Master Thesis
The relationship between board diversity and firm performance
A natural experiment based on the Norwegian gender quota for corporate boardsName: Fabienne Hogenbirk Student number: 10526927
Thesis supervisor: Dhr. Dr. S.R. Arping Version: final
Date: 1 July 2018 MSc Finance
2
Statement of Originality
This document is written by student Fabienne Hogenbirk who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is 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
Abstract
The objective of this paper is to examine if and how board diversity affects the performance of companies. Data on 99 public limited companies in Norway is collected, which totals 865 firm-year observations over a sample period between 2005 and 2013. Two sets of difference-in-difference tests and an OLS-regression are designed to test for a potential relationship. The paper uses the introduced women quota of 2008 as natural experiment. The short run results indicate that the law revision had a negative effect on the performance of companies that had to reform their boards, companies that in general increased their number of female directors and companies that changed to a critical mass of female board members. The effect of the women quota turns positive in the long run for companies that changed to a higher proportion or number of female directors. The results are not always statistically significant, thus more research is required before any definitive conclusions are made. This paper has implications for both theory and practice, as many countries introduced gender quotas during recent years. The paper contributes to existing literature on corporate governance with a set-up that has not been conducted before. Key words: Norway, Gender Diversity, Board of Directors, Firm Performance.
Table of Content
1. Introduction ... 6
1.1. Background information ... 6
1.2. Research question ... 7
1.3. Methodology ... 8
1.4. Contribution to the literature ... 8
1.5. Results ... 9
2. Literature review ... 10
2.1. Board tasks and corporate governance theories ... 10
2.1.1. Tasks of the board of directors ... 10
2.1.2. Agency theory ... 11
2.1.3. Stakeholder theory ... 12
2.1.4. Resource dependency theory ... 13
2.2. Characteristic differences ... 13
2.2.1. Important differences for board positions ... 13
2.2.2. Benefits of boards with women ... 15
2.3. Diversity and firm performance ... 16
2.4. The critical mass theory ... 19
3. The Norwegian Gender Equality Act ... 22
3.1. The cause of the law revision ... 22
3.2. The requirements of the quota………...22
4. Methodology ... 24
4.1. Data and sample selection ... 24
4.2. Variables ... 25 4.2.1. Dependent variables ... 25 4.2.2. Independent variables ... 25 4.2.3. Control variables ... 25 4.3. Types of regressions ... 26 4.3.1. First regression ... 26 4.3.2. Second regression ... 27 4.3.3. Third regression ... 28 5. Results ... 30 5.1. Descriptive statistics ... 30
5.2. Kernel Density estimation ... 34
5.3. Results first regression ... 34
5.4. Results second regression ... 36
5.5. Results third regression ... 38
6. Robustness tests ... 40
6.1. Different time-periods ... 40
6.1.1. First regression ... 40
6.1.2. Second regression ... 41
6.1.3. Third regression ... 41
6.2. Different measurements of performance ... 42
6.2.1. First regression ... 42
6.2.2. Second regression ... 42
6.2.3. Third regression ... 43
5
7.1. Discussion of results ... 44
7.2. Limitations and Future research directions ... 44
7.3. Conclusion ... 46
Reference list ... 47
Appendices ... 52
Appendix 1: List of PLCs ... 52
Appendix 2: List of regions in Norway ... 53
Appendix 3: List of SIC Codes by industry ... 53
Appendix 4 ... 54 Appendix 5 ... 54 Appendix 6 ... 55 Appendix 7 ... 56 Appendix 8 ... 56 Appendix 9 ... 57 Appendix 10 ... 58 Appendix 11 ... 59 Appendix 12 ... 60
6
1. Introduction
1.1. Background information
Over the years, gender diversity within corporates and especially the role of women on the board of directors became an ongoing debate in academic literature and governmental decisions. The European Commission proposed to regulate a minimum level of gender diversity for public companies of the European Union (EU) by 2020. The proposed regulation is based on Article 157(3) of the Court of Justice of the European Union, which ensures gender equality in employment and occupation. In 2012, women managed only 13.7 per cent of all corporate board seats within the largest listed companies of the EU. The European Commission argues that sustainable growth and employment may be harmed by this degree of gender inequality, which as a consequence could affect the efficiency of the EU financial markets (European Commission Proposal, 2012).
As gender diversity among the board of directors faces increased attention, many countries started to adopt minimum gender requirements for their listed companies. Norway was the first country that introduced a quota with as potential penalty to become dissolved from the Oslo Børs Exchange, if these governmental requirements are not met. The Norwegian government requires that public listed companies have at least 40 per cent of women on their boards, but the exact number of sexes depends on the size of the board. France, Belgium and Italy support the equality concept of Norway and thereby they also introduced a women quota. Potential penalties for these countries are the elimination from their local exchange market, fines or even banishment from paying their directors. The governments of the Netherlands, Germany, Spain and the UK only advise their companies to seek for gender equality, although no penalties apply if the suggestions are not pursued (‘‘Ten years on from Norway’s quota for women on corporate boards’’, 2018).
Earlier investigations examined the presence of women on boards and their relationship to the effectiveness of the decision-making process of companies. Research shows that women and men differ in several behaviour- and skills-characteristics, but this difference depends on the situation in which people operate. An irrefutable effect of a gender mixture on firm performance is still not discovered, which makes it interesting that countries introduce quotas (Nielsen & Huse, 2010a).
7
1.2. Research question
The literature review of Chapter 2 explains the results of earlier investigations, with as main focus board diversity, characteristic differences and firm performance. Erhardt, Werbel, and Shrader (2003) state that ethnic and gender diversity increase the effectiveness of the monitoring role of the board, as more opinions and insights are taken into consideration. Therefore one can assume that men and women differ in their characteristics and that the difference influences the performance. Chapter 2 answers the questions:
• ‘Do men and women differ in their characteristics and does this difference influence the decision-making process?’
• ‘What are the benefits of female board members?’
The relationship between the proportion of women on boards and the firm performance is studied in the empirical section of this paper and thereby the first question is answered:
1) ‘Do the requirements of the Companies Act after 2008 show a positive effect on the performance of PLCs in Norway, compared to the period before 2008?’ Two more tests will investigate if related theories also show a connection. First, an increase in female board members to firm performance is observed. Second, the change to a ‘critical mass’ is examined, which assumes that a change to at least three female board members, from previously zero, one or two women, has an influence on the performance of firms. The two additional questions are:
2) ‘Does an increase in the number of female board members show a positive effect on firm performance, whether the minimum of the Companies Act is reached or not?’
