing behaviour and norms. This could make the style of board members diffuse and lower the effect of gender board diversity.

Based on these findings, it is expected that H0 will partly be rejected. It is expected that the effect of a gender board quota or increase in women in boards is depending on its performance objective. Since high-tech firms are highly depending on their ability to innovate, it is expected that enforcing gender board quotas on high-tech boards will negatively affect their broader firm performance. This leaded to the following hypothesis:

H0: There is no relationship between gender board quota, female board participation and performance

(10)

J.J. Marinussen - s1270567 - Public Administration: Public Management 2.2 Relationship between women on boards and firm performance

The relationship and differences between gender board composition and the benefits for firms have become an increasingly popular topic. There are several studies that suggest that a higher degree of women on boards will lead to increased levels of firm performance (Smith et al., 2006; Campbell & Mínguez-Vera, 2008; Nielsen & Huse, 2010; Lückerath-Rovers, 2013; Byron & Post, 2015; Bennouri et al., 2018), decreased firm performance (Jianakoplos & Bernasek, 1998; Lau & Murnighan, 1998; Bøhren & Strøm, 2010; Ahern and Dittmar, 2012) or is depending on the reach of a critical mass (Torchia et al., 2011; Joecks et al., 2013).

First, the positive relationship between women on board and firm performance will be deepened. Bennouri et al. (2018, p.288) studied the relationship between women in the board of directors and firm accounting (ROA, ROE) and market-based (Tobin’s Q) performance of French firms. Results showed that an increase in female board representation positively affects ROA and ROE. However, this increase lowers Tobin’s Q. In addition, the study revealed including attributes (monitoring attributes, board capital attributes) of female board members to improve the relationship with Tobin’s Q. These results imply that with the increased presence of women on boards accounting measures will increase, but with the right attributes it will lift its effect on market-based performance. Byron and Post (2015, p. 436), who conducted a meta-analysis focusing on the percentage of female board members related to performance (CSP1_{) of corporates, found that a positive relationship exists between the level of women on boards and} CSP. Findings showed that women have a higher tendency to participate in activities concerning CSR2 and social reputation. This study highlights the positive impact of firms to society. In case firms aim to achieve higher levels of CSP, it is suggested by Byron and Post (2015, p.428) to focus on female status improvements across the firm and country. In countries with a higher gender parity, it is expected that firms are more likely to focus on CSP and improve female board representation Lückerath-Rovers (2013, p.506), who studied the effect of gender board diversity on performance across Dutch firms with and without female directors, found that firms with female board members showed higher performance (ROE). In addition, findings of her study support that an increase in female board participation will improve innovativeness, transparency and reputation of the firm. This indicates that high-tech firms that are innovation driven will benefit significantly from appointing more female directors. Nielsen and Huse (2010, p.143), who conducted a survey across Norwegian firms to determine the contribution of women

1_{CSR (Corporate Social Performance): combination of results-based measures focusing on CSR, workforce diversity,}

environmental responsibility, philanthropy and codes of ethics (Byron & Post, 2015, p.433).

(11)

in boards, found that an increase in female board participation will increase board effectiveness, which would be achieved by creating open debates, reduce conflicts and focus on quality of developments. It is claimed that the main difference in board contribution between women and men can be explained by leadership style. In addition, effects of female board participation are subject to its board task and processes. Campbell and Mínguez-Vera (2008, p.447), who did research after the relationship between gender diversity in boardrooms and financial performance of Spanish firms, found that gender board diversity (Blau and Shannon index) positively affects firm performance (Tobin’s Q). In this case not an increase of the absolute amount of women, but a diverse gender composition is crucial for effective monitoring by the board and the improvement of Tobin’s Q. Smith et al. (2006, p.22), who conducted a panel study across Danish firms, found evidence for a positive link between the proportion of women in top management jobs (board of directors, top executives) and firm performance (Gross value added/net turnover, Profit on primary operations/turnover, Ordinary result/net assets, Net result after tax/net assets). Findings revealed that the difference in effect of women in board positions on performance can be linked to deviating educational background. Women in top management positions with a university showed to be active in firms with higher performance.

At the other hand, various researchers found evidence for that the increase in female board representation will affect firm performance negatively (Jianakoplos & Bernasek, 1998; Lau & Murnighan, 1998; Bøhren & Strøm, 2010; Ahern and Dittmar, 2012). Ahern and Dittmar (2012, p.160), who studied the impact of imposed gender board quota on performance of Norwegian firms, found that with the introduction of this quota the value of firms decreased significantly. At the moment of announcement a decreased stock price reaction took place. Since the quota announcement in 2002 was quite unusual, firms faced abrupt and costly limitations. In order to cope with this quota, less experienced female board members were hired that damaged performance. Overtime, this also decreased long run firm value (Tobin’s Q). Since in the meantime various countries imposed a gender board quota, it is expected that other countries will experience the introduction of a quota less abrupt. Bøhren & Strøm (2010, p.1303), who analysed politics of corporate governance by studying the economic rationale of both existing and future regulation within Norwegian boards, found no support that change in gender percentage within boards will enhance firm value. Since heterogeneity of boards would decrease effectiveness of decision making, proof was found that low gender diversity creates more value (Tobin’s Q3_{, ROA}4_{, ROS}5_). Therefore Bøhren & Strøm (2010, p.1305) advise firm owners to focus on less diversity to enhance

3_{Tobin’s Q: market value of assets divided by its book value.}

4_{ROA (Return on assets): earnings from operations after taxes divided by the accounting}

value of assets.

