Gender equality in the board room : a critical assessment of the consequences of the gender quota in Norway

Amsterdam Business School

Gender equality in the board room: a critical assessment of the consequences of the gender quota in Norway

Name: Nicole van Schaijk

Student number: 10891811

Thesis supervisor: dr. S.W. Bissessur Second supervisor: dr. ir. S.P. van Triest

Version: Final Version

Date: 21 January 2016

Word count: 13020

MSc Accountancy & Control, specialization Accountancy Amsterdam Business School, University of Amsterdam

Statement of Originality

This document is written by N.H.T. van Schaijk 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.

Table of Contents

1.4 Structure of the thesis 10

2 Literature Review and Hypothesis Development 11

2.1 Monitoring and Corporate Governance 11

2.2 Corporate Governance and Diversity 14

2.3 Regulation and Corporate Governance 16

2.4 Hypotheses 19 2.4.1 CEO turnover 19 2.4.2 Board interlock 20 3 Research Methodology 22 3.1 Sample Selection 22 3.2 Variables 24

3.2.1 Early versus Late adopters 24

3.2.2 CEO turnover 25

3.2.3 Board interlock 25

3.2.4 Control variables 25

3.2.5 Regression 28

4 Tests and Results 30

4.1 Summary statistics 30

4.2 Results 32

4.2.1 CEO turnover 32

5 Concluding remarks 37

5.1 Conclusion 37

5.2 Limitations 38

5.3 Future research 39

Abstract

Lately, the trust in the quality of financial reporting has been lost as a result of multiple scandals and the economic crisis. Both the regulatory and institutional environments have an impact on this phenomenon and require change in order to regain that trust. Standard setters have made a start in improving the financial reporting quality and changes to corporate governance practices are made that are intended to serve the same purpose. In 2003 Norway has implemented a gender quota that requires 40% of the directors on the board to be female. This is one means to improving the firm’s corporate governance and to break the glass ceiling preventing women from reaching top business positions. Using a sample of companies listed on the Oslo Børs, a regression model is employed to test the hypotheses that are developed to research the effect of the gender quota on corporate governance issues, specifically CEO turnover and board interlocks. The empirical findings are that compliance to the quota in the last possible year (2008) has a positive effect on the frequency of CEO turnover and a negative effect on the number of board interlocks.

1

Introduction

1.1

Research topic

Today the financial news is dominated by publications on accounting scandals, the economic crisis and the lack of trust in the financial reporting quality. The standard setters, politicians, regulators, auditors and businesses are working on the reestablishment of trust in among others the financial reporting quality and general economy. In order to achieve this, businesses must provide the users of the financial statements with high quality. Financial statements are considered to be of high quality when providing timely, relevant and reliable information (Barth, Landsman, & Lang, 2008)

The determinants and effects of the quality of this information, i.e. financial reporting quality, have been widely researched. An important notion that is addressed when assessing financial reporting quality is “transparency”, which can be defined as the extent to which financial reports reveal an entity's true underlying economics, including but not limiting to the risks it faces and how the entity manages them, in a way that is readily understandable by those using the financial statements (Barth & Schipper, Financial Reporting Transparency, 2008).

Standard setters like the IASB and FASB provide a conceptual framework that among others addresses the notion of financial reporting transparency and is developed to improve financial reporting quality on a global level.

However, financial reporting quality is not only influenced by standard setters, and prior research has noted that more weight should be given to institutional influences on the financial reporting quality, as opposed to standard setting (Ball, Robin, & Shuang Wu, 2003). The environment in which the standard setting and financial reporting takes place is called the institutional environment. Preparers’ (e.g. managers, auditors) financial reporting incentives depend on the sources of demand for, and political influence on, financial reporting (Ball, Robin, & Shuang Wu, 2003).

The institutional environment one operates in is affected by the systems of laws, regulation, norms and the enforcement of these regulations, which all mold the socio-economic activity and behavior. The institutional environment can be defined as the set of economic, social, political and legal conventions that establish the basis for the course of business (Dickson, 2004). According to Ball et al. (2003), quality is what a country’s institutional environment demands. Hence, the trust in the financial reporting quality cannot be reestablished by merely mandating and changing accounting standards, since that would also imply that financial reporting quality among countries that mandate e.g. IFRS would be similar, which, as proven by Ball et al. (2003), is not the case. Thus, the institutional environment has a huge impact on the financial reporting quality and in order to increase the quality and trust in the financial reporting, changes should be made to that institutional environment for it to be strong enough to have an effect on the financial reporting quality.

As described above, the need to improve financial reporting quality also necessitates the strengthening of the control of management by setting up good governance structures and intensifying law enforcement, as both these notions make part of the institutional environment. The Board of Directors is at the heart of corporate governance, as they are selected by shareholders to monitor and control decisions made by management in developing business strategies and setting policy objectives. Also, the board is responsible for monitoring the integrity of the firm’s financial statements (Petra, 2007).

The firm’s Board of Directors can be structured in many different forms to suit the type of organization. They ensure effective planning and resource management through regular meetings and good monitoring and control (Arfken, Bellar, & Helms, 2004). In order to be the most effective monitors, boards should be composed of professionals who bring skill, diversity and experience to a company to complement the other directors (Hillman & Dalziel, Boards of Directors and Firm Performance: Integrating Agency and Resource Dependence Perspectives, 2003), i.e. diversity in among others age, background, ethnicity and gender. Zooming in on the last aspect, having more women on the board who bring a specific set of skills, experience and mind set, enables the board to be more effective monitors of management and hence contribute, improve the financial reporting quality.

Prior literature has shown that gender diversity in the board decreases the likelihood of financial restatement, since female board members enhance the ability of the board to monitor financial reporting, maintain independence, promote individual thinking and keep a questioning mindset that prove fruitful for the quality of financial reporting (Abbott, Parker, & Presley, 2012). Independent board members are believed to be better monitors and as Carter, Simkins & Simpson (2003) cite, one argument is that diversified boards increase board independence, since they provide differences in e.g. gender, background. This in turn might probe them to question more decisions of management as would directors with a traditional background.

This thesis will examine the effect of board gender diversity on financial reporting quality in Norway for the period of 2000-2011.

1.2

Gender diversity

Recently, significant progress has been made in Europe on gender diversity. According to a research conducted by Egon Zehnder (2014) only 7,6% of the European boards included no female members in 2014, which is a large improvement compared to 2006 when 32,3% of the boards had no female representation. Also, 31,8% of the new board appointments were going to women in 2014 and women make up for 20,3% of the board positions. In the US this percentage is slightly higher at 21,2% and 18,3% of the women in Canada have board positions in the public companies (Egon Zehnder, 2014). The rise of number of women breaking the so-called glass ceiling did not happen overnight and needed some help from the politics by means of legislation, in this Norway has taken the lead.