3) ‘Does the change to a critical mass of women show a positive effect on firm performance, whether the minimum of the Companies Act is reached or not?’
8
1.3. Methodology
After the introduction and transformation-period of the quota between 2003 and 2007, some public limited companies (PLCs) had to reform their boards to meet the new requirements of the revised law by 2008. The unexpected change in the Norwegian law is useable as natural experiment, which provides this paper the possibility to conduct a difference-in-difference (DiD) analysis for the comparison of performances before and after the law revision (Abadie, 2005). The main regression covers the period between 2005 and 2009, as 2007 is the year that the law was revised. More over, two additional tests cover the years 2007 and 2008 and the period between 2005 and 2013.1
Control and treatment groups are designed to answer the first question. The control group is not affected by the change in the law, as this group of companies already met the required number or 40 per cent of women on their boards. The treatment group had to change their boards, as the law requirements were not met beforehand. The DiD analysis shows if there exists a relationship between a higher proportion of women on boards and the performance. Different treatment and control groups are designed to answer the third question. The treatment group in this analysis contains companies that changed from zero, one or two women on their boards to at least three women. The control group did not reach a critical mass and this group only consisted of one or two women after the law became active.
The dataset consists of 99 unique Norwegian PLCs that operate between the years 2005 and 2013. The sample requires that companies operate until at least the year 2010. Information about the companies is provided in Chapter 4. The dependent variables are the return on assets (ROA), return on investments (ROI) and return on invested capital (ROIC)2, as all are found to be one of the used accounting measurements (Erhardt, Werbel, & Shrader, 2003).
1.4. Contribution to the literature
Theory on the design of boards and its relationship to firm performance is so far weakly developed (Becht et al., 2003). This paper contributes to the existing literature, as the set-up and DiD analyses of this paper have not been used before. In
1 The comparison between the years 2007 and 2008 and the period between 2005 and 2013
are part of the robustness tests.
2 The performance measurements of ROI and ROIC are used as robustness tests for the
9
general, there has not been found a clear relationship between gender diversity on boards and the performance of firms. The law revision in Norway allows the conduction of two DiD analyses and the results contribute to the existing literature on corporate governance. Besides, Dale-Olsen, Schøne, and Verner (2013) compared the performance of Norwegian non-finance PLCs, which had to change their boards, to non-PLCs, which do not have any gender requirements, between the years 2003 and 2007. In contrary, this paper creates different control and treatment groups with a wider event-window, thereby new results are provides to the existing literature. More over, Torchia, Calabro, and Huse examined Norwegian firms and found that the change to a critical mass results in the possibility to enhance the investment activities of firms. Besides, they find that this relationship is mediated by board strategic tasks, which refers to the degree in which the board and its members are involved in strategic processes (2011). The results of the third question in this paper contribute to the existing literature on the critical mass theory.
Moreover, this research is interesting from a governmental point of view. As discussed in Section 1.1., many European countries introduced women quotas without that they have knowledge how quotas affect the performance of firms and the country’s economy (‘‘Ten years on from Norway’s quota for women on corporate boards’’, 2018). This paper will therefore contribute to the process of governmental decisions and to the question whether quotas are a strategic move for countries. 1.5. Results
This paper finds evidence for particular relationships between female board members and firm performance, but this connection is variable and sometimes insignificant. The robustness tests support the negative relationship between a critical mass of women on boards and firm performance if other periods and performance measurements are observed. Inconsistent evidence is found for a connection to the proportion and number of women on boards if wider post-law periods and different performance measurements are observed. Tables and detailed information are provided in Chapter 5 and 6.
10
2. Literature review
This chapter explains previous studies on several theories. First, the tasks and obligations of the board of directors are discusses, continued by several board-related theories. Secondly, general characteristic differences between male and female directors, which are crucial for the firm performance, are reviewed. At last, previous research on the relationship between board diversity and firm performance is discussed, in addition to the critical mass theory.
2.1. Board tasks and corporate governance theories 2.1.1. Tasks of the board of directors
Ahead of the discussion of all theories it is essential to understand what role the board of directors has to fulfil. The board of directors has several tasks and responsibilities to successfully perform their function. It is their obligation to advise the CEO (e.g. during the process of mergers and acquisitions or if policies are designed) and to monitor the CEO and management. Besides, the board has the responsibility to hire the CEO. To tackle the agency problem, which is discussed in Subsection 2.1.2., the board of directors determines what compensation is necessary to make the CEO work in the interest of the company. The board has the responsibility to assess the overall direction, performance and strategy of a company and board members are responsible for the internal governance system. The advisory and monitoring roles of board members are in conflict with each other, as the board provides advises to the CEO, but likewise monitors the same advises (Boland & Hofstrand, 2009). Garratt explains that boards have to focus on ‘conformance’ in the short run and ‘performance’ in the long run. ‘Conformance’ consists of the role to meet accounting requirements and the interests of all stakeholders and regulators, but also the supervision of managers. In the long run it is necessary to design a well-functioning policy and strategy to meet long-term goals and to become profitable for the future, which is covered under ‘performance’ (2001). Klein adds that also independent, inside and/or executive board members have a positive effect on the accounting and stock market performance of firms (1998).
Corporate boards can have a one- or two-tier structure. Two-tier (dual) boards consist of a management, who are led by the CEO and who are responsible for the strategy of a company, and the board of directors, who are led by the chairman and
11
who are responsible for monitoring the company. One-tier, or unitary (single), boards consist of executive and non-executive directors, thereby tasks are less controlled but decisions are easier made. Companies in the US only use the one-tier board structure, but most European countries use the two-tier structure. Corporates in Norway can use both board structures, but particularly the two-tier structure is most common for PLCs. Norway is therefore an exception, as most countries mandate their corporates one of these structures (Belot et al., 2014; Jungmann, 2006). The two-tier board structure is designed to ensure independency among boards. Boards with a two-tier structure appear to be more efficient, because tasks are divided and afterwards analysed by a second group (Bezemer et al., 2014).