(12)

financial performance, but suggests that gender board diversity should instead be a part of a political program to become beneficial for society. Lau and Murnighan (1998, p. 336), who did research after diversity within groups by taking into account individual member characteristics and fault lines, found that more homogeneous groups tend to have less conflict and will be quicker and more effective in decision-making, which will reduce costs. However, more diverse groups showed to be more critical and present more creative and innovative ideas. The high-tech industry is highly reliant on innovation power, which suggests that the level of diversity could be more beneficial to some performance measures than others. In addition, Jianakoplos and Bernasek (1998, p.629), who studied the investment behaviour between single men, women and married couples, found that women are significantly more risk averse during financial decision making. People that showed to take more risks, proved to have more household wealth. Since the high-tech industry is highly unpredictable and reliant on investment, it is more likely that women will take less risks than men to benefit from financial opportunities. Thus, women will contribute to fewer firm performance than men. However, these findings may have changed overtime with the change of their socio-economic position.

In addition, there are some studies that suggest that it takes time to reach a critical mass before a positive effect on firm performance takes place (Torchia et al., 2011; Joecks et al., 2013). Joecks et al. (2013, p.70), who conducted research after gender diversity in the Boardroom and firm performance of German listed firms, found support for an U-shaped relationship where a ‘critical mass’ of at least 30% of women in the boardroom is necessary to reach higher levels of firm performance (ROE6_{than that of} boardrooms fully occupied by men. Findings suggest that a magic number of three or more women should be attracted to the boardroom to improve ROE. This study showed that a critical mass was reached overtime, where measuring and the inclusion of female board members took place over several years. Torchia et al. (2011, 312), who studied the critical mass of Norwegian corporate boards, found that a higher female board representation will improve innovation performance of the firm. This study suggests that when the absolute number of women occupying board chairs increases, the firm its level of innovation will increase. In order to determine a firm its level of innovation, board members were asked to share their perspective on firm innovation7_{. Findings showed that when a critical mass of 3 women as} been reached the firm will score significantly higher on innovation. Also in this case this amount was reached incrementally overtime.

6_{ROE (Return on equity): calculated by dividing net income by book value on equity.}

7_{Innovation measure that rates the perspectives of board members on: a) being the first firm in the industry to develop an}

innovative management system; (b) being the first firm in the industry to introduce new business concepts and practices; (c) considerably changing the organizational structure to facilitate innovation; (d) implementing development programs for personnel to facilitate creativity and innovation (Torchia et al., 2011, p.306).

(13)

Based on these studies, it is expected that high-tech firms will show a positive relationship between women on boards and firm performance and H1 will be supported. Since the success of high-tech firms is highly depending on their ability to innovate, it is expected that more women in the boardroom will lead to more innovations. Although in high-tech innovativeness is strongly linked to other firm performance (Gërguri‐Rashiti, 2015, p.96), it is expected that it will be necessary for boards to reach a critical mass in order to unlock the financial benefits of their firm. It is prospected that this magic number of percentage will change overtime. This leaded to the following hypothesis.

H1: An increased degree of women on boards will positively affect firm performance in high-tech after a

critical mass has been reached.

2.3 Relationship between quota, women on boards and performance

The introduction of gender board quota in one country triggered the discussion to impose a quota in other countries. Several studies suggest that the effect of gender board quota on the degree of women in boards and performance of firms is context dependent (Kanter, 1977; Terjesen & Singh, 2008; Lückerath-Rovers, 2015; Koch, 2015; Terjesen, 2015; Ferrari et al., 2016; Palá-Laguna & Esteban-Salvador, 2016).

Various countries decided to impose a non-compulsory gender board quota. Palá-Laguna and Esteban-Salvador (2016, p.401), who conducted research on gender board quotas in Spanish context, found that a voluntarily gender quota suggested by EU directives (40%) increased the presence of women in boards of directors. However, the Spanish Good Corporate Governance Code and Effective Equality Act were introduced in 2007 and emphasis was laid on self-regulation by firms, no compulsory gender-balance norm has been set and reached. Palá-Laguna and Esteban-Salvador (2016, p.403) indicated that a gender quota by law, focusing on penalties in case of violations and not self-regulation and voluntary compliance, is necessary to enable change. In addition, it was highlighted that countries that adopted self-regulation, due to roots in common law, that national law will not enact. According to Ferrari et al. (2016, p.23), who did research after gender quotas in Italian boards of firms listed on the Italian stock exchange and their performance, found that legally binding gender quotas has opened up the selection process to women. In July 2011, gender quotas for boards of directors in Italy were introduced by law (Law 120/2011). In August 2012, this law - also called the “Golfo-Mosca” law - became binding and after this period gradual change took place. In comparison to the Norwegian law, Ferrari et al. (2016, p.12) characterize the Italian law as a collaboration between the private and public sector for collective action, gradual process that gives conservative Italian firms time to adapt, sanction policy that functions as a