In 2003 Norway adopted a law requiring approximately 40% women on the board of directors of public limited companies1_{. The reaction of the corporate sector at that time was at the}

time summed up by Ingunn Yssen, Director for Gender Equality (2003) as:

“The corporate sector is not against an increased number of women, but they don’t like to be deprived of the right to choose men.”

In January 2006 the law became compulsory, since most firms did not comply voluntarily. Firms that did not comply by January 2008 would be dissolved. This lead to a number of firms switching corporate statuses or incorporated in another country, however the ones that remained complied and by 2007 the percentage of female board members reached 40%. Norway took the lead in working towards the goal of more gender equality and not much later among others Iceland, France and Finland followed. Similarly, Great Britain and Sweden have threatened to impose quotas if firms do not appoint more female board members voluntarily.

Recently, the European Union has adopted a similar proposal to that of Norway stating that by 2020 at least 40% of the members of a board of directors should be from the under-represented sex. In practice, this proposal obliges public companies to have at least 40% female members in the board of directors by 2020. In Belgium, a similar law has been accepted in 2011 that by 2017 at least 1/3 of the members of the board of directors should be from the under-represented sex. Both the directive of the EU and the law in Belgium reveal a pressure towards more gender equality in board of directors, which calls for more information on the implementation and consequences of these quotas as the companies and public are, despite previous research on the impact of gender diversity, still divided in two opposite fronts, the opponents and the proponents. In order for us to find sufficient evidence on this subject a closer look should be taken at Norway, the leader on the road to board diversity.

If discrimination is the source of gender inequality then quotas can help businesses overcome that prejudice by forcing more exposure to talented women in authoritative positions. Beaman et al. (2009) found that once the public learns that women can lead effectively, gender

1 _{Specifically, if a firm has two or three members, both sexes should be represented; four or five members, both sexes must have two}

representatives from each sex; six to eight members, both sexes must have three representatives of each sex; nine members must have four representatives of each sex; and more than nine members must have 40% of each sex.

bias will diminish. Public policy (quota) has proven to have a role in influencing the businesses’ belief systems and prejudice. However, what are the implications for implementing such a quota and what are the consequences?

Previous research has focused on the impact of gender diversity in the board on stock market valuation, accounting performance, the behavior of the board and their decision-making process (among others Ahern and Dittmar, 2012 and Matsa and Miller, 2013). Some researches as indicated above have already been conducted in Norway and mainly focus on the effects of the gender quota on firm performance. The results are providing a mixed picture (Ahern, Kenneth, & Dittmar, 2012). This paper will add value to the research base on the effect of law enforcement on corporate governance and ultimately financial reporting quality by operationalization of this concept as described below by introducing two different questions of which the answers are also of worth to the European politics and businesses.

1.3

Research question

The purpose of this research is to examine further the implementation and consequences of the gender quota in Norway. The first question relates to the effect early versus late adoption of the gender quota has on the frequency of CEO turnover. The second and last question addresses the effect of the early versus late adoption of the gender quota on the degree of interlocking between directors. As characteristics of the firms that replaced male board members with female board members in the early stage of the gender quota. The early adopters are the firms that have adopted the quota voluntarily, whereas the late adopters are the ones that adhered to the quota after it was mandated.

As indicated before, this paper contributes to two fields of research, namely corporate governance and political mandates. It provides new evidence on the impact of mandated quotas on two phenomena known to have a big effect on financial reporting quality and firm performance. Also it provides insight to which firms are hesitant to obey to the quota and which companies are not. It is fruitful for the countries that have to adhere to the European legislation to have more information on these developments.

1.4

Structure of the thesis

First I will elaborate on the relevant literature and the hypothesis development. In section 3 the method of measurement of the variables of interest will be carefully explained as well as the sample selection. The fourth section reports on the research methodology, followed by an analysis of the empirical results in the fifth section. The last section includes the most important results and conclusion of this study.

2

Literature Review and Hypothesis Development

2.1

Monitoring and Corporate Governance

The IASB as standard setter has achieved global adoption of the IFRS standards, including the members of the EU that have mandated adoption of IFRS as of fiscal year 2005. Prior literature has generally found a positive impact of IFRS adoption on financial reporting quality. However, taking into account the recent scandals and impact on trust in the financial reporting quality indicates that more factors play a role in this phenomenon. This is confirmed by a study of (Soderstrom & Sun, 2007), who found that financial reporting quality is largely dependent on the institutional environment of a country, including its legal and political system, financial reporting incentives and corporate governance practices. Cohen et al. (2004) refer to this as the corporate governance mosaic as visualized in figure 1.

This is also noted by (Farber, 2005), who investigated changes in corporate governance mechanisms following the detection of fraud, which led to the similar conclusion that the credibility of the financial reporting system and quality of these corporate governance mechanisms are interlinked. Consistent with prior studies, Farber (2005) found that fraudulent firms had poor governance compared to non-fraud firms. In other words, poor governance makes for poor decisions and poor financial reporting quality. These corporate governance mechanisms are intended to oversee managerial actions and reduce agency costs.

As shown in figure 1 the board has influence on not only management, but also on the external auditor and is responsible for establishing an audit committee, that subsequently discharges three oversight roles, namely financial reporting, oversight of external auditors, and oversight of internal controls (Cohen, Krishnamoorthy, & Wright, 2004). Together these facets of the organization, as illustrated below, have an impact on the quality of the financial reporting.

Figure 1

It is argued that important decisions are often initiated at committee level. These committees provide the board a means and structure for effective governance by facilitating special tasks and addressing important corporate concerns (Carter, D'Souza, Simkins, & Simpson, 2010). The establishment of committees enables the members to focus more on these particular tasks and be more effective in the execution thereof. Various characteristics of the board and audit committee and their influence on the effectiveness of the corporate governance mechanisms have been previously researched. It is for instance shown by Vafeas (2001) that the members serving on the audit committee are relatively junior board members, whose effectiveness is greatly dependent on the level of expertise and knowledge of financial reporting. However, prior research results imply that the audit committee should be vested by a board with more power and sufficient expertise to be an effective monitor of management and have a positive influence on the financial reporting quality (Carter, Simkins, & Simpson, 2003).

The external auditor plays an essential role in the promotion of financial reporting quality. It is shown that independent and more active boards demand a higher quality auditor who has proven to be more effective in limiting management to engage in earnings management and the provision of better financial reporting quality (Cohen, Krishnamoorthy, & Wright, 2004). The

Audit Committee Board of Directors

Internal Auditors External Auditors Management

Financial Reporting Quality

Courts & Legal System Financial Analysts Legislators Regulators Stock Exchanges Stockholders

number of outside directors on the board is also positively related to the quality of the audit and hence more effective monitoring of management (Beasley & Petroni, 2001).