2.1.2. Agency theory
The designed board structures do not solve all the issues a board faces if they fulfil their function. The board of directors has two roles that actually are in conflict with each other. This conflict is labelled as the ‘agency conflict’. Onetto describes this conflict for companies as a conflict of interests between the board, management and shareholders. The concept of the agency theory is that there are two parties that do not strive for the same objectives. To explain, the ‘agent’ has to work in the interest of the principal and the ‘principal’ has to commit to the agent with the right compensations for his or her actions, effort and work. In our case, we have the management as ‘agent’ and the board of directors as ‘principal’, but the board also has an obligation towards the shareholders where it faces the title of ‘agent’, and where the shareholders are the ‘principals’ (2007). The conflict plays a role in many businesses and it is hard for directors to manage opposite roles. The problem of dealing with several stakeholders and various interests is captured in Subsection 2.1.3.
Optimal contracting is one possible solution for the agency conflict, which offers the right incentives and compensation packages to make managers work in the interest of the shareholders (Larraza-Kintana et al., 2007). The ‘behavioural’ agency theory is a modern addition to the original agency theory. Decisions by board members are often based on personal assumptions, which means that human behaviour and individual preferences play an important role in the decision-making process of board members and the management. The behavioural agency theory assumes that behaviour-based incentive packages result in a similar firm performance as the outcome-based package. For example, the outcome-based incentive package
12
(optimal contracting) measures the CEO’s ability based on the returns of the firm and these results determine if CEO’s are rewarded. The behaviour-based incentive package differs in that the board supervises the CEO based on his behaviour, decisions and actions, which are assumed to have a positive effect on firm performance if the CEO has the right level of knowledge and experience (Larraza-Kintana et al., 2007; Wiseman & Gomez-Mejia, 1998). But also board members dare to take on more risks if they are not directly punished for unfavourable short run results. In the long run, risks may have a positive effect on the performance if the right investments are made (March & Sevon, 1988).
2.1.3. Stakeholder theory
The next discussed board-related theory is the stakeholder theory. This theory captures the role of the board to value preferences of all stakeholders. Some stakeholder examples for a company are its employees, shareholders, clients or customers, but also the media and local governments (Clarkson, 1995). Wang and Dewhirst add that there is actually a specific order in which directors build their interest of importance: customers and the government come first, followed by the stockholders, employees and at last the society (1992). The stakeholder theory is often in conflict with the shareholder theory, which ensures shareholders that short run profits are maximized to increase their returns. In contrary, the stakeholder theory tries to maximize the long-term company value, which has as consequence that short run profits are not always as high as desirable. The role of the board is to balance between both the share- and stakeholder theories (Hill & Jones, 1992; Jensen, 2002). The degree in which a director tries to achieve all objectives depends on his or hers functional characteristics. For example, non-CEO directors tend to be more stakeholder-orientated than directors with a CEO function. The same applies for inside and outside directors, as outsiders tend to actively work in the interest of all stakeholders (Wang & Dewhirst, 1992). Despite that the intention of directors that try to achieve all desires of each of the stakeholders sounds ideal, the actual practise to accomplish all objectives is hard. Directors and managers cannot follow one main objective, as they need to achieve several objectives by each of the stakeholders. It is a tough task to balance between all these objectives, what strengthens the concept of the agency theory. This theory only has a single-valued objective that has to be accomplished (Jensen, 2002).
13
2.1.4. Resource dependency theory
Another governance-related theory for the board of directors is the ‘resource dependency theory’. This theory was originally developed by Zahra and Pearce (1989) and later revised by Pfeffer and Salancik (2003). The concept is that companies rely on external resources, which can influence the performance. It is required for companies to build long-term relationships with external parties. The role of the board is to maintain these relationships and to evaluate what steps are necessary in the process of building those. Board members should have the right level of knowledge and skills, as the performance of firms depends on their characteristics and relationships. Background, gender but also a network of contacts can have an influence on a board’s capabilities. The composition of the board can therefore be seen as a strategic composition, which makes it necessary to understand how especially gender influences these characteristics and long-term external relationships. Pfeffer and Salancik mention several benefits of board members related to firm performance: they provide advices, offer easier access to information and resources and they bring their legislation-knowledge (2003). Recent studies indicate that the characteristics and nature of board members are most important during a company’s early life cycle stage and during economically heavy times (Hillman, Withers, & Collins, 2009). At last, the board size is important for the resource dependency theory. Assumed is that larger boards operate more efficient than smaller boards, because external connections to resources are easier accessible in a large group of people (Jawahar & Mclaughlin, 2001).
2.2. Characteristic differences
The debate around the concept of glass ceiling during the ‘80s in the US resulted in that men more often obtained boards functions than women, as it was assumed that women did not have the right qualities for these positions (Adams & Funk, 2012; Burgerss & Tharenou, 2002). Subsections 2.2.1. and 2.2.2. explain what characteristic differences are observed in earlier investigations.
2.2.1. Important differences for board positions
Previous studies provide characteristic differences between men and women, however not all differences are important to fulfil a board function. One of the indicated differences is the rational decision-making process if men and women are in a relational situation. Found is that the ethical standards play a bigger role in
14
considerations and decisions for women than for men, but this gap converges with age and experience to each other. The ethical instinct of women makes them better in building up long-term trust and relationships than men, which connects to concept of the resource dependency theory (Dawson, 1997).
One of the main arguments behind the glass ceiling theory was that women tend to be more risk-averse than men if they hold a board position. The glass ceiling theory was, and still is, an ongoing discussed topic for particularly US corporates. The glass ceiling theory assumes that women are not suitable for board positions, which withholds them from receiving top management and board positions (Goodman et al., 2003). Board members, hence especially female members, that take on too less risks can negatively affect the financial decisions and performance of a company. Also the risky investments, which are crucial for an outstanding performance and potential success, are not considered. But in actual practice there is statistical evidence that women do not tend to take lower risks than men under controlled economic conditions, which is in contradiction to the glass ceiling theory. The same applies for investment and insurance decisions (Schubert et al., 1999). Decisions and thus success are based on the involvement and professional experience of board members, not their gender (Nielsen & Huse, 2010b). The base on which men and women make their decisions perhaps does differ. Adams and Ferreira argue that the attendance of women at board meetings is higher, which offers women the opportunity to obtain necessary information before they make decisions. Men tend to base their decisions on personal experience and knowledge (2009).
There are more characteristic differences that matter for board positions. Leadership and managing skills can differ for men and women, as it is found that women are more open in their interaction with colleagues and their preference for a wider participation during the decision-making process (Fiermani, 1990). Bass and Avolio measured the leadership performance of men and women for six of the Fortune 500 firms based on the four I’s of the Multifactor Leadership Questionnaire (MLQ), which is designed by Avolio, Waldman, and Yammarino (1991). The four factors in the MLQ are idealized influence (charisma), inspirational motivation, intellectual stimulation and individualized consideration. The results indicate that women score higher than men with their idealized influence, their inspirational characters and individual consideration (1994).