(14)

warning system with penalties varying between 100,000 to 1,000,000 euro and temporary policy that should become superfluously after the gender-balance has been reached. Terjesen et al. (2015, p.245), who studied the institutional factors that precede gender board quotas, found support that binding gender board quotas have a positive effect on female board representation since gender policies within firms tend to change along with its institutional environment of their country. Diffusion theory suggests that policy developments of one country are influenced by the policy decisions of other countries (Simmons et al., 2008, p.34). After the introduction of either non-binding (self-governance, soft) or binding quota, the discussion in many other countries followed. This study highlights the importance of linking policy developments to political institutions, such as most dominant political parties, legislation, practices and mimetic isomorphism related to boards of directors (Terjesen et al., 2015, p.234). Lückerath-Rovers (2015, p.84) studied the effect of gender board quotas across Dutch firms. This study showed that with the introduction of a soft gender board quota of 30% accompanied with a ‘comply or explain’ principle the degree of women participating boards increased slightly and did not reach its target by 2014. Therefore Lückerath-Rovers (2015, p.94) expects that binding gender board quotas are expected to be imposed. A sanction of providing an explanation within the annual reports showed not to create the intended effect within Dutch context. This study suggest next to ethical, some economic reasons relevant to our study to impose a gender board quota. Firms would be benefited by gender board quotas because it would make better usage of the female human capital, where a decrease in firm performance would be caused by the failure of firms to select and allocate decisions of the right people. In addition, by hiring the level of female board member a higher level of representation of stakeholders can be reached. Countries would be benefitted by diminishing any form of discrimination, where potential board members are dismissed on the base of their gender. It is expected that more women on boards will result in more economic power of women and fair societal output. Koch (2015, p.60), who researched the effect of gender board quotas of Germany firms, found that Germany although it has announced to impose gender board quotas before it fails to follow up on this. Instead, a flexi-quotas was introduced what is considered a balance of binding gender board quota and corporate governance. Koch (2015, p.73) suggests that the responsibility to deal with gender board quotas should be with its national authority and not subject to corporate co-determination. Although, quota is considered a suitable tool to appoint more women in boards, it is assumed that in the German context insufficient suitable candidates could be found. Therefore no binding quotas were imposed, but women favoured over men when during selection both are equally qualified. In an earlier study of Terjesen and Singh (2008, p.14), where research was done after the link of female participation in corporate boards and their context across various countries, found that differences in female board representation across firms are caused by their national context (social, political, economical). Findings show that when a country presents higher levels of female board

(15)

participation, other women are more likely to occupy senior management positions. In addition, it was found that these higher levels will lead to a smaller gender pay gap. At the other hand, in countries where females gained political power earlier, females have a lower tendency to occupy board chairs. These countries would have more room to support shifts of women participating in corporates. The study of Kanter (1977, p.967) is known for its ‘tokenism’ within groups, which could be considered as a precursor of the gender board quota. During this research a field study was conducted in a large industrial organisation. Findings show that in a skewed group distribution, female tokens are used to change group proportions (skewed, tilted, balanced) and reduce underrepresentation. In this industrial context it was found that women often had to prove their technical abilities more than men. Although most women did not have to undertake much effort to prove their presence, this often backfired on their achievements. By imposing tokens the interaction dynamics within groups, such as in the board of directors, will change. High-tech firms are known for their high levels of male domination. With the introduction of female tokens within these boards, tokenism resulted in an increase in female group representation. However, providing women a group positions did not trigger other women to become active in this group. Attracting more women would create the threat of status undervaluation. This study highlights the importance of understanding group composition based on its social and cultural contexts to unravel interaction dynamics. Again, this shows that enforcement of a gender board quota in high-tech is crucial and will lead to more inclusion of women in board decisions, especially when these activities are less likely to be undertaken by firms themselves.

Based on these findings, it is expected that H2 will be supported. Findings emphasize that gender board quotas are context dependent and not much is known about the threshold of women in boards of high-tech firms. Although it is expected that in the future more binding gender board quotas will be imposed, it does not secure the right candidate for the position, preferred group dynamics and the attraction of more women to tech boardrooms. Since the threshold of the degree of women in high-tech boards is still unknown and the lack of women in high-high-tech still high, it is expected that setting an appropriate gender board quota will be difficult and takes more time. For example, high-tech firms will struggle more to achieve a gender board quota of 30% in the Netherlands than in the services or consumer goods sector. Therefore it is expected that firms in the high-tech sector are currently benefitted more by soft or hard gender board quota than self-governance. Although it seems important to keep autonomy as a high-tech firm to appoint board members, it is expected that more women in boards will boost firm performance. In order to test whether this assumption is true, the following hypothesis was developed.