Prior research shows that Chief Executive Officers and the management team affect the quality of financial reporting as well as the investment decisions that can either optimize or destruct shareholder’s value (Goel & Thakor, 2008). The Board of Directors is installed to hire and fire management, but also monitor their decisions and actions on behalf of the shareholders, hence, when this mechanism is performing in a poor manner, the reliance of shareholders on both the quality and control is misplaced as corporate governance is believed to have an impact on the precision and quality of information provided. The theoretical underpinning of this concept is related to the agency theory, which describes the relationship between principals (shareholders) and agents (management) from which a conflict of interest arises due to the separation of ownership and control (Hillman & Dalziel, 2003). In the absence of a good monitoring system, agents are inclined to pursue their own interests at the expense of those of the principals, hence creating agency costs and reducing the financial reporting quality, when engaging in fraudulent activities and/or lacking the incentive to fully disclose on the true underlying economics.

The board is dependent upon the information that is provided to them by the CEO and is in turn responsible for assessing the precision of this information. Given the number of corporate fraud incidents the accurate provision of information is however not always guaranteed. The emphasis on accurate information by for instance the Sarbanes Oxley Act is an attempt to improve the quality of the information provided by executives to boards and investors (Song & Thakor, 2006). Assessing and monitoring the effectiveness of the internal control environment are essential parts of the responsibilities of the Board of Directors, because this tightens the control on information provision and hence would improve the quality of this information. However, when the Board of Directors and CEO are not independent of on another, above mechanism does not guarantee the accuracy and transparency of the information at all (Song & Thakor, 2006). The effectiveness of the board is believed to depend on its independence, since in case of sufficient independence the board should be effective in its role of monitoring, hiring and firing the CEO (Hermalin, 2005).

The most effective monitors are believed to be diverse boards that are composed of professionals with various points of view, experiences, knowledge and leadership skills that are

necessary to align management’s interests and perform effective monitoring (Hillman & Dalziel, 2003). It is suggested that gender diversity is one of the aspects that increases the effectiveness of monitoring, which subsequently increases the financial reporting quality and hopefully helps restore the trust therein.

2.2

Corporate Governance and Diversity

The well-known resource dependence theory refers to the ability of board members to bring different resources to the board and highlights the four types of benefits that can come from boards: (1) advice and counsel, (2) communication and information channel for both external as internal, (3) access to commitments or support from important external elements, and (4) legitimacy (Pfeffer & Salancik, 2003). The agency theory on the other hand refers to the monitoring function of the board of directors as described earlier. Expertise, experience and network of the board member are found to be positively linked to the advising and counseling role of the member (Matsa & Miller, 2013), whereas an independent outsider member is assumed to be a better monitor (Adams & Ferreira, 2009). This broad spectrum of activities and responsibilities that is supposed to be carried out by one organ, the board of directors, implies the need for multiple and diverse types of directors in order for the board to be successful in both the monitoring as well as the consulting part of the job (Hillman, Nicholson, & Shropshire, 2008). Multiple identities enable the board to be flexible in adapting to non-routine situations, bring more experience which increases legitimacy and the availability of information, in other words, diversity improves the provision of resources of the board. In its turn, the flexibility of the board also enables the board to view their monitoring activities from different perspectives, which can lead to an improvement of the board’s overall monitoring function (Hillman, Nicholson, & Shropshire, 2008) and hence an improvement of the corporate governance practice within a firm. Carter et al. (2003) show that the relationship between board diversity and shareholder value creation is essential in conducting good corporate governance. Additionally, they find that having minorities on the board is positively related to the number of outsider board members, which has appeared to be an improvement of the monitoring role of the board. The board of directors as a group combines a mix of skills and knowledge which collectively represents a pool of social capital for the corporation. The notion of diversity within corporate governance lies in the board composition and the varied combination of attributes, characteristics and expertise added by each

individual board member in relation to decision making and general board process. Diversity in the boardroom hence refers to the mix of human social capital. Social capital represents skills that an individual has acquired in the course of training and experience. The social capital that is contributed by the board is a measure of the value added by that board in executing its governance function. Also, the diversity within a board reflects the structure of the society as a whole and firms can benefit from this diversity in obtaining and using the right perspectives that are required by the current business environment. In the widest sense, diversity in the boardroom is represented by differences in age, gender, ethnicity, culture, religion, independence, experience, skills, knowledge, expertise etc. (van der Walt & Ingley, 2003)

Gender diversity in the boardroom, and the lack thereof, is often discussed and has gained momentum during recent years. According to Terjesen et al. (2009) recent studies have revealed that boards that include more female members perform significantly different. For instance, 94% of boards with two or three women have proven to explicitly monitor the implementation of corporate strategy. This is in contrast to only 66% of all-male boards. Also, boards with female members are likely to ensure more effective communication among the board and its stakeholders. In the presence of female directors it has also become evident that male directors change their language and moderate their masculinity, which led to better governance (Terjesen, Sealy, & Singh, 2009). By including women in the board an organizational and governance advantage is gained. As stated by Seierstad and Opsahl (2011), women are an under-used pool of talent, have a distinct set of perspectives, a broader point of view, and extended legitimization bases. Which as previously discussed is needed for the board to successfully fulfil all roles (i.e. monitoring and advising roles) this governing body is expected to fulfil. Female board members have proven to provide greater oversight and monitoring of management’s strategies and actions through the promotion of better board attendance, occupying seats on the auditing and corporate governance committees and requiring management to be accounted for poor performance (Gul, Srinidhi, & Ng, 2011). Research has shown that by appointing more female directors and establishing a gender-diverse board, this board could act as a substitute mechanism for otherwise weak corporate governance practices (Gul, Srinidhi, & Ng, 2011). Summarizing, the inclusion of female directors is believed to lead to better quality corporate governance.

2.3

Regulation and Corporate Governance

According to Denis and McConnell (2003) a country’s legal system has a large effect on the governance structures that are adopted by the companies, and on the effectiveness of those governance systems. This is in line with the results of a study conducted by Ball et al. (2003), who show that financial reporting quality is largely dependent on the preparer incentives. Preparer incentives are determined by the surrounding political and market forces, as also described in the governance mosaic and displayed in figure 1. However, an overload of regulatory requirements can lead to a suboptimal system (Roe, 1991). This is confirmed by the research conducted by Larcker et al. (2011), who show that certain regulations that impacted governance practices of the affected firms had a negative performance outcome. In other words, the corporate governance system is, as mentioned before, dependent on market and political forces. Hence, when the political forces are too strong and binding, this may impede the system to be influenced by the market forces. The number of political constraints that are to be placed on a company may have its limits and part should be left to the market for corporate governance to be optimal. It is argued that the market provides management with incentives to change because of the penalty that is laid upon firms with poor governance and performance. However, for the market to provide that incentive, there should be mechanisms in place to facilitate change. The recent corporate scandals have placed doubt in the minds of investors and the public that companies and markets can be self-regulatory and hence change needs to be brought upon by introducing regulations that are meant to improve corporate governance and restore the trust therein.