15
To conclude, typical gender characteristics do influence the decisions and actions boards undertake. Gender can have an influence on knowledge and behaviour characteristics, which beforehand define if a woman becomes a board member and how women operate if they are. It is complicated to tell how specific characteristics influence the way someone operates and performs, as it is often a combination of characteristics that matters (Huse & Solberg, 2006).
2.2.2. Benefits of boards with women
Although most of the boards worldwide consist of only men and just a small fraction of women, studies suggest that women do have positive influences on actions taken by the boards they are part of. Terjesen et al. demonstrate that boards with female directors are actively involved in decisions that are important for a company’s strategy, such as monitoring and supervising policy implementations and other operations. Moreover, women tend to review their own board performance and they do not solely rely on their networks, but also on experts if they do not have the knowledge themselves. Women also value non-performance measures besides the accounting performance, such as customer satisfaction and organised development programmes. Boards that consist of females could therefore provide new insights in measurements of performance, which could diversify them from their competitors in the long run (2009).
Another benefit is that female directors often bring international diversity into boards, which provides a competitive advantage in the globalizing world of today. Internationalization offers more insights in the monitoring process, as the knowledge and experience of several national and business cultures are taken into consideration. Besides, women often have a MBA degree what makes them highly educated compared to their male competitors (Singh, Terjesen, & Vinnicombe, 2008).
Burke provides several reasons why women still are not chosen as directors. He explains that women are often unqualified for these positions because of their lack of experience and leadership knowledge (1996). The persistent problem of women that are unqualified keeps repeating itself, as women do not receive equal chances to obtain a board position and to thereby gain any experience. This process is described as the ‘career pipeline’, where men and women still have other opportunities in and expectations of their careers (Schweitzer et al., 2011). Another reason is that companies simply do not search for more women to include to their boards or that
16
companies do not know where to find qualified female directors. A last reason is that companies might be afraid to change their board structure if boards previously consisted of only men (Burke, 1996).
So far there has been less evidence that highly qualified women obtained board positions in Norway after the law revision in 2008. There is actually found that women with a board position before the revision are more qualified than women that were hired after the new requirements. It is doubtful that the law truly changed something for women in businesses (Bertrand et al., 2014).
2.3. Diversity and firm performance
Sections 2.1. and 2.2. explained how men and women differ in their characteristics and how this difference could affect the tasks of board members. Even more important is to discuss studies that test for a relationship between diversity and a company’s performance. Earlier investigations point out several relationships between female directors and firm performance. Carter et al. find a positive relationship between women on boards and firm value. Besides, they find that the fraction of female directors increases if the size of a firm increases, thus if the firm shows growth potential for the future. The same result is found for other minorities on boards, such as African Americans, Asians and Hispanics (2003). The positive relationship does not necessary suggest that introducing a women quota for corporates results in an improved firm performance or governance, as there exists no direct relationship between firm performance and a quota (Adams & Ferreira, 2009). Erhardt, Werbel, and Shrader did a similar study with a smaller group of large US companies. They used the percentage of women and minorities (African, Hispanic, Asian and Native Americans) as measure of observable demographic diversity. Their results suggest that a degree of board diversity leads to a higher organizational performance, measured by the return on assets and the return on investments (2003).
In contrary, Ahern and Dittmar found a negative relationship between the mandated number of female directors and firm performance. Tobin’s Q and the share price are used in this study as performance measurements. The paper suggests that the share price of firms with no female director reacts significantly more negative than the prices of firms with at least one female director, if an extra female director is included to the board. One explanation for this value loss is that the hired female directors are less experienced than the exiting male directors. Besides, these female
17
directors are on average 8 years younger (2012). Bøhren and Strøm find a similar negative relationship to firm performance if the number of female board members increases. A provided argument is that a higher level of diversity can negatively affect the decisions-making process, as more different opinions need to be considered what decreases the efficiency (2007).
Several studies were unable to find a significant connection between firm performance and board diversity. Rose measured diversity as the percentage of women, foreigners and people with different educational backgrounds. His paper uses Danish listed companies and firm performance was measured by Tobin’s Q, but no clear relationship was found. The paper still suggests that there should be other reasons for boards to be become more diversified than only because it would be better for their firm performance. For example, a reason should be that women receive the chance to gain board experience and develop themselves (2007). Carter et al. found in another test a positive relationship between the return on assets and gender and ethnic minority diversity for major US corporations, though this relationship became insignificant and neither positive nor negative if Tobin’s Q was used. The paper therefore does not support a law to have a minimum number of women and ethnic minorities on boards, as no actual relationship was found (2010). Table 1 summarizes all the relationships between firm performance and board diversity that are mentioned above in this section. Table 1: Summary of the results of the relationship between firm performance and board diversity.
Author(s) (year) Sample Main result
Adams & Ferreira (2009) 1,939 S&P 500, S&P
MidCaps and S&P SmallCap firms over the period of 1996-2003.
Positive relationship between female directors and value creation. No evidence that a women quota improves firm performance.
Ahern & Dittmar (2012) 248 publicly listed Norwegian firms over the period of 2001-2009.
Negative relationship
between mandated number of women on boards and firm performance (measured by Tobin’s Q and share prices). Bøhren & Strøm (2007) Listed non-financial
Norwegian companies over the period 1989-2002. Total of 2,207 firm-time
observations.
Negative relationship is found between board diversity and firm
performance (measured by Tobin’s Q).
18
The effect of the women quota in Norway can more over depend on the current performance level of companies. The new requirements of the Norwegian law could cause that companies have to deviate from their initial, optimal, performance level as they are obligated to restructure their boards to meet the gender equality requirements. As an example, costs of that company increase temporarily as women often have less experience, which results in a lower performance than their initial performance optimum. Dohmen and Falk explain that the likely risk aversion of women and a lower productivity can both influence the performance of companies. In this case, a change in the board composition after the law revision requires companies to deviate from their optimal, which negatively influences their performance (2011). The other way around, companies that did not achieve their optimal performance yet could possibly reach this level now if more women are involved in their businesses. Discussed in Section 1.1. was that the European Commission declares that the efficiency of the EU financial market may be harmed by gender inequality, as this negatively affects sustainable growth and employment (European Commission Proposal, 2012). The effect of the
Continuation of Table 1 Carter, Simkins, & Simpson
(2003)
638 ‘Fortune 1000’ firms for the year 1997.