(16)

and performance of high-tech firms more positively than countries that do not enforce a gender board quota.

2.4 Diffusion of gender board policy limits innovation

Overall it can be stated that how gender board quotas are developed and imposed are country and in some cases firm dependent. Various studies support that institutionalist perspectives and diffusion theory help to explain the variation and similarities of gender board quotas across different EU member states (March & Olsen, 1989; Powell & Dimaggio, 1991; Finnemore & Sikkink, 1998; Mahoney & Schensul, 2006; Simmons et al., 2008; Allemand et al., 2014; Terjesen et al., 2015).

Several studies support that policy development is context dependent and tends to follow initiators, which will create similar policy and organisations. As mentioned before, Terjesen et al. (2015) did research after institutional origins of the enactment of gender board quotas. This study revealed that three institutional origins that explain why variation exists amongst countries with or without a binding gender board quota. First, the nation-state in question should have already adopted policies with the purpose to attract more women to the workforce. Second, in case a nation state is dominated by a left-leaning political coalition, it is more likely that a binding gender board quota will be enacted. Third, a country with an institutional policy legacy towards the achievement of gender parity is more likely to support board quotas regarding this matter. In other words, how gender board quotas develop across different countries is path dependent and subject to prior policy, political coalition and policy legacy. Allemand et al. (2014), who studied women of European boards from an institutionalist perspective, found that an increase of women in the board of directors could be assigned to coercive and normative pressures over the years. Firms that experienced these institutional pressures, often aim to meet these pressures and expectations to gain legitimacy in society. During this study, it was shown that the introduction of a binding gender board quota will speed up the process of achieving gender board equality within corporate firms and preparing suitable female candidates. Simmons et al. (2008) explain policy reform and adoption with usage of diffusion theory. Diffusion theory suggests that policy developments of one country are influenced by the policy decisions of other countries. To explain why some countries adopt a gender board quota also similarities could be sought, which makes policy, law and governments more diffuse. This makes that an initiator of a successful national policy is likely to be followed by other countries during similar policy reform. Such as the successful introduction of the corporate board quota law in Norway resulted in the adoption of similar legislation in Western-European countries (Belgium, France, Italy and the Netherlands). Mahoney and Schensul (2006) view path dependency as a crucial part

(17)

of historical institutionalism in the context of political research. They defined path dependency as history impacting future effects as the result of critical junctures, contingent events, lock-ins, self-reproducing and reactive sequences. This made it interesting to look at different events and processes in the macro and meso environment, which shaped decision-making and behaviour of different actors. According to Finnemore and Sikkink (1998), who studied how international norms are shaped and enable political change, found that the process of norm development needs to go through three stages (norm emergence, norm cascade, internalization) before it may become a law, influenced by different actors, motivations and mechanism. First, during norm emergence, norm entrepreneurs within an organisation have altruistic and ideational motivations that result in empathy and commitment to the norms, which leads to persuasion of others. After a acceptance of a critical mass, a tipping point has reached that leads to norm cascade. Secondly, norm cascade shows that states, international organisations and/or networks adopt the new norms by motives to improve their legitimacy, reputation and confidence. These different actors, motivations and mechanisms determine whether one stage will follow up another. Thirdly, institutionalization, socialization and demonstration of norms often leads to the introduction of laws, profession and bureaucracy to conform and protect to these norms. When a norm such as importance of gender board equality is conformed, this could lead to the introduction of legislation and bureaucracy to protect this norm. Thus, in order for a nation-state to adopt a gender board quota law, it will need to experience board public support for gender board equality improvements and almost an institutionalized habit in society for a state to undertake measures or even introduce a law. DiMaggio and Powell (1983) suggest that institutionalization often takes place as a process of isomorphism and is a matter of rational choice, which makes that organisations will become the similar overtime. How organisations change would often be based on how others behaved in history. DiMaggio and Powell identified three isomorphic processes, which lead to similarities between organisations: coercive, mimetic and normative-leading processes of change and make organisation diffuse. However, isomorphism may provide a template of how organisations should behave in certain social constructs, it provide limited handles to deal with new situations. According to March and Olsen (1989, p.22) institutions are shaped by ‘the believes, paradigms, codes, cultures and knowledge that support rules and routines’. How actors behave is based on logics of appropriateness bound to rules, norms and expectations posed by their social constructs (March & Olsen, 1995). In comparison to other industries, high-tech firms are expected to be more innovative. However, this may be different in some countries. For example, the norm in a country like Germany makes expectations of innovation even higher, due to its dominant reputation in innovation worldwide (Breznitz, 2014). In addition to others institutions, public organisations focus next to results, performance and outcomes also on purposefulness of their behaviour (Powell & Dimaggio, 1991)

(18)

Enforcing similar policy with regard to gender board quotas, either soft or hard, will increase similarities between high-tech firms and limit their ability to innovate. For both high-tech firms and national governments it is purposeful to stay innovative to stay competitive. As mentioned before, the high-tech industry is highly dependent on innovation, is male-dominated, and characterized by a skewed group dynamic (Kanter, 1977; Gërguri-Rashiti et al., 2015). Dilevko and Harris (1997) found that the technology sector shows stereotypical men and female roles. Based on this study, patterns were found that show that men are more often portrayed as deep thinkers and visionaries, where females often showed the simple usage of the product (Dileck & Harris, 1997, p.722). Next to this, innovation performance considered an important precursor of financial success of high-tech firms. In order to unlock innovation it could be suggested to adopt similar gender board quota policy. However, studies suggest that difference could be attributed to difference in context and it is unclear whether this type of policy is suitable to a high-tech context. It is both in the best interest of firms and national governments to protect their knowledge economy and innovation power. In order to get a deeper understanding of the threshold of women in boards and its possible effect on innovativeness of high-tech firms the following hypothesis was included.