As set out by Lipton and Lorsch (1992) boards should be given the opportunity to develop and adopt means to effectively improve corporate governance on their own initiative instead of this being imposed by legislation. A proposal that can be easily and more important, voluntarily implemented should be provided to these boards. In 2003 the Norwegian government put such a voluntary proposal was forward where boards were left the choice of selecting 40% female directors to diminish the gender inequality and improve the effectiveness of corporate governance. Early adoption of this proposal should then, in relation to the above researches, have a positive impact on the corporate governance practices, as it is left to market forces, in other words voluntarily adopted by the firms, which could signal quality and eagerness to improving

corporate governance and inequality issues. In 2006 the proposal however was adapted to be mandatory, since too less firms acted on the proposal voluntarily. The adoption of the mandatory regulation is hence believed to have a less positive effect on the late adoption firms’ governance and performance than the voluntary adoption had on those early adopting firms’ practices.

In this study, the focus lies upon the consequences of the early versus late adoption of this regulation. Following the mandating of the gender quota in Norway in 2006, a number of studies have been published, exploiting this natural experiment to analyze the effects of the enforcement of the gender quota that was introduced to create more equality between the sexes and improve the corporate governance system.

Firms who were subject to the new legislation that mandated 40% female members in the board were hesitant at first. They were hesitant since the quota took away their freedom in selecting the board members as previously quoted. Hence it is shown that the disruption of the market forces was clearly noted by the companies of subject, but was this legislation an overkill and did it lead to a suboptimal system or was it a mechanism that was put in place at the right time?

The prior literature examined the effects on firm performance as well as the effects on corporate governance. Norwegian boards consist of average five to six members and if the Norwegian firm has over 200 employees, these employees have the right to elect one-third of the board members. The quota rules are separately applicable to each group of members, being either elected by the employees or shareholders. Even though part of the board is elected by employees, which might enlarge the possible agency conflicts within a firm, the corporate governance in Norway is indicated to be strong compared to other countries (Ahern, Kenneth, & Dittmar, 2012). The gender quota in Norway has resulted in a large number of inexperienced women having a seat on the board, which subsequently affected the stock performance in a negative way (Ahern, Kenneth, & Dittmar, 2012). It is evident that the women elected as board members after the gender quota are less experienced than their male colleagues, due to their absence on boards previously and hence the lack of opportunities to gain that experience.

However, it has been noted that female board members are more educated than male board members. Education is namely one of the key measures that makes it possible for minorities to distinguish themselves and gain widespread recognition in the world of executives in order to

start their career and gain experience (Hillman, Cannella, & Harris, 2002). This statement is also supported by Storvik and Teigen (2010), who find that female directors appear to be younger, however more educated than male directors. According to the widely researched agency theory and resource dependence theory, which are both used to explain board governance, the gender of a board member would not have any impact on ones performance as a board member. However, Nielson and Huse (2010) applied gender differences theories to the context of boards and indicate that there are some gender related differences in leadership styles that have a profound effect on the effectiveness thereof. They found that the presence of female directors lead to less conflicts and more board development activities, ultimately being positively related to the board’s strategic control and effectiveness. It is also found that more women on the Norwegian boards imply a stronger stakeholder and long-term orientation (Matsa & Miller, 2013). The impact of the gender quota on independence status, age, qualifications and experience has also been previously researched in other settings. It is believed that female directors are more likely to be outsiders, compared to their male colleagues. Where an outside director has proven to be more independent and a better monitor than an insider (Adams & Ferreira, 2009). Generally, outside directors are not tied to the management and hence more active monitors than inside directors. This implies a positive change in corporate governance.

Although previous research has created a large pool of knowledge regarding this subject, I attempt to add more depth to this research base and expand it by looking at two different and well-known concepts in corporate governance. The first set of questions that is investigated, deals with whether the effect of early adoption of the quota on the frequency of CEO turnover is significantly different than the effect of late adoption. Third, the effects on the independence of the boards are further investigated by asking the question whether the number of board interlocks has increased as an effect of the early versus late adoption of the gender quota, assuming the pool of experienced women is smaller than the pool of their more experienced male competitors. In general, should early adoption be seen as an indicator of the responsiveness of an organization towards wider social issues and hence does this have a signaling purpose? Are these early adopters viewed to be pioneers or does early adoption of the gender quota mostly depend on the industry and the availability of these women?

2.4

Hypotheses

The conceptual framework, as elaborated on in the previous sections of this thesis and which treats the effects of regulation on financial reporting quality is operationalized by zooming in on the effects of early versus late adoption of the gender quota in Norway on two corporate governance issues, namely CEO turnover and board interlocks. Prior research has already revealed several determinants of these two phenomena, however the effect of the gender quota has not yet been researched. Below the hypotheses development is set out.

2.4.1 CEO turnover

A Chief Executive Officer shapes and is responsible for strategic decisions, hence CEO turnover is believed to have a significant impact on firm performance and the governance of a firm (Shen & Cannella Jr., 2002). As mentioned previously boards play a major role in the monitoring of management in order to minimize the agency conflict and associated costs. Also, the board is likely to influence organizational performance. This notion comes from suggestions that a firm’s Board of Directors contributes to the corporate governance of that firm in that it selects and evaluates the top management, who in their turn set company objectives and oversee the day-to-day business (Wagner III, Stimpert, & Fubara, 1998). The effectiveness of the board’s monitoring activities may manifest itself in disciplining management as the board is the shareholders´ first line of defense against incompetent management (Weisbach, 1988). CEO turnover, meaning firing a CEO and hiring a new one, is one way of disciplining them. Hence, a possible measure for monitoring effectiveness is the CEO turnover. There has been growing interest in this corporate governance issue and the body of research is expanding rapidly. Research has shown that CEOs that are approaching the age of retirement are turned over more frequently which is viewed as a normal course of business. Furthermore, the larger a firm, the more frequent a CEO turnover takes place (Eriksson, Madsen, Dilling-Hansen, & Smith, 2001). Prior literature also suggests that board characteristics have and influence on the frequency of CEO turnover and specifically that insider directors are less likely to remove members of the management team, than are outsider directors. Insiders can be defined as those who are current or former employees or officers, their associates, or their family members. Whereas outsiders do not have any of these ties to the firm (Wagner III, Stimpert, & Fubara, 1998). Research has indicated

