Bachelor Thesis Economics & Business
“Does having a choice in board structure affect firm performance?”
A comparative study on corporate board structures in the Netherlands and the United States.
Name Student: Faas Gelens Student number: 10034714
University: University of Amsterdam Faculty: Economics and Business Track: Finance and Organization Field: Organizational Economics Supervisor: Silvia Dominguez Martinez
Table of content
1. Introduction………. 2
2. Literature Review……… 4
2.1. Corporate Governance and the Board of Directors……….. 4
2.2. The Two-Tier Board………. 5
2.3. The One-Tier Board……….. 6
2.4. Differences in the two Board Structures ……….. 7
2.5. The Anglo-Saxon versus the European Continental Approach……… 8
2.6. The Case for the Netherlands………....9
2.7. The Case for the United States………. 11
2.8. Prior Research and Hypotheses……… 12
3. Research Methodology……….... 17
3.1. Sample, Data and Timespan………. 17
3.2. Variables of Interest……….. 18 3.2.1. Dependent Variables……….. 18 3.2.2. Independent Variables………... 19 3.2.3. Control Variables………... 19 3.3. Research Method……….. 20 4. Results………. 23 4.1. Descriptive Statistics……… 23 4.2. Regression Results……….... 24
4.3.Testing of the Hypotheses………. 26
4.4. Robustness Checks………... 27
4.4.1. Robustness Check using the ROE………..27
4.4.2. Testing Hypotheses……… 29
4.4.3. The Regressions for each year Individually………... 29
5. Discussion……… 30
6. Conclusion………... 32
References………... 34
1. Introduction
On the 1st of January 2013 the Dutch act called Act “Bestuur en Toezicht” has entered into effect. This act legally allows Dutch listed companies to choose between a one-tier board structure and a two-tier board structure. The Netherlands now join the few countries in which companies are free to choose between the two types of corporate board structures. Before this act the one-tier board structure was allowed in the Netherlands, but there was no legal basis for this. The main goal of the Act is to improve the international attractiveness of Dutch listed companies. In the United States companies are obliged to apply a one-tier board structure. Economists have been debating about which of the two corporate structures is the best. Both of the structures seem to have their advantages and disadvantages. The major difference between the two corporate structures is the way in which the monitoring directors are situated in the company.
The question that arises is whether the fact that Dutch companies have a choice in board structure will have an effect on their firm performance. This thesis explores the relationship between the freedom in choice of corporate board structure and firm performance by comparing Dutch and American listed companies. In addition this research investigates if one of the two board structures can be seen as the best. In contrast to prior research, which mainly focuses on which board structure can be seen as superior, the focus of this thesis lies on the effect the freedom of board structure has on the firm performance.
The results of this thesis show a negative relationship between the fact that a company is Dutch and the firm performance, using the Return on Assets as firm performance measure. However, no significant relationship was found when the Return on Equity was used as performance measure. Because of these inconsistent findings, no conclusion can be made regarding the relationship between freedom in choice of board structure and the firm performance. Neither did the results of this thesis give sufficient information to make conclusions about the superiority of one of the two board structures. The results of this thesis show that differences in the performance of the firms are largely explained by factors as type of industry and firm size.
This thesis is organized as follows. In the second section the differences in the two corporate board structures will be described. Possible advantages and disadvantages of the two main approaches will be discussed. The corporate governance of the two countries of interest will be summarized briefly. At the end of the second section prior research will be discussed and two hypotheses are formulated. In the third section the research methodology is
described. The fourth section analyzes the results of the regressions that were performed on the designed models. Several checks will be performed to check for the robustness of the findings. The fifth section discusses the results that were found. Finally this thesis will come a conclusion and makes suggestions for further research.
2. Literature Review
In this section a theoretical background on the different board structures is given. Firstly corporate governance and the board of directors are discussed. Secondly the two types of board structures are described and compared. Thirdly the different global approaches are discussed. Additionally the corporate governance in the Netherlands and the United States will be described. Finally some prior research related to this thesis is discussed and the hypotheses of this thesis are formulated.
2.1. Corporate Governance and the Board of Directors
In a company with a separation of ownership, the shareholders, and control, the managers, some problems can arise. The first problem that could arise is that the management is not acting on behalf of the shareholders but is maximizing its own return or is taking excessive risk. This problem is referred to as the principal-agent problem (Jensen 1976). The second problem is the so-called free-rider problem. In this problem the costs for the shareholders to monitor the management of the company are exceeding the benefits of being able to exert influence. The non-monitoring shareholders will gain the same advantages as the monitoring and so this will result in a situation in which none of the shareholders is monitoring (Jungmann 2006).
Corporate Governance is trying to solve these problems. Corporate governance is the framework of rules and practices by which a board of directors ensures accountability and transparency to its stakeholders. Shleifer and Vishny (1997) refer to corporate governance as the ways in which suppliers of finance to corporations assure themselves of getting a return on their investment. Corporate governance structures are sets of institutional arrangements that tend to align the interest of the management with that of the risk bearing shareholders. The main institutional arrangement assigned with a governance function is the board of directors. This corporate body has the right to fire, hire and compensate the senior management and has the task to resolve conflicts of interest between decision makers and the risk bearing shareholders. For the majority of the countries, corporation law states that listed business corporations should be under the guidance of a board of directors. Economic theory and corporation law are mostly not very dictating concerning the size, composition and the other tasks of the boards like frequency of the board meetings (Baysinger and Buttler 1985).
Although differences in board characteristics arise across the world, two main structures can be found. There can be made a distinction between the two-tier and the one-tier board.
2.2. The Two-tier Board
When a company applies a two-tier board structure, two organizational layers are assigned. There is the management, or executive board, and a separate supervisory, or non-executive board. Each board has its own separate meetings, to which only members of the concerning board attend. Both of the boards have their own individual tasks. The management board is responsible for the day-to-day operation of the firm. It manages the firm’s affaires, sets up goals and long-term guidelines and allocates the resources. The management board fully exists out of executive directors. These executive directors have the task of running the firm. They design, develop and implement strategic plans for the company. The management board acts autonomously and is not bound by orders of the shareholders or the supervisory board. The executive directors report to the head of the management board the Chief Executive Officer (CEO). This CEO leads the management board and acts as the chairman of this board. The chairman leads the board’s meetings. The members of the management board are assigned and fired by the supervisory board (Jungmann 2006).