Positive relationship between number of women and other minorities on the board and firm value (measured by Tobin’s Q).
Carter, D’Souza, Simkins, & Simpson (2010)
641 S&P 500 over the period of 1998-2002. Gathered from the IRRC database.
Positive significant
relationship between gender and ethnic minority diversity and return on assets, no relationship found with Tobin’s Q.
Erhardt, Werbel, & Shrader (2003)
137 ‘Fortune 1000’ large US companies over the period of 1993-1998.
Positive relationship between board diversity and firm performance. Used measurements of
performance are the return on assets and return on
investment. Rose (2007) Listed Danish firms over the
period of 1998-2001. Total of 443 firm-time observations.
No relationship is found between firm performance (measured by Tobin’s Q) and female board representation. Same for foreigners and educational background.
19
women quota on the performance therefore depends on the current stage in which companies are operating.
The number of female members and therefore a change in the board composition can also indirectly affect the firm performance. The studies of Table 1 focused particularly on the direct effect of female directors on the performance, though their presence can indirectly have an effect as well. Isidro and Sobral explain that female board members can influence a company’s performance through a better financial performance, but more over by the improvement of other business parts that indirectly improve the firm value. The paper suggests that the presence of women indirectly, but positively, influences the non-financial decisions of a company, especially the observance of the ethical and social policies that companies implement. Social and ethical compliance are measured by the consistence of an ethical or social responsibilities-committee within a company. If companies have such organised committees, then it often is the case that a set of well-functioning ethical and social principles is designed, implemented and monitored. The positive effects result in higher firm values, though the firm’s accounting measurements are possibly not even changed. The paper does not find any direct relationship between the presence of women on boards and firm performance (2015). The paper by Isidro and Sobral indicates that the women quota of Norway should result in a higher indirect firm value.
Although different relationships are found in earlier investigations (see Table 1), specific studies with listed Norwegian companies show that there exists a negative relationship between board diversity and firm performance after the quota was introduced (Ahern & Dittmar, 2012; Bøhren & Strøm, 2007). A similar result is therefore expected in this paper. More over, followed by Dohmen and Falk (2011) it is expected that companies deviate from their performance optimum if they are required to reform their boards. New hired female directors and potential costs could push companies away from their initial optimal performance level.
Hypothesis 1: A higher level of gender diversity within the board of directors has a negative influence on the firm performance.
2.4. The critical mass theory
The studies of Section 2.3. indicate that (gender) diversity affects firm performance if the percentage of board diversity increases. In contrary, the critical mass theory assumes that a specific number of female board members has to be reached to obtain
20
an effect on performance. The concept of the critical mass theory is that the performance improves if the number of female board members increases from zero, one or two to at least three (Torchia, Calabrò, & Huse, 2011).
The practical concept of a critical mass is that female directors tend to operate more involved and innovative when they are part of a group of females. The critical mass theory is in line with the ‘tokenism’ theory, which assumes that boards can have a majority of one gender that dominates the minority of the other gender. This minority group of females is labelled as token and in practice they fill up seats without that they have an actual voice in decisions. Being part of this, or actually any, minority creates a feeling of discomfort and self-debt, what often results in a higher pressure to perform better to the majority group. A high performance pressure has unfavourable effects on the actual performance of a minority group (Kanter, 1977a, b). The critical mass in perfectly summarized by Kramer et al., as quoted:
“If there is only one woman, and that woman is a more reticent sort, she may be less willing to speak up and in some way less effective; in addition, the rest of the board (all male) tend to view her as symbolic of all women. If you add one more woman, you start to adjust that balance. If the two women work well together, you can adjust it hugely. If there are at least three women, the likelihood is very good that at least two will work well together and can have impact beyond their numbers.” (2006, p. 22)
Torchia, Calabrò, and Huse indeed investigate a positive relationship between three or more female directors and the company’s level of organizational innovation. The paper cannot find a similar relationship if boards consist of less women (2011). Still to keep in mind, more women on boards do not fix all obstacles boards are facing. For example, a higher mixture of both genders on boards solves the inequality problem between men and women, but it does not change other governance-related problems for corporates (Konrad, Kramer, & Erkut, 2008). Joecks, Pull, and Vetter also find a significant positive relationship for German corporations between three women on the board and firm performance, compared to a non-significant positive relationship to zero or two women. The paper finds similar results for a percentage of women between 20 and 40, as this often results in three women as well. The paper explains that boards must reach the ‘magical number of three’ before they can take full advantage of a diversified
21
board (2013).
Hypotheses 2 and 3 are developed based on Section 2.4. This section indicates that the performance not necessary improves after an increase in female members, but only if a specific number is reached. The critical mass theory assumes that the firm performance increases if more women are included in corporate boards. The European Commission (2012) more over declared that gender equality within corporates results in an efficient EU financial market (Section 1.1.). At last, Isidro and Sobral add that the presence of female board members can indirectly affect the firm value by the observance of ethical and social standards. The indirect effect also indicates that more women on boards improve the total firm performance (Section 2.3.). The second regression in this paper will test if an increase in female board members indeed has a positive effect on firm performance.
Hypothesis 2: An increase in female board members has a positive influence on firm performance.
The critical mass theory assumes that boards must reach a specific number of female members to perform better. Hypothesis 3 is designed to test this theory. The positive expectation is based on Section 2.4. and the proposal of the European Commission (2012).
Hypothesis 3: The change from zero, one or two to at least three female board members has a positive influence on firm performance.
22
3. The Norwegian Gender Equality Act
This chapter explains the structures of Norwegian companies, how the law for gender equality among PLCs was revised and what the requirements of the Companies Act are.
3.1. The steps of the law revision
Norway has two types of limited companies. First, there are private limited companies, also named Aksjeselskap (AS), and secondly there are public limited companies, Allmennaksjeselskap (ASA), which are also labelled as PLCs. ASA companies are much larger than AS companies, as it is required for ASA companies to have a ten times larger minimum amount of capital. ASA companies have the possibility to list their shares on the public stock exchange. Only ASA companies are subjected to the regulations for gender equality among boards, so a company that changes from an ASA to an AS structure does not have meet the requirements any longer. Moreover, Norway has companies with a general partnership structure, labelled as ANS and DA, sole companies and at last companies that are registered or listed abroad, labelled as NUF (Rasmussen & Huse, 2011).