their careers are not tied to these top managers and have a reputational incentive to remove an ineffective manager (Dahya, McConnell, & Travlos, 2002). Prior literature suggests that there is an inverse relation between the percentage of insider board members and the presence of female directors (Carter, Simkins, & Simpson, 2003). According to Ahern et al. (2012) the gender quota in Norway has also led to more outsider (female) directors. Additionally, a sudden addition of more female directors, which implies a significant change in the composition of the board may alter the decision making process on replacing a CEO or other top manager as was shown by Eriksson et al. (2001). As to a difference between early and late adoption of the quota, research on the impact of regulation on corporate governance has indicated that the voluntary adoption of a proposal signals quality and has a positive effect on the governance practices and firm performance (Larcker, Ormazabal, & Taylor, 2011). Also, it is expected that the more experienced women will be selected first due to their knowledge and expertise. The pool of these experienced women is larger for the early adopters relative to the late adopters, hence more likely that the early adopters select the more experienced women to be part of their board. A more experienced board has a positive impact on the effectiveness of the corporate governance, hence it is expected that this also is visible in the number of CEO turnovers as this is one of the indicators of good monitoring (Weisbach, 1988). This is confirmed by the research results of Hermalin (2005) who states that greater board diligence leads to shorter CEO tenures, which is related to more frequent CEO turnovers.

Based on above literature review, I expect the frequency of CEO turnover to increase more for an early adopter relative to a late adopter, which leads to the development of the first hypothesis.

Hypothesis 1: The frequency of CEO turnover increases more for early adopters of the gender quota relative to late adopters.

2.4.2 Board interlock

Interlocks (shared directors) have been claimed to be devises for intercorporate collusion or cooptation, for bank control over corporate decision making and for the aggregation and advancement of the collective interests of the corporate elite (Davis, 1996). Existing literature suggests that interlocked boards cause earnings management contagion across firms (Chiu, Teoh, & Tian, 2013). Multiple board interlocks also raise independence questions and questions about

the monitoring effectiveness of busy board (Bizjak, Lemmon, & Whitby, 2009) (Ferris, Jagannathan, & Pritchard, 2003). However, it is also shown that board interlocks could potentially affect corporate governance in a positive manner. Interlocks could for instance be of advantage to outside directors, who would otherwise have an informational disadvantage, by spreading information on and insights of the industry and general economy, as well as information on good governance practices. Due to the relatively limited pool of experienced women, interlocks between boards of directors of different firms become more likely after the introduction of the quota. The gender quota serves as an external shock which creates a temporal shortage of skilled and qualified women that are eligible for these top positions. An easy way to comply with the gender quota is to select from the few available women with already governance experience. For the early adopters of the gender quota the pool of skilled and qualified women is deemed larger than the remaining pool for any late adopter, as it is assumed that the most eligible of the available women will have already been offered a board position. A late adopter has to choose from a more limited amount of women. Additionally, these women can already be part of a board and due to shortage the repeated use of a select few women creates an increase in the number of board interlocks. Acknowledging the importance of the subject and implications of board interlocks on corporate governance the following hypothesis is formulated based on above discussed assumptions.

Hypothesis 2: The number of board interlocks is higher for late adopters of the gender quota relative to early adopters.

3

Research Methodology

Below it is described how the concepts and hypotheses as developed above are ultimately researched. First, the method of sample selection is elaborated on, second the different variables of interest are extensively treated and finally the model that is used is set out.

3.1

Sample Selection

In Norway a public limited firm is known as ‘Allmennaksjeselskap’, abbreviated ASA. Another form of a limited liability company known in Norway is the private limited company, Aksjeselskap, abbreviated AS. The key differences between these two forms are that ASA firms are much larger (minimal capital requirements 10 times larger than for the AS firms) and more important for this study is that only the ASA firms are subject to the gender quota. The research of Ahern and Dittmar (2012) shows that as a consequence of the mandated gender quota, many public limited firms delisted or incorporated in another country to avoid complying with the mandate. Consequently limiting the number of positions available for women and also restricting the number of firms in the treatment group of this research. Based on Ahern and Dittmar’s retrieval from Statistics Norway there were 529 public limited firms in 2001 and in 2009 there were only 351 public limited firms listed, which means that in 2009 there are less than 70% as many public limited firms as there were in 2001. The resistance of companies towards the mandatory gender quota is notably visible, however sufficient data is available to conduct this study. Below I will describe the method used for data extraction, however first a timeline is sketched.

The gender quota was imposed on public limited firms in Norway in 2003, however the quota was only mandated in 2006 due to the fact that most firms did initially not comply voluntarily. Companies had until January 2008 to meet the requirements under the threat of being dissolved for non-compliance. In the below schedule a timeline is shown to illustrate the process of the quota in Norway.

Figure 2 Voluntary Gender Quota Mandatory Gender Quota Deadline Gender Quota 2002 2003 2006 2008 2010 Early Late Adoption

The firms viewed as early adopters in this research are the firms that reached the 40% target before the beginning of 2006. The quota was already introduced in 2003 as a voluntary measure, hence the firms that reached the 40% target in the voluntary period are viewed as early adopters. The firms that complied between 2006, when it became mandatory, and 2008, when non-compliance was penalized, are viewed as late adopters.

In order to be able to successfully separate the early adopters from the late adopters and capture the full effect on board interlocks and CEO turnover, I analyze a panel of Norwegian public limited companies in the years 2003 to 2010 using data from DataStream, Thomson One and annual reports. DataStream Asset4 database contains all information concerning environmental, social and governance subjects. I collect the information on percentage of female board members, board interlocks and relevant control variables from DataStream. Specific information on the CEO and the rate of turnover is obtained from Thomson One. In case certain information is not available here, the annual reports are used to collect the missing data on CEOs.

Following these procedures yields an ultimate sample of 114 firm-year observations for the period 2003 till 2010 for 30 unique Norwegian firms. Table 1 below described the sample selection. The initial sample of 201 firm-year observations on gender, turnover and interlock data was drawn from both DataStream and Thomson One as mentioned above. After excluding the firm-year observations with missing control variables, the full sample is reduced to 114 firm-year observations.