The supervisory board exists out of non-executive directors only. These non-executive directors are not an employee of the firm and are not involved in the firm’s executive management. The task of the supervisory board is to monitor the management board and to promote the interests of the shareholders of the firm. Things being inspected by the supervisory board are the financial reporting, the remuneration of directors and the risk that is taken by the management board. The supervisory board acts on the behalf of the owners of the firm, the shareholders. The members of the supervisory board are nominated by the supervisory board itself and elected by the shareholders on the shareholders meeting (Maassen 1999). The composition of the board is mostly well balanced. The non-executive directors get selected on characteristics like expertise, experience and personal background. On the head of the supervisory board is the chairman. The supervisory board has the right to elect and fire the members of the management board. When the firm is not performing properly the supervisory board has the right to fire executive directors. The supervisory board usually includes a share of independent outside directors. These directors have no monetary or pecuniary relationship with the firms except for their commission fees. The difference
between a non-executive director and an independent director is that a non-executive director is allowed to hold shares of the firm (Monitoring Commissie, 2008a).
The supervisory board acts like a counselor for the management board. The management board can explain and discuss their new strategies with the supervisory board. The supervisory board will provide comments and criticism with an aim on improving the strategic vision of the management board. Important decisions made by the management board need to be approved by the supervisory board. Major management decisions that alter the assets, earnings or financial situation of the firm need to be presented to the supervisory board. Examples of these decisions are large share issues, important acquisitions and big investments (Jungmann 2006).
In its task to monitor the management board the supervisory board is dependent on the information that is provided by the management board and information the non-executive directors have collected themselves. In many countries in which a two-tier board structure is commonly used, the corporate governance code recommends the supervisory board to apply committees. These committees are composed out of independent non-executive directors that have expertise on a certain field. The different committees all have their field that they supervise in particular. Examples of commonly used committees are audit committees, remuneration committees and risk committees. The audit committee keeps oversight on financial reporting and disclosure. The remuneration committee is supervising arrangements made concerning the remuneration packages provided to directors. The risk committee is responsible for providing oversight and advice to the supervisory board concerning the current and future risk exposures and company’s future risk strategy (Collier 1996).
2.3. The One-tier Board
In the one-tier board structure only one organizational layer is assigned. This board of directors consists out of the executive directors, which are entrusted with the task of managing the firm, and the non-executive supervising directors. Where in the two-tier board structure these managing and supervising directors are separated in their own board, in the one-tier board structure they are in the same single board. They attend to the same meeting. As both the managing and supervising directors are on the same board, they receive the same information at the same time. Because of this the supervising directors are able to react faster. Non-executive directors are more closely involved in the management of the firm. The non-executive directors are supervising throughout the total decision-making process of the
management directors whereas in a two-tier board structure the supervising by the non-executive mostly takes place ex post. Jungmann (2006) refers to this as the non-non-executives working in a retrospective manner in the two-tier board structure and in a reactive manner in the one-tier board structure.
In many counties that apply the one-tier board structure publicly listed firms are legally obliged to assign board committees. For example, an audit committee is required for companies that are on the stock exchange in the United States. The assignment of committees like a remuneration committee and a risk committee are strongly recommended by the Stock Exchange Committee, the institution that draws up regulations for public listed firms in the United States (SEC, 2002).
In some countries it is possible, when applying a one-tier board structure, to combine the role of CEO with the role of chairman of the board. This is called CEO-duality (Maassen 1999).
2.4. Differences in the Two Board Structures
The main advantage of the two-tier board is the division of roles between the directors. There is a clear separation between the directors that manage the company and the directors that supervise the managing directors of the company. In the one-tier board this division could get blurred. The non-executive directors of a unitary board are more involved in the daily business of the company than non-executives in a two-tier board, since in a one-tier board all the directors receive the same information. However this involvement could be harmful to the supervising role the non-executives have. Jungmann (2006) stated that non-executive directors in a unitary board are well aware of their strategic role but less of their monitoring role. Members of a one-tier board fulfill both managerial and supervisory roles. According to Jungmann these members should take decisions and, at the same time, monitor these decisions. In a one-tier board structure the supervisory directors are less independent of the management. This problem does not arise in a two-tier board structure.
The main advantage of the one-tier board is the fact that non-executive directors are closer involved in the operating of the business. In the two-tier board the supervisory directors are not involved in the decision making, except for the important decisions that need their approval. In the two-tier board structure the supervisory board has to rely on the information reported by the management directors. There exists an asymmetry in information. In a unitary board the non-executives no longer have to wait for the reporting of the management board.
Both the executive and non-executive directors receive information at the same time. This leads to a close involvement in the decision-making from the beginning in contrast to the retrospective decision-making in the two-tier board (Jungmann 2006). In a two-tier board non-executives often doubt their decisions as a result of lack of information. In a one-tier board the non-executives can intervene quicker which may result in a more efficient way of decision-making. Big decisions, that in the two-tier board need approval of the supervisory board, can be evaluated faster in the one-tier board structure. This can be an advantage when quick decisions have to be made about business opportunities (Maassen 1999).
Companies applying a two-tier board show to have fewer meetings between executive and non-executive directors compared to one-tier boards. This may make it difficult for directors to build trust relationships and this may be bad for the communication and flow of information between the boards (Spencer Stuart 2013). Turnball (2000) states that the in a one-tier board the relationship between management- and monitoring directors could get too close and be harmful to the independence of the supervising directors. Aste (1999) states that the two-tier board can be a benefit when looking at company mergers. When two companies merge both of the CEO’s could get a leadership position in the firm. One could be elected for head of the management board and the other could be elected for head of the supervisory board. In his paper Turnball (2000) summarizes a few possible weaknesses of the unitary board. First he states that in unitary control increases corruption. Directors are able to influence and maintain their private benefits and performance of the firm. As an example they are able to influence their remuneration package. Secondly he states that in a unitary board the non-executive directors could suffer from an information overload. According to Turnball, supervising directors in a unitary board structure are unable to monitor, direct, remunerate and retire the management objectively.
2.5. The Anglo-Saxon versus the European Continental Approach
Regional and international developments have resulted in two leading approaches to the organization of corporate bonds across the world. First there is the Anglo-Saxon one-tier board and secondly there is the continental European two-tier model. Many variants of the two main approaches can be found. Countries applying variants of the one-tier board structure are the United States, the United Kingdom and Canada. Examples of countries applying the continental approach are Germany, Finland and the Netherlands (Maassen 1999).