The official change of the law made its introduction with several revisions. During 1999 the Norwegian Gender Equality Act was revised for the first time. After this revision the quota and its political and regulatory aspects were designed, though companies were not required to truly change their boards. After a while, the Norwegian Public Limited Liability Companies Act was developed and ready for its implementation. The Companies Act states five requirements with as main objective to reach gender equality among corporate boards. The requirements are discussed in Section 3.2. The Companies Act is applicable for all PLCs that are listed in Norway on the Oslo Børs Exchange. The Companies Act became official by December 2003, though by that year the Companies Act only suggested companies to voluntarily reform their boards. If companies did not meet the threshold of 40 per cent of female board members by July 2005, then the quota would be enforced as of January 2006 with a two-year transformation period. By July 2005 females managed only 13 per cent of the board seats and as a consequence the law became official and sanctioned. As of January 2008 PLCs have to meet the gender requirements for corporate boards to not become dissolved from the Oslo Børs Exchange. Companies that became
23
public any moment after January 2006 must meet the requirements from the moment they became public (Dale-Olsen, Schøne, & Verner, 2013). Figure 1 pictures the timeline of the key changes between the period of 2003 and 2008.
Figure 1: Timeline of the Companies Act between the period of 2003 - 2008.
3.2. The requirements of the quota
The Companies Act requires that PLCs have at least 40 per cent of both sexes in their boards. The exact number of each of the sexes depends on the total number of board members. The Companies Act states the following requirements for PLCs:
1. If a board consists of two or three member, then at least one member of both of the sexes must be present;
2. If a board consists of four or five members, then at least two members of both of the sexes must be present;
3. If the board consists of six, seven or eight members, then at least three members of both of the sexes must be present;
4. If the board consists of nine members, then at least four members of both of the sexes must be present. If boards have more than nine members, then at least 40 per cent of both sexes must be present;
5. The requirements of 1-4 likewise apply for the elections of deputy board members (Norwegian Public Limited Liability Companies Act, 1997).
The quota requirements of Section 3.2. are used to design the control and treatment groups of Subsection 4.3.1. The results are used to answer the first question and hypothesis 1. December 2003: Companies Act becomes active July 2005: Decision to sanctionate the requirementes January 2006: Start of transformation period January 2008: Companies Act officially in practice
24
4. Methodology
This chapter first discusses the sample selection of this paper. Explained is where and how data is collected. Secondly, all the used dependent and independent variables are described. At last, Section 4.3. describes the three regressions and related components. Section 4.3. also explains how the treatment and control groups are designed.
4.1. Data and sample selection
This paper investigates if there is an observable relationship between the number or percentage of women on the board of directors and the firm performance, with as sample all PLCs in Norway that operate between a specific time-period. The revision in the Norwegian law in 2008 is used as natural experiment to test if the change in the number or percentage of women on boards has an effect. The total sample uses the period between 2005 and 2013. The sample starts with the year 2005, as this was the last year that companies voluntarily could choose how many men and women they preferred for their board seats. 2013 is the last year of the sample and this year provides the possibility to observe long run effects on firm performance. The regressions in this paper cover the period between 2005 and 2009, as the two years before the year of the change (2005/2006) and two years after the year of change (2008/2009) are observed, 2007 is the year of the law revision. The robustness tests cover the periods between 2007 to 2008 and 2005 to 2013, to see if similar results are visible.
The active PLCs in Norway during the sample period are obtained from the Orbit database, with as restriction that companies cannot be incorporated after 2005 and that the companies have to remain listed and active until at least the year 2010. This results in a list of 99 PLCs. The Datastream database is used to gather data on the sexes of board members. Unfortunately, the provided data is very limited, what makes it necessary to hand-collect missing data. This data is collected with the same steps as Ahern and Dittmar (2012). The annual reports are used to decide how many male and female board members each of the companies have during the sample period:
1) First use photographs of board members to decide a gender;
2) If no photographs are available, then search for identifying pronouns as ‘he’ and ‘she’, ‘his’ and ‘hers’;
25
3) If no photographs or identifying pronouns are available, then search for the first names of board member and use the ‘First Names’ database from Statistics Norway to find if names are commonly used for men or women.
The Wharton Research Data Services (WRDS) database is used to gather data on the remaining control variables.
4.2. Variables
4.2.1. Dependent variables
The dependent variable is a measurement of performance. The measurement is the return on assets (ROA) and data from the WRDS database is used for the calculation. Following Ahern & Dittmar (2012) and Guest (2009) the formula for the ratio is:
𝑅𝑂𝐴 =𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑝𝑟𝑜𝑓𝑖𝑡 𝑏𝑒𝑓𝑜𝑟𝑒 𝑑𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛 𝑇𝑜𝑡𝑎𝑙 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑎𝑠𝑠𝑒𝑡𝑠
For the robustness tests two different performance measurements are used, namely the return on investment (ROI), which is directly gathered from the Datastream database, and secondly the return on invested capital (ROIC). Data for the calculation of the ROIC is gathered from the WRDS database. The formula for the ROIC is:
𝑅𝑂𝐼𝐶 =𝑂𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑝𝑟𝑜𝑓𝑖𝑡 𝑏𝑒𝑓𝑜𝑟𝑒 𝑑𝑒𝑝𝑟𝑒𝑐𝑖𝑎𝑡𝑖𝑜𝑛 𝑇𝑜𝑡𝑎𝑙 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑖𝑛𝑣𝑒𝑠𝑡𝑒𝑑 𝑐𝑎𝑝𝑖𝑡𝑎𝑙 4.2.2. Independent variables
The independent variable of interest is different for all three regressions. For the first and third regressions the interaction term is important and for the second regression the women variable. Detailed information is provided in Subsections 4.3.1. – 4.3.3.
4.2.3. Control variables
Several control variables are included in this paper and the inclusion is based on earlier studies.
1. The board size: the total number of board members, consisting of both men and women. Guest (2009) found that there exists a negative relationship between the size of boards and firm performance for large firms, as large firms tend to have larger boards compared to small firms. Eisenberg, Sundgren, and Wells find a similar negative relationship between board size and small to medium sized
26
firms (1998). Jawahar and Mclaughlin (2001) in contrary find a positive relationship, as larger boards tend to operate more efficient due to the resource dependency theory. The board size variable is gathered from the Datastream database and completed by hand with annual reports.
2. The number of employees: this variable contains the total number of employees in log. The inclusion of this variable is in line with the similar research by Dale-Olsen, Schøne, and Verner (2013). The variable is gathered from the WRDS database and completed by hand with annual reports. The number of employees is used as measure of firm size.