Table 1

All

years 2003 2004 2005 2006 2007 2008 2009 2010

Number of observations 201 15 23 26 27 28 27 27 28

(-) observations without control variables in year t -87 0 -8 -11 -13 -14 -12 -14 -15

Final sample 114 15 15 15 14 14 15 13 13

3.2

Variables

3.2.1 Early versus Late adopters

What makes a firm comply with a regulation early? This is an interesting and important question as recent theory argues that firms have different needs and expectations of the board of directors and the members’ experience and expertise. Some boards, for instance, function better with more insiders that have firm specific knowledge than outside board members (Harris & Raviv, 2008). However, when you think of the board of a large diversified company, you expect the board to be diversified as well. Hence, when a heterogeneous demand for board members is present in the corporate world, it is expected that the firms that have the greatest need for replacement by new female directors that share a similar knowledge and experience, will be the first companies to comply with the quota, anticipating the shortage of equally qualified women. Whereas the firms where directors are more easily replaced will perhaps be late adopters of the quota. Take for instance R&D intensive firms; experience and knowledge of board members of these companies are essential features and due to the relative limited pool of new female directors with matching characteristics, it is expected that firm with specific values and requirements be the first to select the new female directors.

Voluntary adoption of the gender quota is as mentioned earlier labeled early adoption in this paper, the mandatory adoption is labeled as late adoption. The characteristics of the early and late adopters are believed to differ. Also, the pool of possible female directors is expected to be different for the early versus late adopters, hence different effects on the frequency of CEO turnover and board interlocks is predicted.

As the research focuses on the effect of the gender quota on the frequency of CEO turnover and board interlocks, I will use a dummy variable GDIV where 1 equals the firms that complied between 2003 and 2005, and 0 equals the firms that complied between 2006 and 2007.

Furthermore a dummy variable GDIV08 is included and equals 1 if the board complied with the quota in 2008, before any sanctions would have been laid upon them.

Also, as the essence of the gender quota is that the board should be comprised of a more equal percentage of both male and female members the input will also be drawn from the percentage of females on the board (Ahern and Dittmar, 2012):

3.2.2 CEO turnover

The dependent variable CEO turnover will be measured by a dummy variable, where the value is equal to one if a CEO turnover has taken place during that firm-year and zero when no turnover has taken place. The data is, as mentioned before, drawn from the Thomson One database and manually converted to a dummy variable.

3.2.3 Board interlock

The dependent variable board interlocks will be measured as being the average number of other corporate affiliations for a board member in firm-year t. This variable is, as opposed to the dependent variable CEO turnover, directly imported from the ASSET4 database from DataStream.

3.2.4 Control variables

Based on the existing research base, I have included several variables to control for phenomena that have shown to be significant determinants of CEO turnover and board interlocks. Below the control variables and their predicted relation to either CEO turnover and/or board interlocks are set out.

To estimate the impact of the quota the pre-quota variation in female board representation as an exogenous instrument for the variation in board changes mandated by the quota (Ahern and Dittmar, 2012). The firms that had a greater proportion of female directors face a smaller constraint than the firms with less female members of the board. The pre-quota presence of female directors is measured using the percentage of female board members in 2002, hence prior to the quota that was introduced in 2003. In the regression model the pre-quota diversity is included as GDIV02.

The control variable TRI that is included in the regression estimate, concerns the Total Return Index. The TRI is a key datatype in DataStream that shows the growth in value of a

shareholding over a specified period, assuming that dividends are re-invested to purchase additional units of equity at the closing price applicable on the ex-dividend date. The TRI is calculated as follows:

= *

Where:

Pt = Price on ex-date

TRI is included in the model following Campbell et al. (2011) who similarly included stock returns in their model, since extant studies have found that poor stock returns are related to poor CEO’s decisions. As variation in stock performance is believed to have an effect on the rate of forced turnovers, this is controlled for in my analysis.

The corporate governance score (CG) is included in the analysis and measures a company’s systems and processes that are in place to ensure that its board members and executives act in the best interests of the long term shareholders. It reflects a company’s capacity, through its use of best management practices, to direct and control its rights and responsibilities. A higher CG Score indicated a stronger corporate governance practice within the company and hence is presumed to have a positive effect on the frequency of CEO turnover and a negative effect on the number of board interlocks.

A historic performance measure as Return on Assets (ROA) reflects both the success of a firm’s policies and strategies as well as the CEO’s implementation of those policies and strategies. Farrel and Whidbee (2003) shows that the ROA is significantly and negatively related to the likelihood of CEO turnover, since letting the CEO go might be a reaction to poor firm performance. Hence, in order to control for the effect of a low ROA on the incidence of a turnover, I have included Return on Assets in the model. A measure for firm performance is also controlled for by Westphal et al. (2006), since executives and directors of poorly performing firms might perceive a greater need for ties to other companies. Hence firm performance is believed to have a negative effect on the number of board interlocks, thus controlled for in this model.

The number of employees within in a firm is a commonly used measure for firm size. Firm size is also typically included as a control variable in research on interorganizational networks (Westphal, Boivie, & Chng, 2006). An aspect of interorganizational networks is board interlocks,

hence to diminish the effect of firm size I controlled for firm size in this research. Farrel and Whidbee (2003) included the natural log of the number of firm employees as a proxy for size as several studies find a positive relation between the likelihood of CEO turnover and firm size. In line with prior literature I hence add the log of number of employees as a control variable (logEMPLY).

In line with the research conducted by Gul et al. (2011), the variation in the percentage of female directors across industries suggests the need for controlling for industries in the analysis, hence I included dummy variables for these different industries in the regression model. Also, DeFond and Park (1999) indicates that highly competitive industries have a positive effect on the frequency of CEO turnover, since it is more useful in these circumstances to use Relative Performance Evaluation (RPE), which subsequently improves the board’s ability to identify unfit CEOs. In order to take into account the effect of the differences in industries, these are controlled for using the NACE Rev. 2 code, which is the statistical classification of economic activities in the European Union.

Above described control variables are included in both models. Specific for the turnover model I have included CEO tenure, CEO age and independent board members. For the analysis on the effect on number of board interlocks I have included the number of board meetings as a control variable. Below these variables will each be explained separately.

CEO age and tenure are two variables that are often controlled for in researches on the determinants of CEO turnover. Both age and tenure are related to the frequency of CEO turnover for obvious reasons. CEOs approaching the retirement age are more likely to end their tenure voluntarily (Shen & Cannella Jr., 2002). In order to control for the positive effect of CEO age on the turnover rate this is included in the model. CEO tenure is measured as the number of years an individual has served as the CEO of the firm from which he or she was departing. Campbell et al. (2011) argue that CEO tenure is one of the indicators for CEO perceived ability, hence my analysis should be able to separate the effects of low ability from the effects of the gender quota on CEO turnover.