According to Maassen (1999) the corporate concept of the Anglo-Saxon approach is based on reliance between shareholders and the managers. The Anglo-Saxon approach, also called the shareholder-orientated approach, is based on the belief that self-interest and decentralized power can operate in a self-regulating, balanced manner. Profit maximization and material success are the main goals of entrepreneurs and managers in this approach. The European Continental approach on the other hand is more stakeholder-orientated. This approach not only considers the interests of the shareholders but also takes into account the interest of other stakeholder of the firm. Unlike the Anglo Saxon approach, this approach allows for multiple parties to be involved in the shareholder-manager agency problem, such as trade unions and work councils. Banks and enterprises usually own large proportions of the stock of these companies. For this reason bank representatives are commonly found on the board. In this approach it is important that small shareholders are protected against the parties holding a large amount of the shares (Cernat 2010).
Anglo-Saxon corporate systems are characterized by their dispersed ownership. When comparing the US and UK companies with European companies, the most striking point is the difference in ownership. In the United States most of the holders of the stock are individuals. In the European companies, enterprises and banks hold most of the shares. Another characteristic of the Anglo-Saxon orientated companies is that most of the shareholders only own a small amount of the shares whereas in the European countries the majority of shareholders owns a much larger proportion of the total stock, the so called block holders (Barca en Becht 2001). Another difference between the approaches is the influence of labor and trade unions. In the European countries the influence of the unions is much larger compared to the little influence these unions have in United States.
2.6. The Case for the Netherlands
In the history of the Dutch corporate governance the two-tier model has always been dominant. The supervisory boards, as we know them nowadays, have existed since 1623, when the East Indian Company (The VOC) created this body.
After the industrialization and the development of the capital markets after the Second World War, corporate governance became more important. In 1960 the Verdam Committee was set up, this government committee investigated if there was a need for a improved governance structure, with a focus on the management and supervision of larges companies. In 1965 the committee recommended a two-tier board in which certain rights were assigned to
the separated boards. The supervisory board was responsible for the electing of the management directors, appointment of supervisory directors and the approval of annual accounts. Important decisions by the management board needed to be approved by the supervisory board (Groenewald, 2005).
The Structuurwet of 1971 defines a large (“structuur”) company as a NV (Dutch public company) or a BV (Dutch private company) that has an issued capital and reserves equal or greater than 16 million euros, has a works council (“ondernemingsraad”) and employs at least 100 people. The Structuurwet obligates large companies to apply a two-tier board structure (Jong et al, 2005).
In the late nineties, the stock market boom caused the bankruptcy of several large companies and resulted into a number of accounting scandals. In addition, the public debate about the increases in the remuneration packages of management directors had created doubts concerning the accountability and supervision of corporate policy makers. Therefore, in 2003 the “Code Tabaksblat” was introduced, focusing on transparency. This Dutch code for corporate governance describes the duties and composition of the management and supervisory boards for Dutch listed firms. Although the code is based on the two-tier board structure it also contains a few provisions for companies applying a one-tier structure. The Dutch code is based on a so-called “comply-or-explain” approach. A company should state each year in its annual report how it applied the best practice principles of the Code and, if the best practice principle was not applied, carefully describe why it was not applied (Monitoring Commissie, 2009a). Dutch listed firms are legally obliged to apply the Code (Rijksoverheid, 2013).
In 2004 the Act “Structuurwet 1971” was revised. The revised version of the act had some adjustments that enlarged the influence of the shareholders. In 2008 the code Frijns was introduced. This code did not deviate from the Code Tabaksblat but it made some additions to improve the supervision of the management and the monitoring of the appliance of the code.
On the first of January 2013 the Act “Bestuur en Toezicht” has been applied. This act allows NVs and BVs to choose between the unitary and a dualistic system. The large companies that fall under the Structuurwet will now be allowed to choose the one-tier board structure. Before this act Dutch firms were in fact already allowed to apply the one-tier structure but there was no legal basis. This act now explicitly allows for the appliance of a one-tier board. The goal of the act is to improve the Dutch international competitiveness. The rationale for introducing this change is that by allowing firms to have a one-tier board,
internationally orientated firms can now compete better with Anglo-Saxon structured firms, like American or British companies.
2.7. The Case for the United States
In the history of the American corporate governance the interests of the shareholders always have been of importance. Throughout the 1920s and the 1930s the number of shareholders investing in the New York State Exchange (NYSE) doubled to 30 million. This large grow lead to an increase of the share prices. In this time there were hardly any laws governing the securities. In the Great Crash of 1929 the prices of share plummeted by 83%. In combination with Great Depression of the 1930s this lead to a public demand for federal security laws. Roosevelt created the Securities and Exchange Committee (SEC) and introduced several Security Acts. From the 1920s to the 1930s the idea of the separation of ownership and control was developed. And the possible divergence of interests between owners and managers was brought to light (Frentrop, 2002).
In the 1950s the number of small shareholders grew again. In the 1960s the number of mergers grew. The merger mania reached is pinnacle in 1968. Throughout the 1960s the important of independent directors was growing. From the 1970s the SEC and the stock exchanges began to see the task of the board to monitor the management and demanded even greater director independence. In 1977 the SEC approved the new rule that stated that all listed firms should have audit committees consisting out of independent directors that had no relationship or task in management. However the SEC had no authority to regulate the composition or memberships of the boards and committees as the corporate law was left to the individual states. In the 1970s the board’s functions changed from advising to monitoring. The concepts of “independent director” and “corporate governance” became more important and the arguments for having a chairman separate from the CEO began to grow (Frentrop, 2002).
The 1980s brought new waves of mergers and now hostile takeovers as well. In the 1990s the number of institutional investors increased. Their number nearly increased to 50 % of all ownership of US listed corporations in 1996. From 1990 on, the hostile takeovers declined. New corporate governance mechanisms began to play a role, one of those in particular were the stock option plans by which directors acquired more shares, which should lead to greater involvement of the board. The institutional investors promoted stronger supervision and monitoring by the board and its independent directors, especially through its
composition. Shareholders criticized the CEOs on the reproach that they would underperform and would be overcompensated. The SEC now permitted shareholder to include their criticism on the CEO’s compensation package (Holstrom and Kaplan, 2001).
In 1994 General Motors board published its Corporate Governance Guidelines. These guidelines were later considered as a charter for all firms. The CalPERS, the largest pension fund of California government employees, took initiative to give corporations grades on the compliance of the GM guidelines. In this time the idea was that the board, except for the CEO, should completely consist of independent directors. In the United States more stock option plans were granted to directors with the goal of making them in the interest of the company (Frentrop, 2002).