H3: Countries that enforce a gender board quota will develop more similar policy and limit high-tech

(19)

J.J. Marinussen - s1270567 - Public Administration: Public Management 2.5 Overview of hypotheses & conceptual mechanism

Table 1 gives an overview on the main hypotheses and expectations mentioned before. Figure 1 gives an overview of the conceptual model. How these will be tested, will be further explained in the following chapter.

Table 1. Hypotheses and expectations

Hypotheses Expectation

H0: There is no relationship between gender board quota, female board

participation and performance of high-tech firms



H1: An increased degree of women on boards will positively affect firm

performance in high-tech after a critical mass has been reached.



H2: Countries that impose a gender board quota are expected to affect the degree

of women on boards and performance of high-tech firms more positively than countries that do not enforce a gender board quota.



H3: Countries that enforce a gender board quota will develop more similar

(20)

J.J. Marinussen - s1270567 - Public Administration: Public Management Figure 1. Conceptual mechanism

* This conceptual model exists of three levels of analysis: country, firm and time. See appendix IV for more information regarding Time-level.

(21)

3. Methodology

This chapter elaborates on the sample, data collection, operationalization of measurements and statistical analysis. During this study a multilevel analysis was conducted based on longitudinal data. Since different occasions of measurement over time are nested within organisations within countries.

3.1 Sample

Table 2 shows that multilevel research data regarding the population is hierarchical en requires a multistage sample (Hox, 2010, p.5). Since this study exists of three levels, where Time (Level 1) is nested in Firms (Level 2) that are nested in Countries (Level 3), the sampling procedure took place in three stages. First, a sample of 11 EU countries was selected based on the choice of their administration to enforce some form of self-regulation, soft or binding gender board quota. Observations within the same union are expected to be subject to similar economic and political opportunities and threats. It is also likely that gender board quotas are diffuse to countries within the same union. Secondly, in order to study the high-tech sector, a sample of large firms was drawn from the sub-sector ICT (Appendix I). Since especially this sub sector is characterized by high male domination, data of this sector was considered interesting for this study. Organisations from the same sub sector are expected to experience higher ICC’s between measures of board compositions and performance than in other sectors. Thirdly, the years of exposure to self-governance or quota were selected. Overall, organisations were exposed between the years 2006-2016. Since the number of available years varied per firm over time, different moments of exposure to a gender board quota were measured. Although the year of exposure may differ, the years of exposure is likely to have a higher correlation due to time needed for organisations to adjust to the new norm. Multilevel modelling makes it possible to include these different repeated measures, since there is no need for balanced data (Hox, 2010, p.79). Including longitudinal data in multilevel analysis makes it possible to explain variance more detailed. This resulted on the selection of the following countries (11), firms (17-739) over time (2006-2016) suitable for multilevel analysis. For more information about the sample selection see Appendix II.

(22)

Table 2. Data hierarchy

Level number Level Groups

1 Time 11

2 Firm 17-3798

3 Country 11

3.2 Data collection

With usage of database Orbis - Bureau van Dijk a multilevel dataset of 12,925 observations was collected. This data was made suitable for statistical analysis on three different levels: Time, Firm and Country. First just 11 EU member states, which adopted a self-regulating, soft or binding gender board quota, were selected. Since the moment of exposure to some form of regulation or quota differs, only country-year data between 2006-2016 was used. In addition, Luxembourg was excluded from this analysis since it lacked enough relevant data (Snijders & Boskers, 1993; Maas & Hox, 2010). For the country selection see Appendix VII. In addition, since the high-tech sector is considered too large only large firms of its ICT sub sector were selected. According to Gërguri-Rashiti et al (2015, p.99) large firms in ICT tend to undertake more innovation than others. In order reduce firm size and industry effects, only large ICT firms were used for data analysis. Initially this dataset provided data only on firms-level, where data about firm performance measures (ROA, ROE, Tobin’s Q, Patents) did no need further adjustment. Data on firm-year level was required to identify WOB, which required the manual connect of firm names to board gender and names. In some cases the data did not provide the gender of the board member. In some case (n=53) names of board members were used to define the gender. For example, the surname ‘Peter’ was classified as a typical male name. In addition, the accumulation of males and females made it possible to determine the amount of roles in boards (BOD roles) and compare them to the actual amount of people within boards (BOD people). By dividing the amount of females (F) by BOD roles, the degree of women in boards (WOB) could be determined. Finally, this data on firm-year level required the merging of 87 datasets to make it suitable for multilevel analysis.