The percentage of independent board members (INDEP) is part of the governance structure and believed to have a positive effect on the frequency of CEO turnover according to Shen and Cannella (2002) as independent board members are believed to be better monitors and more

likely to dismiss a CEO in case of bad performance, since they also have no further ties to this CEO or the firm. The percentage of independent board members as reported by the company is included as a variable to control for the impact of the governance structure.

Based on the research of Fich and White (2005) the number of board meetings appears to have a negative effect on the number of interlocks. It is assumed that when a board is more active (as evidenced by more meetings) there may be less time to seek other directorships or executive positions in another firm. Hence to control for this effect the number of board meetings (BMEETING) is included in the regression analysis.

Below, in table 2, all variables and their descriptions are shown for reference.

Table 2

Definition

TURNOVER Dummy for CEO turnover in firm-year t

INTERLOCK Average number of other corporate affiliations for the board member in firm-year t EarlyAdopt Dummy for percentage of female directors > 40% in firm-year 2003 - 2005 GDIV Percentage of female directors in firm-year t

GDIV08 Dummy for percentage of female directors > 40% in firm-year 2008 GDIV02 Pre-quota percentage of female directors (firm year 2002)

TRI Total Return Index in firm-year t

CG Corporate Governance Score in firm-year t ROA Return on Assets in firm-year t

TENURE CEO tenure measured in years in firm-year t

INDEP Percentage of independent board members in firm-year t AGE Age of CEO measured in years in firm-year t

BMEETING Number of board meetings in firm-year t

logEMPLY Natural log of number of employees in firm-year t

3.2.5 Regression

Above described variables are incorporated in two regression formulas in order to test the hypotheses that are developed. Below the regression formulas that are used are stated.

Hypothesis 1 is tested using the following logistic model:

= + + + ∗ +

08 + 02 + + + + +

Hypothesis 2 is tested using below model:

= + + + ∗ +

08 + 02 + + + + +

4

Tests and Results

4.1

Summary statistics

Table 3 presents summary statistics for the gender diversity, interlock, turnover and control variables.

The below summary statistics contain results where the unit of observation is firm-year. The indicator values TURNOVER, EarlyAdopt and GDIV08 are each zero-one dummy variables. The mean of the EarlyAdopt variable represents the proportion of firms classified as early adopters, hence only approximately 1,5% is classified as being an early adopter.

Gender Diversity over the years has a mean of 25% and median of 30% as also shown in Table 3. The gender diversity, tenure and age figures are in line with prior research on board characteristics and CEO turnover (e.g. Campbell et al. (2011) and Ahern et al. (2012)).

Figure 1 depicts the upward trend in the average percentage of female board directors. It is however shown that in the sample the target of 40% is not reached in the years of 2008 and later. This has minimal implications for the research as early adopters are classified as having reached the target of 40% before 2006 and are then compared to the late adopters.

Table 3 Mean Median TURNOVER (indicator) 0,15 INTERLOCK 1,84 1,62 EarlyAdopt (indicator) 0,15 GDIV 25,23 30,00 GDIV08 0,38 GDIV02 20,05 20,00 TRI 1.557,68 202,70 CG 51,31 54,18 ROA 5,48 6,31 TENURE 6,00 4,00 INDEP 59,59 64,71 AGE 51,13 50,00 BMEETING 12,30 12,00 logEMPLY 10.873,17 5.171,50

Note: Industries are based on four-digit NACE Rev. 2 code and controlled for by using dummy variables but not reported. Refer to 3.2.4 for variable definitions.

Figure 1 0,00% 5,00% 10,00% 15,00% 20,00% 25,00% 30,00% 35,00%

Average percentage of female

directors

Percentage of female directors

4.2

Results

As previously described, the effects of early versus late adopters of the gender quota on board interlocks and CEO turnovers are examined in a regression framework that controls for other determinants of interlocks and turnovers. Below the results of these analyses are being treated for the two models separately.

4.2.1 CEO turnover

Table 4 contains a summary of the CEO turnover model. A logistic regression analysis was conducted to predict the effect of implementation of the gender quota on the CEO turnover. A test of the full model against a constant only model was statistically significant, indicating that the predictors as a set reliably distinguished between the firms that either fulfilled the 40% criterion and the firms that did not (chi square = 37,174, p < ,05 with 20 degrees of freedom). Nagelkerke’s R² of ,745 indicated a moderately strong relationship between the predictor and the prediction. Although there is no close analogous statistic in logistic regression to the coefficient of determination R² as is used in a linear regression for indicating the explanatory value of the model. The Nagelkerke modification attempts to imitate the R². It ranges from 0 to 1 and is quite a reliable measure of the relationship. Rather than using a goodness-of-fit statistic, you can also look at the proportion of cases the model has managed to classify correctly. For this the prediction overall success percentage can be used, which tells us how many of the cases where the observed values of the dummy variable CEO turnover was 1 or 0 respectively have been correctly predicted. The overall percent is 93% (97,8% for no turnover and 75% for a CEO turnover). In a perfect model the overall percent would be 100%, hence the model which predicted 93% correctly has a high explanatory value. The Wald test is used in empirical studies to determine whether a certain predictor variable X is significant or not. In this model the Wald criterion demonstrates that only the fulfillment of the gender quota during 2008 and CEO tenure made a significant contribution to prediction (p = ,035 and 0,083 respectively). Early or late adoption was not a significant predictor. The firms that have reached the target of 40% female representation in the board of directors in 2008 have a negative effect on CEO turnover. This is derived from

the sign of the coefficient as presented in table 4. A minus sign in front of the coefficient indicates a negative effect, the absence of the minus sign indicates a positive effect on the frequency of turnover.

The dummy variable EarlyAdopt does not have the predicted sign, however it is also not significant, hence there is no explanatory value and the sign might be due to coincidence. The results show no empirical relation between early versus late adoption of the gender quota and CEO turnover. The adoption of the quota in 2008, before sanctions would apply however does have significant impact on CEO turnover as mentioned earlier and which can be viewed in table 4. Fulfillment of the quota as per 2008, which can be classified as an extremely late adopter, has a negative effect on CEO turnover. This can be interpreted as the board being worse monitors and not effective in their corporate governance practices. These results can be seen as support of the null hypothesis, being that adoption in 2008 (laggard) has a negative effect on the frequency of CEO turnover. This was as expected and can possibly be explained by a smaller pool of available women, that also have the right characteristics like experience and knowledge (Ahern, Kenneth, & Dittmar, 2012). A more experienced board is believed to have a positive impact on the effectiveness of the corporate governance (Weisbach, 1988). The expectation that laggards of the adoption of the quota would include less experienced women in the board, which subsequently leads to less effective monitoring and hence less CEO turnovers was correctly phrased and is supported by the results as indicated below.