The 2001 and 2002 scandals of Enron, Tyco and Worldcom showed that higher compensation packages for directors is not a guarantee of better management. That is why President Bush signed the 2002 Sarbanes-Oxley Act. The Act goes in very much detail about the composition and the roles of directors on the board (Nordberg, 2011).
The current United States corporate governance is mainly determined by the legislation of the Sarbanes-Oxley Act of 2002 and the Securities and Exchange Commission (SEC), the New York State Exchange (NYSE) and the NASDAQ that have drawn up more detailed regulations. Both the NYSE and the NASDAQ listed firms must comply with the Sarbanes-Oxley corporate governance requirements. In the United States a one-tier board is the norm. Deviating from the one-tier board structure is not possible since there is no federal legislation for the two-tier board structure (Owen, 2003). The Sarbanes-Oxley regulations use a “comply-or-explain” approach in some of the parts, but mostly the regulations are stated by US regulation. Violating the requirements of the Sarbanes-Oxley rules will result in penalties or imprisonment. The approach has a much greater emphasis on regulatory enforcement than on voluntary compliance, as is the case in Netherlands. One of the main requirements of the Sarbanes-Oxley Act was that all listed companies should maintain a wholly independent audit committee including solely outside directors. The rules for the two listings in the US differ in some points. The NYSE specific rules can be found in the NYSE’s Listed Company Manual and those of the NASDAQ can be found in the NASDAQ Marketplace Rules.
2.8. Prior Research and Hypotheses
Summarizing, the two approaches both have their advantages and disadvantages. In the one-tier board structure the non-executive supervising directors are closely involved in
management of the company. This involvement leads to vast decision-making. However, this involvement could be harmful to the independence of the supervising directors. The two-tier board structure is focused on the clear separation between the management directors and the supervising directors. Which is to the benefit of the independence of the supervising directors. However, the supervisory board is not involved in the management of the firm, which leads to an asymmetry in information. Economists have debated the strengths and weaknesses of the two board structures. The main difference between the two models relates to the question if it is desirable to have the supervising directors involved in the management of the company (Jungmann 2006, Maassen 1999). Many researchers have tried to find which of the two board structures is performing the best.
In his research Jungmann (2006) empirically tests the effectiveness of both the models by comparing companies from Germany with companies from the United Kingdom. Companies in Germany apply the two-tier board structure and companies in the United Kingdom apply the one-tier board structure. Jungmann tries to find if one of the two models can be seen as most efficient and superior. His paper examines the relationship between the board turnover and the financial performance of companies listed on the stock exchange in Germany and the UK. In each country, he randomly selects a sample of 25 companies that are listed on the London Stock Exchange or the German Stock Exchange for the period 1994-2003. The total sample consisted out of 397 observations. In order to test for efficiency he uses board turnover. Board turnover is defined as the number of executive directors who leave the board during the financial year, divided by the total number of board members at the beginning of the financial year. He uses the loss for the financial year, loss of profits, abnormal negative share price performance and a dividend cut or omission as indicators of bad financial performance. If two of the four financial performance measures are positive, a company is considered to be “performing poorly”. The first hypothesis he tests is: “There is an inverse correlation between board turnover and financial performance”. This hypothesis tests if the non-executives directors in the UK and the member of the supervisory board in Germany effectively exercise their power to unseat the management. The second hypothesis Jungmann tests is: “It is impossible to deduce the superiority of either of the two systems from the degree of inverse correlations between the board turnover and the financial performance”. In order to test the first hypothesis he compared the average board turnover in years not following a year of “poor performance” with the average board turnover in years following a year that did fall in the definition of “poor performance”. Jungmann found an increase in average board turnover in years of poor performance. In Germany the average
board turnover rose by 3.3% and in the UK it rose by 3.73%. His sample was large enough to be approximated by the normal distribution. He found that the increase was significant at a 99% confidence level. The data supported the hypothesis that the board turnover and firm performance are inversely related. Jungmann states that both systems, the one- tier and the two-tier system, are effective means of control, and non-executive directors, as well as the members of the supervisory board, attend to their duty of removing incompetent directors. Jungmann states that the second hypothesis also is accepted. The data did not allow him to judge the superiority of either of the two board structures. Both show an inverse correlation between the financial performance and board turnover. According to Jungmann this correlation is so statistically significant that it is impossible to come to the conclusion that one system would be more effective than the other. According to Jungmann the strengths and weaknesses of the two systems must always be seen in the context of both business and legal environments. Accounting standards, taxation principles and other circumstances must be taken into consideration. Finding the optimal system depends on the legal, historic and cultural constrains a company encounters.
A closely related empirical research is that of Millet-Reyes and Zhao (2010). They compared the performance of one-tier and two-tier structured companies in France. French law allows firms to choose between the two board structures, which is similar to the law in the Netherlands. Their sample consisted out of 665 firms covering 174 French companies over the period 2000-2004. They used the Operating Cash Flow (OCF), Return on Assets (ROA) and Tobin’s Q as proxies for firm performance and the type of board structure as independent variable. In many French companies related family holds a large proportion of the shares. In their regression they controlled for this fact. In addition Millet-Reyes and Zhao control for differences in size, growth and the amount of leverage between the firms. Industry dummies have been added to the model to control for industry effects but the results on these dummies are not reported. They tested whether the structural differences between the one-tier and two-tier board have an impact on firm performance. They expected the board structure to have a significant impact on the operating and stock performance of the French companies. Millet-Reyes and Zhao find a significant positive relationship between both the OCF and ROA and the fact that a company applies a two-tier board structure. A 5% significance level is used. Between the Tobin’s Q and the type of board structure no significant relationship is found. These findings indicate that the choice for a two-tier board structure would improve the OCF and ROA of a company. In addition Millet-Reyes and Zhao introduce regression that includes a dummy variable that indicates if the company is owned by institutional investors. These
institutional investors, mostly banks, represent both equity owners and lenders. They state that the ownership by large institutional shareholders, like banks, could improve firm performance by optimizing the cost of debt and their active monitoring. On the other hand these banks could be increasing the cost of debt if they will misuse their inside-position. The regressions including the institutional investors dummy shows different results. A significant negative relationship is now found between the Tobin’s Q and the two-tier board structure. In contrast to the first regression, the second regression shows a significant negative relationship between the two-tier board structure and the OCF. The second regression showed no significant relationship between the ROA and the two-tier board structure. The results of the research of Millet-Reyes and Zhao stated that there is a significant difference in the performance of two-tier structured firms compared to one-two-tier structured firms in France. In the first regression the two-tier board structure was improving the firm performance. However, when the dummy variable indicating institutional block holders was added, a one-tier board was preferred above the two-tier board structure. This finding indicates that the institutional block holders are in a position to manipulate the two-tier boards. French institutional investors, mostly banks, are able to misuse their position when the board structure is more complex and less transparent, like in the two-tier board structure. Concluding, Millet-Reyes and Zhao find a significant relationship between the type of board structure and the performance of the firm. However, these findings of the two regressions are inconsistent. In their first regression they found that a two-tier structure is preferred to a one-tier board structure. In their second regression the contrary was found. Millet-Reyes and Zhao do not find consistent results indicating that one of the two board structures can be seen as superior.