3. Total value of assets: this variable contains the total asset value as measure of firm size. Dale-Olsen, Schøne, and Verner (2013) use the lagged log value of the total assets and this paper will therefore too. Data on total assets for 2004 is gathered to create the lagged variable.
4. Industry specific variable: this variable is the four-digit code of the Standard Industrial Classification (SIC). This dummy variable is gathered from the WRDS database and this variable is used as industry fixed effect (FE).
5. Region specific variable: this variable consists of fourteen dummies for each of the used regions in this study. The list of regions in Norway can be found in Appendix 2. The information is gathered from Bureau van Dijk within the WRDS database, which provides the names of cities where the companies have their headquarter. By hand all regions are determined, based on the cities and locations. Regions are used as FE.
4.3. Types of regressions
In this section all three regressions are described. Each test has a different regression, as their main independent variable differs. Standard errors are clustered at the firm level for each of the regressions.
4.3.1. First regression
The first regression tests if the control and treatment groups differ in their performances before and after Companies Act became active. For this test two groups are designed:
1. Treatment group: this group of companies had to change their board composition, as these companies did not meet the requirements of the Companies Act before the law became active in 2008. These companies had to include more women to their boards if the board size would remain unchanged.
27
2. Control group: this group of companies did not have to change their board composition, as these companies already met the requirements of the Companies Act before the law became active in 2008.
Table 2 shows the regression models of the first test. Table 5 summarizes the description of each of the used variables in the models.
Table 2: Regression models of the first test.
Model 1: ROAit = β0 + β1*Changei + β2*Changei*Lawt + FEi + FEt + 𝜀it
Model 2: ROAit = β0 + β1*Changei + β2*Changei*Lawt + β3*Employeesit + FEi +
FEt + 𝜀it
Model 3: ROAit = β0 + β1*Changei + β2*Changei*Lawt + β3*Employeesit +
β4*Board Sizeit + FEi + FEt + 𝜀it
Model 4: ROAit = β0 + β1*Changei + β2*Changei*Lawt + β3*Employeesit +
β4*Board Sizeit + β5*Assetsit + FEi + FEt + 𝜀it
4.3.2. Second regression
The second test measures the relationship between an extra female director and the firm performance. In this regression all firms are included, also if they did not meet the requirements of the Companies Act. This regression observes whether there exists a general relationship between female directors and the performance of firms. Table 3 shows the regression models of the second test. Table 5 shows the description of each of the used variables in the models.
Table 3: Regression models of the second test.
Model 1: ROAit = β0 + β1*Womenit + FEi + FEt + 𝜀it
Model 2: ROAit = β0 + β1*Womenit + β2*Employeesit + FEi + FEt + 𝜀it
Model 3: ROAit = β0 + β1*Womenit + β2*Employeesit + β3*Board Sizeit + FEi +
FEt + 𝜀it
Model 4: ROAit = β0 + β1*Womenit + β2*Employeesit + β3*Board Sizeit +
28
4.3.3. Third regression
The critical mass test uses two different groups for the DiD analysis. The time component again uses the period before the Companies Act was enforced, the period between 2005 and 2007, and the period after the law became enforced, from 2008 and onwards. The treatment component differs.
1. Treatment group: this group of companies changed from zero, one or two to three or more female board members after the law became active. If companies in the treatment group changed to a critical mass for only some of the years after the law became active, then only the years that the company met the critical mass are used.
2. Control group: this group of companies did no reach a critical mass after the law became active. The control group still consists of zero, one or two female board members after the law. Companies that already had and kept a critical mass before and after the law became active are deleted from the sample.
Table 4 shows the regression models of the third test. Table 5 shows the description of each of the used variables in the models.
Table 4: Regression models of third test.
Model 1: ROAit = β0 + β1*Criticali + β2*Criticali*Lawt + FEi + FEt + 𝜀it
Model 2: ROAit = β0 + β1*Criticali + β2*Criticali*Lawt + β3*Employeesit + FEi +
FEt + 𝜀it
Model 3: ROAit = β0 + β1*Criticali + β2*Criticali*Lawt + β3*Employeesit +
β4*Board Sizeit + FEi + FEt + 𝜀it
Model 4: ROAit = β0 + β1*Criticali + β2*Criticali*Lawt + β3*Employeesit +
β4*Board Sizeit + β5*Assetsit + FEi + FEt + 𝜀it
Table 5: Description of variables used in the first, second and third regression. ROA: Return on Assets
Change: 1 if companies did not meet the requirements, 0 otherwise Law: 1 for the year 2008 and onwards (revision), 0 otherwise Women: Inclusion of 1 female director
Critical: 1 if companies changed to a critical mass, 0 otherwise
29
Continuation of Table 5
Critical*Law: Interaction variable between the critical mass and period after revision Employees: Total number of employees in log
Board Size: Number of board members
Assets: Total value of assets, lagged and in log FE: Fixed effects – year, region and industry
30
5. Results
In this chapter presents the results of all tests. First the descriptive statistics are discussed, followed by several graphs and the matched analysis of Kelner. At last, the results of the three regressions are presented and discussed.
5.1. Descriptive statistics
Table 6 shows the number of companies for each of the observed years. This paper requires that companies at least must operate between the period of 2005 and 2010. Table 6 therefore shows that the number of companies decreases after 2010, starting from 99 and decreasing to 86 companies in 2013. All the company names that are used in this sample are reported in Appendix 1. The average board size remains constant, although a small increase from 6.4 members in 2005 towards 6.8 members by 2013 is visible. The proportion of female board members almost doubled between the period of 2005 and 2013 from 21.5 per cent to 41.3 per cent. In the bottom part of Table 6 there is a separation for the number of female board members and its percentage as part to the total board size. The table shows that the percentage of companies with none female board members decreases to zero from 2007 and onwards, while the percentage of boards with at least three women strongly increases from 16.2 per cent in 2005 to almost 55 per cent in 2013.
Table 6: Percentages of women on the board of directors for each of the observed years.