Looking at the control variables in the model, the only variable that is deemed significant in explaining CEO turnover is CEO tenure. CEO tenure has a positive effect on the frequency of turnover, as can be seen in table 4. This is in line with prior research, which indicates that the likelihood of a CEO turnover increases in case he or she has a longer tenure. The other control variables show an insignificant impact on the independent variable, CEO turnover, hence no conclusions on the relation to the independent variable can be drawn.

Table 4

Variable Coef Wald

Constant 79,756 ,000 logEMPLY 4,764 2,085 GDIV02 ,290 ,000 GDIV -,012 ,021 EarlyAdopt -6,243 ,000 GDIV08 -6,096** 4,437 TRI ,006 1,038 CG -,114 1,233 ROA -,213 ,981 TENURE 1,746* 3,004 INDEP ,117 1,934 AGE ,032 ,038 Chi-square 37,174** Nagelkerke R² ,745 Prediction criterion 93

Note: Industries are based on four-digit NACE Rev. 2 code and controlled for by using dummy variables but not reported. Refer to 3.2.4 for variable definitions. **, * indicate significance at a p-value of less than the 5% level (2-tailed) and 10% level (2-(2-tailed), respectively.

4.2.2 Board interlock

The explanatory power of the second model, which tests for the effects of the gender quota on the number of board interlocks is indicated with the R², this is set out in table 5. R² indicates that model 2 explains 60% of the number of board interlocks, which in practice is an acceptable percentage. The adjusted R² is ,47 and is the best estimate of the degree of relationship in the basic population as here is adjusted for the number of predictors in the model and the value of the model only increases when a new term enhances the model above what would be obtained by probability.

The Durbin-Watson value is 1,7 which is roughly equal to 2 and indicates that the residuals are uncorrelated.

As shown in table 5, and in consonance with model 1, the dummy variable GDIV08 is solely significant. The gender diversity in 2008, which was the year to either oblige to the quota or be subject to sanctions, has a negative relation with the number of board interlocks. This means that the second hypothesis should be rejected. This could be explained by the following, the most known and experienced female directors or women in top management functions are chosen first by the early adopters or during 2006 and

2007. In 2008 the pool of women is less, however in order to still comply with the regulation women have to be represented in the board room. The women that are still “available” are the ones with less knowledge and experience (Ahern, Kenneth, & Dittmar, 2012), hence also no other affiliations like a top management function or directorship in another firm, which subsequently leads to less board interlocks. This negative relation is significant at a level of ,05. The significance level indicates whether the relation is due to coincidence or indeed has explanatory power on the independent variable board interlocks. The t-statistic is used to determine a p-value that indicates how likely we could have gotten these results by chance, if in fact the null hypothesis were true. By convention, if there is less than 5% chance of getting the observed differences by chance, we reject the null hypothesis. As mentioned above, the only statistically significant relation exists between adoption of the gender quota in 2008 and board interlocks. Table 5 also shows the VIF (Variance Inflation Factor) values of all the variables. The VIF values are added to the analysis to check for multicollinearity issues. All variables have a VIF value lower than 5, which then in practice can be viewed as no to little indication for multicollinearity. The pre-quota gender diversity (GDIV02) is deleted from the regression analysis, due to a high VIF value.

Apart from the GDIV08 the other variables show no significance in the model and hence the signs might be due to chance and hence are not further discussed. The difference between the effects of early versus late adoption on board interlocks can thus not be fully supported, apart from the laggards that have a negative effect as previously elaborated on. This means that the null-hypothesis is rejected.

Table 5

Variable Coef t-Stat VIF

Constant ,350 ,448 logEMPLY ,142 1,550 3,837 GDIV ,005 ,640 2,792 EarlyAdopt ,328 ,847 4,051 GDIV08 -,444** -2,538 1,394 TRI -2,179E-7 -,012 3,480 CG ,006 1,108 1,959 ROA -,004 -,327 2,258 BMEETING -,015 -,607 2,884 R² ,601 Adjusted R² ,474 Durbin-Watson 1,705

Note: Industries are based on four-digit NACE Rev. 2 code and controlled for by using dummy variables but not reported. Refer to 3.2.4 for variable definitions. **, * indicate significance at a p-value of less than the 5% level (2-tailed) and 10% level (2-(2-tailed), respectively.

The results of both models show opposite effects on two corporate governance mechanisms. The adoption of the gender quota in the last possible year shows both a negative effect on the frequency of CEO turnover and the number of board interlocks. The first indicates a worsened quality of corporate governance, whereas the second indicates a betterment of the corporate governance practices within a firm. Prior literature has shown that disciplining management by means of firing indicates good corporate governance (Shen & Cannella Jr., 2002) and multiple board interlocks raise independence questions and questions about the monitoring effectiveness of busy board (Bizjak, Lemmon, & Whitby, 2009). Ahern et al. (2012) have presented their results showing that the board of directors became less experienced as a result of the gender quota. The lack of experience in the board of directors could be explaining the results that indicate less CEO turnovers and board interlocks for the boards that adhered to the gender quota in the last year (2008) and hence had less experienced women available for selection.

5

Concluding remarks

5.1

Conclusion

The purpose of this paper is to provide empirical evidence on the impact of one measure of board diversity, the presence of female board members, on corporate governance and in particular the frequency of CEO turnover and the number of board interlocks. The primary motivation for this study is to provide more insight on the effects of a gender quota, since multiple countries including the Netherlands and Belgium have set deadlines for listed companies to select a certain percentage of female directors to be part of the board. The Norwegian gender quota that was introduced in 2003 and mandated in 2006 serves as the perfect natural experiment to be able to conduct the research and provide evidence to help the public firms acknowledge the added value and/or pitfalls of the gender quota and ultimately a more diverse board. The gender quotas serve as a means to helping women break the so-called glass ceiling. Gender diversity has previously proven to result in improved monitoring compared to a homogenous board (Abbott, Parker, & Presley, 2012), hence diversity in the board could also be beneficial for poor governance. Which leads me to additional motivation for the study, namely that due to recent scandals in the world of corporate business the trust in the financial reporting quality is lost. The corporate world is in their quest for a means to restoring the trust, which can not only be recovered based on new and stricter financial reporting initiatives, however also implies the need for changes in the institutional environment. Part of the institutional environment is corporate governance and legislation. An improvement in the corporate governance of a firm could ultimately lead to better quality of financial reporting and help restore the trust therein. The gender quota leads to more diverse boards, which is believed to be an improvement of corporate governance. However, is it more beneficial for corporate governance when changes thereto are left to the market or when this is imposed on through new legislation, like a gender quota.

This paper contributes to the board-level gender diversity literature by incorporating the aspect of voluntary versus mandatory adoption of the quota and by