Dutch companies are allowed to choose between the two board structures similar as in France. This research is trying to find if the fact that the company has a choice in board structure, compared to a company without this choice, is related to the performance of a company. It is examined if companies in the Netherlands, which have a choice in board structure, perform differently from companies in the United States, which do not have this choice. In contrast to Millet-Reyes and Zhao (2010), the relationship between the freedom of choice and the firm performance is the main research interest. Millet-Reyes and Zhao were interested in the difference in performance between the two board structures in France. The effect of the freedom in choice of board structure is examined by comparing the firm performance of Dutch and American companies of similar size and in the same industry.
In order to test the effect of the freedom of choice in board structure, the following hypothesis is designed:
(1): There is a significant relationship between the fact that a company is
Dutch, and therefore has a choice in board structure, and firm performance, as compared to similar American companies.
If a significant relationship is found then this relationship will be analyzed. If no significant relationship is found then possible reasons for this will be discussed.
In line with the findings of Millet-Reyes and Zhao (2010), who found that none of the two board structures could be seen as superior, the second hypothesis is designed:
(2): The appliance of a one-tier board structure, versus the two-tier board structure, does not generate differences in the firm performance.
After the hypotheses are tested, this thesis will discuss relevant results and will make suggestions for further research.
3. Research Method
This section will discuss the research methodology used in this thesis. The first subsection gives information on the selection of the sample, data and the timespan that is used in this research. The second subsection will discuss the variables of interest. Finally the research method is presented.
3.1. Sample, Data and Timespan
This thesis is trying to find a relationship between the fact that a company is Dutch, and thus has a choice in the type of board structure, and the performance of a firm, as compared to similar American companies, that do not have this choice. The sample that is used contains the years’ 25 Amsterdam Exchange Index (AEX) public listed firms and 25 comparable S&P 500 listed firms of that year, for the period 2009-2012. The AEX is composed out of the 25 companies with the biggest market capitalization in the Netherlands. The composition of this index changes over time. The changes in the AEX for the period 2009-2012 can be found in Table 1a in the Appendix. A list of companies in the Netherlands applying a one-tier board structure can be found in Table 1b in the Appendix.
The AEX listed firms are linked to similar S&P 500 firms on the basis of size, measured by market capitalization, and industry. The firms included in the sample can be found in Tables 1c-f in the Appendix for the years 2009-2012 respectively. In some cases the US linked companies considerably differ in size, the main reason for this is that there is only one firm on the S&P 500 that is in the same industry as the Dutch company. Remarks about the differences are given in the Tables in the Appendix. The total sample consists out of 83 companies of which some are in the sample for multiple years. In total, 196 observations are included in the sample, 4 of the 200 observations were left out because of missing data. The performances measures of the Dutch company Aperam could not be found for 2012. For the United States, the performance measures of the McGraw-hill Companies were incomplete for the period 2009-2011.
Financial data about the performance of the performance of the firms is obtained from the Wharton Research Data Services database (WRDS). WRDS contains Bureau van Dijk that contains AMADEUS, a database with data on company financials. Data concerning the type of board structure and the composition of the board was found in the Board Indices of Stuart (Stuart 2008-2010) that provide detailed information about the public listed firms’ boards in
the Netherlands and the United States. Missing information was found in the annual reports of the in the sample included firms.
Data on the period December 31st of 2009 – December 31st of 2012 is used in this research. 2009 was the first year after the Dutch Corporate Governance Code was revised. The year 2013 is left out because of the unavailability or incompleteness of the information.
3.2. Variables of Interest
In this section the variables of interest of this thesis will be discussed. The dependent, independent and control variables of this research will be presented.
3.2.1. Dependent variables
Millet-Reyes and Zhao (2010) use three different dependent variables: Tobin’s Q, the OCF and the ROA. The OCF is a cash flow-based performance measure, the Tobin’s Q is a market-based performance measure and the ROA is an accounting-market-based performance measure. Dybrig and Warachka (2012) find that the relationship between firm performance and Tobin’s Q suffers from endogeinity. They state that underinvestment leads to inefficiency and lower firm performance but increases Tobin’s Q. Underinvestment could inflate Tobin’s Q and for this reason this performance measure is not used in this thesis. The Financial Accounting Standard Board (FASB), an organization designed by the SEC for setting accounting standards in the United States, state that the use of accounting based valuation improve the ability of earnings to measure firm performance (FASB, 1978). They state that accrual based measurements provide a better indication of enterprise performance than cash flow-based performance measures. For this reason the OCF is not used as dependent variable in this thesis.
Similar to Millet-Reyes and Zhao (2010) the ROA is used as dependent variable in this thesis. The ROA gives an indication of how profitable a company is in proportion to its total assets. It provides information on how efficient the assets are used by the management of the company to generate income. The ROA is calculated by dividing the net earnings of a company by its total assets. To check for robustness, an additional regression will be
performed using the ROE as dependent variable. The ROE is an indicator of how profitable a company is in proportion to the total equity that is provided by the shareholders. The ROE gives an indication of how efficient the money invested by the shareholders is used. The ROE
is calculated by dividing the net earnings of a company by the shareholders equity. The difference between the ROA and ROE is that the ROA also takes account of how well the liabilities of a company are used. The advantage of using the ROA and ROE is that companies of different size can be compared. The ROA and ROE should be strongly correlated and differences between the two performance measures can be explained by the level of liabilities a company has. Both the regressions are expected to show consistent results.