(23)

J.J. Marinussen - s1270567 - Public Administration: Public Management 3.3 Variables and model specification

Table 3 presents a summary of the variables used. For this study various independent (Quota, WOB) and dependent (ROA, ROE, Tobin’s Q, Patents) variables were used. In order to identify the strength and direction of these variables four different models were developed. Since multilevel models include variables that can be linked to clear grouping criteria, the variables are assigned to each level.

Table 3 Summary of variables, definitions and sources.

Variables DV/IV/C9 _Level _Definitions _Sources

ROA DV 2 Backward-looking measure ROA calculated by: Net Income / total assets. For each measure of ROA year-end values were used.

Orbis - Bureau van Dijk, (Demsetz & Villalonga, 2001)

ROE DV 2 Backward-looking measure ROA calculated by: Net income / book value on equity. For each measure of ROE year-end values were used.

Tobin’s Q DV 2 Forward-looking measure Tobin’s Q calculated by: Total Market Value of Firm/ Total Asset Value. For each measure of Tobin’s Q year-end values were used.

Patents DV 2 Innovation measure based on the total amount of patents granted per firm. For each measure of Patents year-end values were used.

Orbis - Bureau van Dijk, (Gërguri‐Rashiti et al., 2015)

WOB IV 1 The percentage of women occupying board roles in the Board of Directors of each firm. WOB was calculated by: total amount board occupied by women / total amount board occupied by women and men. For each measure of WOB year-end values were used.

Orbis - Bureau van Dijk, Ahern & Dittmar (2012, Pletzer et al. (2015).

Quota IV 3 The required percentage of women occupying boards enforces by their local authority is the shape of self-governance, soft or binding gender board quota. The percentage of gender board quota per year was used.

European Commission, Deloitte, Catalyst, Allemand et al. (2015).

Time IV 1 The moment of exposure to a gender board quota. The year of exposure was labelled 0, the first year after 1, second year after 2, etc.

Orbis - Bureau van Dijk

Year IV 0-4 The fiscal year of firm performance, WOB and presence of gender board quota.

Orbis - Bureau van Dijk, European Commission, Deloitte, Catalyst BOD roles C 2 The total amount of

Board roles assigned to the Board of Directors (BOD) of each firm. This is an indicator of board size.

Orbis - Bureau van Dijk, Pletzer et al. (2015)

BOD people

C 2 The total amount of board members participating in the Board of Directors (BOD) of each firm. This is considered an indicator of board size.

Orbis - Bureau van Dijk, Pletzer et al. (2015)

(24)

J.J. Marinussen - s1270567 - Public Administration: Public Management 3.4 Multilevel modelling and statistical analysis

Multilevel analysis allows to examine the variance taking place at separate levels of the hierarchical system. This variance was observed by testing different models including Time, Firm and Country-level predictors. The model building strategy has been bottom-up, where fixed regression coefficients of the lowest level were added first and the random variance components of the higher levels second (Hox, 2010, p.56). According to Hox (2010) multilevel models are more able to show power and will limit bias of standard errors than OLS regressions. This resulted in the development of the following model equations. These models were used to be estimated with statistical software Stata, allowed to respect the hierarchical structure of the data and to perform suitable commands (Albright & Marinova, 2010). See Appendix VI more information regarding multilevel analysis and V for Stata commands.

Multilevel models: Model 0: Empty Ytfc = β0 + β1Year + ε Model 1: Time-level Ytfc = β0 + β1Year + β2Time + ε Model 2: Firm-level

Ytfc = β0 + β1Year + β2Time + β3WOB + ε Model 3: Country-level

Ytfc = β0 + β1Year + β2Time + β3WOB + β4Quota + ε

Ytfc = Firm performance (ROA, ROE, Tobin’s Q, Patents)

β0 = Random intercept

β1Year = Year variable β2Time = Time variable

β3WOB = Women on boards (%) β4QUOTA = Gender board quota (%)

ε = Error term

The empty model includes only the intercept and variable βYear, where at level 1 (Time-level) βTime was added, at level 2 (Firm-level) βWOB and at level 3 (Country-level) βQuota. In order to re-examine the relationship in the context of large ICT firms across Europe, empirical analysis was conducted.

(25)

4. Results

During this multilevel analysis different empirical findings were retrieved. First, this chapter gives an introduction of the most important descriptive statistics (4.1). Second, the correlations across variables were provided (4.2). Third, the intraclass correlation coefficients were discussed that explain the variance that could be attributed to difference in level (4.3). Fourth, the findings of our multilevel regression analysis were explained (4.4). Subsequently, these findings were analysed by a robustness check.

4.1 Descriptive statistics

Graph 1 shows that throughout the years 2006-2016 more ICT firms were exposed to gender board quota policy. The graph reveals that when exposure to one quota levelled off, another higher quota was

introduced. This supports that increased gender board quotas are depending on developments in and the decisions of other national governments.