Year 2005 2006 2007 2008 2009 2010 2011 2012 2013
Number of companies 99 99 99 99 99 99 95 90 86
Board size 6.4 6.7 6.6 6.7 6.6 6.7 6.7 6.9 6.8
% of women on boards 21.5 29.2 40.0 40.5 40.5 41.0 41.5 40.5 41.3
% firms with 0 women
on the board 23.2 9.1 0.0 0.0 0.0 0.0 0.0 0.0 0.0
% firms with 1 woman
on the board 31.3 23.2 3.0 3.0 4.0 4.0 9.5 5.6 5.8
% firms with 2 women
on the board 29.3 36.4 45.5 43.4 47.5 44.4 33.7 40.0 39.5
% firms with at least 3
women on the board 16.2 31.3 51.5 53.5 48.5 51.5 56.8 54.4 54.7
Table 7 shows the descriptive statistics of each of the components of the regressions. The ROA, ROI and ROIC are in decimals and must be multiplied by 100 to obtain a return in percentage. The three the measurements of performance are winsorized at 0.5
31
percentile on both sides for potential outliers. Table 7 shows that the measurements differ and that the mean of ROA is between the means of the ROI and ROIC (.0056<.0276<.0926). The large difference in the average returns can result in inconsistent evidence for the robustness tests, as the measurements are not perfectly similar. The ‘Change*Law’ variable is a dummy that takes the value of 1 if it contains the years 2008 and onwards and if companies had to change their board compositions. The ‘Critical*Law’ variable is a dummy that takes the value of 1 if it contains the years 2008 and onwards and if companies changed to a critical mass. The women variable shows the total number of female board members. This variable has a maximum of five females during the entire sample and the average number is half of the maximum, 2.5 women. The number of employees is measured in log and winsorized at 0.5 percentile on both sides. The number of employees provides an average of almost 6.2. The board size has an average value of 6.67 rounded and has a minimum of 2 and maximum of 12 members. The value of total assets is measured in lagged log and winsorized at 0.5 percentile on both sides, which results in an average value of 7.4.
Table 7: Descriptive statistics of variables.
Variable N Mean SD Min Max
ROA 865 0.0276 0.2962 -2.3161 0.5567 ROI 815 0.0056 0.3838 -2.912 0.8917 ROIC 863 0.0926 0.4511 -2.8935 2.3091 Change*Law 865 0.1595 0.3664 0 1 Critical*Law 757 0.2708 0.4447 0 1 Women 865 2.4948 1.0727 0 5 Employees 861 6.1713 2.2092 0 10.74193 Board Size 865 6.6659 1.9408 2 12 Total Assets 861 7.442 2.2214 2.2341 14.4208
Figure 2 presents the types of industries and the percentage of companies in each of these industries. Nine dummies are created for each the industries that are included in this sample. The companies are grouped by the sector they are operating in, based on their SIC Code. Most companies are operating in the manufacturing industry; this group represents almost 40 per cent of all companies. The sectors of public administration, retail trade and agriculture, forestry and fishing each represent only one company and
32
these are therefore the last common industries in this sample. Detailed information about the types of industries is provided in Appendix 3.
Figure 2: Types of industries and the percentage of companies within these industries.
Figures 3 and 4 show graphs for all of the companies for each of the years between the period of 2005 and 2013. In figure 3, we see that the average number of women per board in 2005 doubles to almost 3 by 2007. This average remains constant from 2007 and onwards. The total number of board members does not reveal a major change. The board size is stable between 6 and 7 members. Figure 4 presents the difference in the proportion (in percentage) of female board members between the treatment and control group for the first regression. The control group shows a stable ratio of females per board between 40 and 45 per cent for each of the years. The treatment group in contrary shows a large increase in the proportion between the years 2005 and 2007, with a rise of 15 to almost 40 per cent. This increase makes sense, as the treatment group did not meet the requirements of the Companies Act and this group thus had to reform. Over the period between 2007 and 2013 the proportion of women remains stable around 40 per cent.
0 5 10 15 20 25 30 35 40
Manufacturing Transportation & Public Utilities Mining Finance, Insurance & Real Estate Services Construction Agriculture, Forestry & Fishing Public Administration Retail Trade
Percentage of Companies
Types of Industries
33
Figure 3: The number of women on the board of directors. This figure presents the average number of (female) board members for each of the years between 2005 - 2013. The board size is graphed as the dark grey line and the number of female directors as the light grey line.
Figure 4: The proportion of women on the board of directors. This figure presents the proportion of female board members for each of the years between 2005 - 2013. The treatment group is graphed as the dark grey line and the control group as the light grey line. Figure 4 is related to the first regression in this paper.
1 2 3 4 5 6 7 2005 2006 2007 2008 2009 2010 2011 2012 2013 M em be rs
Number of Women/Board
Board Size Women 15 20 25 30 35 40 45 50 2005 2006 2007 2008 2009 2010 2011 2012 2013 In %Proportion of Women on Boards
Treatment Control
34
5.2. Kernel Density estimation
The Kernel Density estimation compares the control and treatment group of the first regression for the years 2005, 2006 and 2007, hence the period before the Companies Act became active. Table 8 reports the average characteristics of each of the variables. The statistics in Column 1 correspond to firms that did not have to change their board composition to meet the law and the statistics in Column 2 correspond to the group of firms that had to change their board composition. Column 3 reports the difference between Column 1 and 2, Column 4 reports the p-value of the difference between the two samples. Table 8 shows that the two groups of firms are similar in their ROA’s, board sizes, number of employees and value of assets, hence the groups do not significantly differ from each other. The control and treatment groups are well formed and useable for the first regression.
Table 8: The Kernel Density estimation for the treatment and control groups. This table reports the Kernel Density estimation for the variables of ROA, the board size, number of employees and total assets between the treatment and control groups. Column 1 reports the mean of the control group before the law was enforced. Column 2 reports the mean of the treatment group before the law was enforced. Column 3 reports the difference between columns 1 and 2. Column 4 reports the p-value of the difference between the treatment and the control group. Variables (1) Control (2) Treatment (3) Diff. (4) p-value ROA 0.041 0.037 -0.003 0.9250 Board Size 6.371 6.408 0.037 0.8596 Employees 5.866 5.717 -0.149 0.5566 Assets 6.845 6.759 -0.086 0.7082
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 5.3. Results first regression
Is there any relationship between the percentage of female directors and the performance of firms? Section 5.3. investigates if any relationship exists with a DiD analysis. In Chapter 2 it became clear that earlier studies suggest several relationships between board diversity and firm performance. The first test will examine if the firm performance improved after boards included more female directors, caused by the requirements of the Companies Act. The results are presented in Table 9.
The ROA decreases if companies restructure their boards. Column 1 reports the DiD variables of the companies that are in the treatment group (Change) before the law