3.2.2. Independent Variables
In this research two independent variables are used. First there is a dummy variable (“NL”) that indicates in which country the company is situated and thus, to which law the company is subjected. This variable is used to test the first hypothesis and will show if there is a relationship between the firm being Dutch and the performance of the firm. A significant relationship is expected between this variable in relation to the firm performance. When looking at the literature, which states the best corporate board structure relies on the constraints a company is subjected to, this relationship is most likely to be positive (Jungmann 2006). Dutch law allows companies to choose between the two board structures in contrast to the US law that states a company should apply a one-tier board structure. This freedom of choice in board structure should result in a positive relation to the firm’s performance. It is expected that the relation on the “NL” variable will be positive.
The second independent variable of interest is a dummy variable (“CBS”) indicating which kind of board structure is applied on the company. In their research Millet-Reyes and Zhao (2010) used a similar dummy variable. This variable is added to test the second hypothesis and will show if there is a relationship between the firm applying a one-tier board and the performance of the firm. It is expected that the one-tier board structure does not generate significant differences in the ROA, compared to the two-tier board. This expectation corresponds to the findings of Millet-Reyes and Zhao (2010).
3.2.3. Control Variables
A number of control variables are included in this thesis. These variables control for other things that might influence firm performance.
According to Zeitun and Tian (2007) there is a significant positive relationship between the ROA and ROE of a firm and the size of the firm. For this reason a control variable (“MarketCap”) is added to control for the differences in market capitalization between the firms. The relationship between this variable and the firm performance is expected to be positive. This control variable will also control for the possible mismatch in market capitalization that occurred when the Dutch companies were linked to the American companies.
Yermack (1996) finds that firms with a smaller board work more efficiently and are higher valuated than those who have a large board. He states that there is an inverse association between the size of the board and firm value. Therefore a variable (“BoardSize”) is added to control for the differences in board size. Based on the existing literature the relationship between this variable and the firms’ performance is expected to be negative. Yermack (1996), Agrawal and Knoeber (1996) and Klein (1998) all report a significant negative correlation between the proportion of independent directors and the performance of the firm. Although these studies use the Tobin’s Q to measure the firm performance it might be necessary to control for the proportion of independent directors on the board. For this reason the “Independent” variable is added. This variable represents the proportion of independent directors on the board. Following the literature this variable should have a negative relationship with the firms’ performance.
Selling and Stickney (1989) state that the ROA is industry dependent. The ROA levels differ across industries, for example, according to their levels of leverage. For this reason a group of variables is added to control for the different industries (“Industry”). The companies are divided into 18 different industry categories. For each of the industries a dummy variable has been made. Differences between industries are expected.
3.3. Research Method
An OLS regression is used to measure the relationships between the country of residence, the board structure and the performance of the company. The regression used can be compared to a Difference-in-Difference research method in which the US firms can be seen as the control group and the Dutch firms can be seen as the treatment group. Dutch and American companies are linked by industry and size. In the regression several control variables are included to control for other thing that might influence firm performance. By using this experimental approach this thesis is trying to isolate the relationship between the fact that the
company has the freedom in choice of board structure, and the performance of the company. Many researches have used a similar experimental approach. Card en Krueger (2005) compared the employment in the fast-food sector in New Jersey and Pennsylvania. They observed the change in employment after a minimum wage increase in New Jersey. Comparing the employment before and after the increase in minimum wage was suffering from omitted variable bias. By including Pennsylvania as control in the difference-in-difference model, the bias caused by variables common to New Jersey and Pennsylvania were controlled for. The same method is now applied to Dutch companies, using the American companies as the control.
The difference between this research and that of Millet-Reyes and Zhao (2010) is that their research does not compare the performance of French companies with companies in a country that do not have a choice in corporate board structure. Their research tries to find which of the two board structures performs best in France.
The following model is used to measure the relationships with the firm performance:
The dependent variable is the ROA and is given in percentages. The independent “NL” variable is a dummy variable indicating the country of residence; it takes on the value 1(0) if the company is Dutch (American). The coefficient is indicating the relationship between the country of residence and the firm performance. The independent “CBS” variable is a dummy variable indicating the Corporate Board Structure that is applied to the firm. This variable takes on the value of 1(0) for companies applying a one-tier board structure (two-tier board structure). The coefficient is indicating the relationship between the type of board structure and the firm performance. The variable “MarketCap” is a variable controlling for differences in size of the companies. Similar to Millet-Reyes and Zhao (2010) a logarithm of the firm size is used. Firm size is measured by the total Market Capitalization of the company in dollars, which is calculated by multiplying the numbers of shares outstanding by the share price. The coefficient shows the relationship between the size of the company and the firm performance. The independent variable “BoardSize” is the variable controlling for differences in the total number of directors. Similar to Yermack (1996) a logarithm of the size of the board will be used. The coefficient will show the relationship between the number of directors on the board and the firm’s performance. The “Independent” variable is the variable controlling for the differences in the proportions of independent directors on the board. The
directors that are labeled as independent can be found in the annual reports of the companies. A director that is independent does not have a monetary or pecuniary relationship with the firm except for his commission fees. This variable is calculated as the proportion of the total number of directors that are independent. The coefficient will show the relationship between this variable and the firm’s performance. The last control variable “Industry” controls for the effects the differences in industries have on the firm’s performance. The companies in the sample have been divided into 18 different categories. For each of these categories a dummy variable is created. The dummy variables can be found in Table 2d in the Appendix. The coefficient on the dummy variable shows the relationship the industry has with the firm’s performance. The “ “ term in the regression indicates the error term.
The coefficient is used to test the first hypothesis and the is used to test the second hypothesis. For both significant regression results are trying to be found. To check the robustness of the findings of the model, a different proxy will be used for the measurement of the firm performance. This regression will be performed with the same model except for the ROA that will be replaced by the ROE. This robustness check is performed to see if the results are robust for different measures of performance. ROA and ROE are usually strongly correlated and therefore no differences in results are expected. In addition, a check will be performed to see if the use of a pooled regression leads to problems. The observations used might be correlated over time since some of the companies are in the sample for multiple years.
4. Results
This section will present and analyze the results of the regression. In the first subsection the descriptive statistics concerning the most interesting variables will be given. In the second subsection the results of the regression will be presented with regard to the hypotheses that are tested in this thesis. Finally the results of the robustness checks will be shown.