Graph 1. Gender board quota throughout the years (2006-2016)

Table 4 provides a summary of the measures of tendency (frequency, mean) and variability (standard deviation, minimum, maximum). Both accounting-based measures (ROA, ROE) show a positive effect of what gender board quota has achieved.

(26)

On average ICT firms from the sample showed to have a ROA of 4.8%, which indicates an effective usage of assets that benefit the operational and financial performance across ICT firms. An average ROE of 11.5% shows that these ICT firms are quite efficient in using investment funds to generate profits. The large difference between ROA and ROE can be explained by the large presence of debt. In the high-technology industry it is considered a large risk to rely on external funding. Due to the fast pace of developments in this sector, inventions could become superfluous rapidly. An increase in debts could increase the risk of bankruptcy, which makes it less attractive for the debt market to provide investments. Therefore, it is surprising that firms from this sample do make usage of this. However, this sample exists of large to very large established firms, which are considered less risky by lenders.

The market-based measures (Tobin’s Q) provide a positive outlook what is expected to be achieved in the future (Demsetz and Villalonga, 2001, p.2719). The relatively high Tobin’s Q of 2.0% shows that stock of the ICT firms are overvalued. This means the stock value is higher than the replacement cost of assets. It is expected that these firms by improving effective usage of their resources will create more value (Demsetz & Villalonga, 2001). For example, better utilization of human capital within the BOD is likely to improve firm valuation.

The innovation measure (Patents) shows a relatively high average innovation output of 74 patents per ICT firm. This measure shows that instead of economic value the ability of large ICT firms to turn inventions into patents is high (Gërguri‐Rashiti et al., 2017, p.97). However patenting tells just a part about the innovation performance of a firm, the amount explains that these firms tend to look beyond their competencies more than others and want to be part of a larger scientific community (Henderson & Cockburn, 1994, p.64).

On average the selected ICT firms showed to have 19.6% of women on boards (WOB). Since the average gender board quota (Quota) is 4.7%, this actual percentage of women is high. However, most of the countries followed self-governance (0%), where some a soft quota varying between 25-30% or binding quota between 20-33.3%.

(27)

J.J. Marinussen - s1270567 - Public Administration: Public Management Table 4. Descriptive statistics

_____________________________________________________________________________________

N Mean SD Min Max

1 ROA 12862 4,782 17,393 -233,250 217,870 2 ROE 12869 11,491 75,701 -990,250 970,440 3 Tobin's Q 3404 2,012 10,201 0,000 284,000 4 Patents 12925 74,130 979,260 0,000 26933,000 5 WOB 11634 0,196 0,177 0,000 1,000 6 Quota 12925 4,726 9,739 0,000 33,300 7 Year 12925 2012,790 2,327 2006,000 2016,000 8 Time 12925 3,514 2,416 0,000 10,000 9 Org. 12925 6463,000 3731,270 1,000 12925,000 10 Country 12925 7,970 3,327 1,000 11,000 11 BOD roles 12925 28,900 60,912 0,000 1708,000 12 BOD people 12730 22,260 29,808 1,000 1176,000 _____________________________________________________________________________________

(28)

J.J. Marinussen - s1270567 - Public Administration: Public Management 4.2 Correlations

Table 5 shows various significant correlation coefficients that imply linear relationships worth mentioning. Correlations were interpreted with usage of Pearson’s R (Schober et al., 2018, p.1765). Looking at variables from the null-model, only Year and Quota show a weak positive correlation (r = 0.2610, p = .000). In addition, Year shows a weak negative relation with BOD roles (r = -0.160, p = .000), females (r = -0.148, p = .000) and males (r = -.158, p = .000). Time did not show any relevant relationship with variables from model 1. However, Time showed weak negative correlations with BOD roles (r= -.212, p = .000), Females (r = -.1870, p = .000), Males (r = -.2119, p = .000) and total amount of firms (r = .1085, p = .000). Looking at model 2, relevant correlations are negligible. However, WOB shows a weak positive association with Females (r = .219, p = .000). Also variables from model 3 did not show any relevant correlations. In addition, Quota showed logically a moderate but negative relation with Country and weak association with total amount of firms (r= -.465, p = .000) and unique firms (r= -.100, p = .000). It should be noted that the moderate positive relationship (r = 0.5924, p = .000) between ROA and ROE is due to similarity of input. Other performance measures did not show any correlation with other variables. However, some weak positive relationship was found between Patents and BOD people (r = 0.124, p = .000), BOD roles (r = .2173, p = .000), Females (r = .237, p = .000) and Males (r = 0.201, p =.000). Since the majority of the correlations between independent and dependent variables are significant but weak to negligible, it is important to look at intraclass correlation coefficients (ICCs).