4.1. Descriptive Statistics
In this section the means and standard deviations of the variables of interest are discussed. Table 2a in the Appendix reports the sample means and standard deviations of the independent, dependent and control variables for the years 2009-2012 separated by country of residence and corporate board structure. In addition the minimum and maximum vales of the observations are given. In Table 2b and Table 2c in the Appendix the means and standard deviations of the most important variables are given for the Netherlands and the United States separately.
The statistics show that the average market capitalization was not very different between the two countries. This indicates that the linkage of the companies was successful. There is quite a big difference in the size of the firms applying a one-tier board structure and firms applying a two-tier board structure. As can be found in Table 2a in the Appendix, the mean of the market capitalization of one-tier board structured firms is more than 12 billion higher than that of two-tier board structured firms. This can be explained by looking at Table 2b in the Appendix, which shows the market capitalization of companies situated in the Netherlands. In the Netherlands, the difference between the average market capitalization of companies that choose the one-tier board structure is more than 35 billion higher than that of companies choosing the two-tier board structure. The companies in the Netherlands choosing a one-tier board are on average much larger than companies choosing the two-tier board. There are no major differences in the average number of directors appointed by the firms comparing the Netherlands and the United States. However, the average percentage of independent directors is approximately 20% higher in the United States compared to the Netherlands. This explains the higher average on one-tier boards since the United States only applies one-tier board structures. This finding indicates that the proportion of directors that is assigned with the role of monitoring the management is higher in the United States compared to the Netherlands.
The difference in the ROA between the two countries is around 2% and the difference in the ROE is around 1%. When looking at companies in the Netherlands, the ROA and ROE of companies choosing a two-tier board structure is are higher. The statistics show that Dutch companies applying a one-tier board on average are much larger in market capitalization but their ROA and ROE are on average lower compared to Dutch companies applying a two-tier board structure. As the literate states, the ROA and ROE vary by industry. Since some industries include more companies than other industries in the sample this might influence the average ROA and ROE. The average ROA and ROE per industry can be found in Table 2d in the Appendix.
Table 2e in the Appendix shows the correlations between the variables. Some remarkably high correlations are found. The “NL” variable is highly correlated with the “CBS” variable. The reason for this correlation is the fact that only a small part of the Dutch companies applies a one-tier board structure. The “CBS” variable is highly correlated with the “Independent” variable. This correlation is explained by the higher average percentage of independent directors in the United States.
4.2. Regression Results
In this section the results of the regression will be discussed. Table 3a provides a shortened version of the results of five different regressions that were performed on the first model, using the ROA as dependent variable. A full version of the results of the first regression can be found in Table 3b in the Appendix. The results of the regression using the robust standard errors can be found in Table 3c in the Appendix.
In Table 3a the results of five different regressions can be found. The first regression is the regression that is done on the full model. The second regression is done on the whole model excluding the control variables for the differences in industry. The third regression is done on the full model excluding the variable controlling for differences in the proportion of directors that is independent. The fourth regression is done on the full model excluding the Industry dummies and the “Independent” variables. The fifth regression is performed on the model excluding all the control variables.
The first regression, performed on the full model, shows a 10% significant negative relative relationship between the “NL” variable and the ROA. Indicating that if the company is Dutch, the ROA is lowered by 2.96%. Furthermore, a positive relationship between the size of the company and the ROA follows from the regression. This relationship is significant at a
1% significance level. Hence, increasing the market capitalization by 1% would lead to a 1.46% increase in the ROA. The finding of this positive relationship corresponds to the findings of Zeitun and Tian (2007). It follows from the regression that the size of the board is negatively related to the ROA, on a 5% significance level. This result concludes that if the number of directors would increase by 1% this would lead to a 4.55% increase in ROA. This negative relationship corresponds to the findings of Yermack (1996). All, except for one, coefficients on the industry control variables are significant as can be found in Table 3b in the Appendix. These findings correspond to that of Selling and Stickney (1989). Remarkably is that the coefficients on the dummy variables for industries are considerably larger than those of the other variables. No significant coefficient was found on the “CBS” and the “Independent” variables. The regression shows an adjusted R2
of 0.46.
Table 3a. Regression Results: NL vs US Companies, ROA as Dependent Variable
N=196 (1) (2) (3) (4) (5)
Variable Coefficient Coefficient Coefficient Coefficient Coefficient
Constant -2.327126 9.690525* -2.517668 13.78965*** 9.149252*** (6.367324) (5.751862) (5.535871) (4.797683) (1.658168) NL -2.961234* -2.961385* -2.91314** -3.944029*** -3.78382** (1.598675) (1.626452) (1.387586) (1.438555) (1.521678) CBS -1.614478 -3.794032** -1.613288 -3.781206** -2.122654 (1.525504) (1.466303) (1.521007) (1.468786) (1.545022) MarketCap. 1.464129*** 1.916859*** 1.4614*** 2.037298*** (.4111806) (.4063479) (.4075768) (.3961033) BoardSize -4.549908** -9.015929*** -4.536557** -9.450706*** (1.849009) (2.047003) (1.830795) (2.022399) Independent -.0023653 .0512136 (.0387019) (.0397942)
Ind. Control YES NO YES NO NO
R2 0.5199 0.1942 0.5199 0.1872 0.0385
Adjusted R2 0.4589 0.1730 0.4620 0.1701 0.0285
Notes: NL= 1(0) if company is Dutch (American). CBS=1(0) if one-tier (two-tier) type. The “MarketCap” and “BoardSize” variables are both a natural logarithm. *, **, ***Significant at the 0.10, 0.05 and 0.1 levels, respectively. Standard errors are in parentheses.
The second regression leaves out the control variables for the industry effects. Leaving out these dummy variables leads to changes in the findings of the regression. The regression now shows a significant relationship between all the variables except for the “Independent”
variable. A negative relationship between the “CBS” variable and the ROA follows from the second regression. This relationship is significant at a 5% significance level. The coefficient on the “BoardSize” variable changes considerably. From the second regression follows a negative relationship between the size of the board and the ROA. This relationship is significant at 5% significance level. The “NL” and “MarketCap” variable still show a significant relationship at a 10% and 1% significance level respectively.
The third regression leaves out the “Independent” variable that is controlling for the proportion of independent directors on the board. The regression shows a positive significant relationship between the “NL” variable and the ROA. Excluding the “Independent” variable leads to a rise in the significance of coefficient on the “NL” variable that is now significant at a 5% level. Both the positive “MarketCap” and negative “BoardSize” variables are significant on a 1% and 5% significance level respectively. All of the industry control variables are significant, except for one. No significant relationship was found on the “CBS” variable. The fourth regression leaves out both the control variables and the “Independent” variables out of the regression. All of the variables now show a significant relationship at a 1% significance level, except for the 5% significant relationship on the “CBS” variable. The directions on the relationships stay unchanged.
Excluding the “Independent” variable and the control variables for industry effects leads to significant results on the “NL” variable. The fact that the model improves when the “Independent” variable is dropped can be explained multicollinearity. In Table 2e in the Appendix the correlations between the variables can be found. The table shows that the “NL” and “Independent” variables are strongly correlated. For this reason a regression including both of these two variables will lead to biased results. Looking at the second regression, leaving out the control variables for control, the findings on the variables of interest changed substantially. The significant coefficients on the dummy variables for the different industries, as can be found in Table 3b in the Appendix, indicate that a large part of the ROA is explained by the industry that the company is in. Leaving out the industry variables will lead to omitted variable bias. For these reasons the third regression shows the highest adjusted R2 and is the most reliable model.
4.3. Testing the Hypotheses
The first hypothesis that is tested is: “There is a significant relationship between the fact that a company is Dutch, and therefore has a choice in board structure, and firm performance, as
compared to similar American companies.” The relationship is expected to be positive. Looking at the coefficient on the NL variable, the regressions show a negative relationship. The third regression performed shows 5% significant negative relationship. These findings indicate that the fact that the company is Dutch will result in a lower ROA. The hypothesis that a significant relationship between the fact that a company is Dutch and the firm performance exist is supported by the results of the regressions performed on the first model. However, the relationship found is negative as a positive relationship was expected. The second hypothesis that is tested is: ”The appliance of a one-tier board structure, versus the two-tier board structure, does not generate differences in the firm performance.” As can be found in Table 3a, no significant relationship was found on the “CBS” variable in the third regression. This insignificant result does provide enough information to make conclusions with regards to the second hypothesis. It follows from the regressions that the relationship between the “CBS” variable and the ROA is largely explained by the type of industry the firm is in.
4.4. Robustness Checks
In order to see how the effects on the “NL” and “CBS” variables change when the regression is modified, another regression is performed on an adjusted model. The ROE replaces the ROA, as the dependent variable. Since some of the companies are included for multiple years it could be possible that the observations are correlated over time. The model will be tested for each year separately to check if the findings of the full sample will differ with the findings for each year individually.
4.4.1. Robustness Check using the ROE
In order to check if the findings of the first model are robust for different performance measures, a second model has been designed:
Table 3d provides a shortened version of the results of five different regressions that were performed on the second model, using the ROE as dependent variable. A full version of the
results of the first regression can be found in Table 3e in the Appendix. The results of the regression using the robust standard errors can be found in Table 3f in the Appendix.
In the first regression, performed on the full model using the ROE, none of the coefficients on the variables is significant except for the control variable for the differences in size and the control variables for industry. Five of the industry variables are non-significant the other thirteen are significant.
Similar to the ROA model, excluding the control variables for industry-effects lead to changes in the results. The second regression shows that all the coefficients are significant except for the “NL” variable. The “CBS” variable is negatively related to the ROE on a 5% significance level. The effects on the control variables are the same for the ROA as for the ROE. The coefficients of the control variables are all significant at a 5% level. The “MarketCap” variable is significant at a 1% level. Noticeable is that, in contrast to the regression using the ROA, the “Independent” variable is now significant at a 5% level. However, leaving out the “Independent” variable, as in the third regression, does not lead to major changes in the effects on the other variables.
Table 3d. Regression Results: NL vs US Companies, ROE as Dependent Variable
N=196 (1) (2) (3) (4) (5)
Variable Coefficient Coefficient Coefficient Coefficient Coefficient
Constant -20.80725 .3883907 -16.15478 17.64304 21.16079*** (15.40535) (14.00184) (13.40828) (11.77907) (3.913978) NL .2475603 -.5349069 -.9267522 -4.671199 -3.631403 (3.867897) (3.959294) (3.360832) (3.531879) (3.591804) CBS -2.339784 -9.009122** -2.368851 -8.955131** -5.628827 (3.690865) (3.569443) (3.683987) (3.606101) (3.646907) MarketCap. 3.260742*** 3.454516*** 3.327377*** 3.961488*** (.9948265) (.9891784) (.9871803) (.9724958) BoardSize -3.858369 -11.08554** -4.184362 -12.91567** (4.473567) (4.983048) (4.434317) (4.965306) Independent .0577527 .2155758** (.0936368) (.0968715)
Ind. Control YES NO YES NO NO
R2 0.4825 0.1206 0.4813 0.0976 0.0134
Adjusted R2 0.4166 0.0974 0.4187 0.0787 0.0032
Notes: NL= 1(0) if company is Dutch (American). CBS=1(0) if one-tier (two-tier) type. The “MarketCap” and “BoardSize” variables are both a natural logarithm. *, **, ***Significant at the 0.10, 0.05 and 0.1 levels, respectively. Standard errors are in parentheses.
The third regression shows no significant relationships on the “NL” and the“CBS” variable. The “MarketCap” variable is significant at a 1% significance level. The third regression is the model with the highest adjusted R2 and the most reliable for the same reasons as for the ROA-model. Noticeable is that, comparing the results of the ROA-model with the ROE-model, the NL variable no longer is significant.
4.4.2. Testing the Hypothesis
With regard to the first hypothesis, no significant coefficient was found on the “NL” variable. Neither a significant relationship was found on the “CBS” variable, which was designed to test the second hypotheses. The results of the regression using the second model do not provide enough information to make conclusions with regards to the hypotheses.
4.4.3. The Regressions for each Year Individually
In order to check if the fact that some of the companies are in the sample for multiple years is a problem regressions have been performed for each year individually. Performing a pooled regression using panel data could lead to biased estimates. In Tables 4a-d in the Appendix, the regression results for each separate year can be found. The first model, using the ROA as the dependent variable, is used.
The regression performed on the data of the years 2009-2012 showed a 5% significant negative relationship between the “NL” variable and the ROA. Furthermore significant relationships were found both on the “MarketCap” and “BoardSize” variables. The regressions for each year separately however do no support these findings. The years 2010 and 2011 show a significant negative relationship between the “NL” variable and the ROA, the years 2009 and 2012 do not. The coefficients on the “MarketCap” and industry variables are significant for most of the years. The findings of the pooled regression that was performed on the first model are inconsistent with the findings of the regressions for each year separately.