(29)

J.J. Marinussen - s1270567 - Public Administration: Public Management Table 5. Correlations (Pearson’s R)

_____________________________________________________________________________________ 1 2 3 4 5 6 7 8 9 10 11 12 1 ROA 1,000 2 ROE 0,592 ** 1,000 3 Tobin's Q 0,047 ** -0,005 1,000 4 Patents -0,005 -0,006 -0,011 1,000 5 WOB 0,051 ** 0,029 ** -0,077 ** 0,029 ** 1,000 6 Quota 0,008 -0,004 -0,064 ** -0,002 0,010 1,000 7 Year -0,044* * -0,005 -0,026 -0,009 -0,015 0,261 ** 1,000 8 Time -0,051* * -0,014 -0,012 -0,003 -0,042 ** 0,031 ** 0,784 ** 1,000 9 Org. -0,048* * 0,006 -0,056 ** -0,042* * -0,043 ** -0,465 ** 0,002 -0,038 ** 1,000 10 Country -0,034* * 0,016 -0,075 ** -0,030* * -0,046 ** -0,496 ** -0,056 ** 0,023 ** 0,933 ** 1,000 11 BOD roles 0,015 -0,001 -0,033 0,217 ** 0,045 ** -0,072 ** -0,160 ** -0,212 ** -0,072 ** -0,095 ** 1,000 12 BOD people -0,005 -0,008 -0,048 ** 0,124 ** 0,068 ** 0,042 ** 0,037 ** -0,084 ** 0,108 ** 0,086 ** 0,361 ** 1,000 _____________________________________________________________________________________ ** Correlation is significant at .01 level (2-tailed).

According to Albright and Marinova (2010, p.11), variance component estimation is important in multilevel analysis. Pearson’s R correlation coefficient solely describes the strength and direction of the relation between variables and does not assume data is nested (Schober et al., 2018, p.1767). ICC takes into account the hierarchical structure of the population and explains similarities within groups (Hox,

(30)

2010, p.4). Since Time is nested in Firms and Countries, it is expected that ICC will be higher during the same moment, firm or country and will differ from the average correlation coefficient.

4.3 Intraclass correlation coefficients

Table 6 shows the ICC of our models at different levels. In order to calculate ICC, the variance of each level was divided by the sum of the variances of all three levels (Albright and Marinova, 2010, p.20). See figure 7 for the ICC formulas and notations at each level (Davis & Scott, 1995, p.100). Overall, it can be stated that with the addition of different variables to the empty model to some extend model fit improves. However, it was found that accounting (ROA, ROE) and market-based (Tobin’s Q) measures are strongly influenced by their shared factors within firms and innovation measure (Patents) by similar influences within countries.

In model 0, only the random intercept and variable Year were added. In order to create a realistic variance estimation for repeated measurements between 2006-2016, next to the random intercept, variable Year was added to the model (Bekerom, 2016, p.57). ICC shows that most of variance in ROA (86,497%), ROE (93.358%) and Tobin’s Q (87.656%) can be attributed to difference over time. Firm-level is able to explain the largest part of the variance in Patents (96.295%) and small proportions of variance in ROA (11.651%), ROE (6.444%) and Tobin’s Q (7.937%). Only very small parts of variance in the performance measures can be explained by Country-level predictor Quota.

In model 1, variable Time was added. Again ICC shows that Time is able to explain the largest proportions of variance in ROA (86.538%), ROE (93.382%) and Tobin’s Q (87.671%). The largest variance of Patents (06.297%) can be explained by differences between firms. With the addition of variable Time, the ability of explaining variance of ROA, ROE, Tobin’s Q and Patents slightly increased. In model 2, Firm-level predictor WOB was added. ICC reveals that again differences in Time accounts for the largest part of the variance in ROA (86.8447%), ROE (93.344%) and Tobin’s Q (86.503%). Again variance in Patents (96.297%) can be explained by differences between firms. However, with the addition of variable WOB this model slightly increases its ability to explain the variance of ROA and Patents.

In model 3, Country-level predictor Quota was added. ICC confirms again that the largest parts of the variance of ROA (86.847%), ROE (93.339%) and Tobin’s Q (87.274%) can be attributed to differences over time. The variance of Patents can largely be explained by differences between firms. Only the proportion in variance of ROA and Tobin’s Q increased slightly with the addition of Quota.

(31)

J.J. Marinussen - s1270567 - Public Administration: Public Management Table 6. Intraclass correlation coefficients

_____________________________________________________________________________________

ROA ROE Tobin’s Q Patents

Model 0 Time 86.497% 93.358% 87.656% 3.561% Firm 11.651% 6.444% 7.937% 96.295% Country 1.851% 0.198% 4.407% 0.144% Model 1 Time 86.538% 93.382% 87.671% 3.561% Firm 11.665% 6.456% 7.938% 96.297% Country 1.797% 0.162% 4.391% 0.142% Model 2 Time 86.844% 93.344% 86.503% 3.857% Firm 12.474% 6.382% 10.026% 96.019% Country 0.683% 0.274% 3.471% 0.124% Model 3 Time 86.847% 93.339% 87.274% 3.856% Firm 12.469% 6.375% 10.100% 96.007% Country 0.684% 0.287% 2.626% 0.137% _____________________________________________________________________________________

Do gender board quotas in high-tech matter? Explaining variance of gender board quota effects on gender board equality and performance of high-tech firms across EU member states

Thesis Public Administration