Board size: Does one size fit all?
Student: Soufyan el Hamdi
Thesis title: Board size: Does one size fit all? Student number: 10662200
Date final version: 15-06-2015 Word count: 12,642
MSc Accountancy & Control, variant Control Amsterdam Business School
Faculty of Economics and Business, University of Amsterdam Supervisor: Dr. Bo Qin
Statement of Originality
This document is written by student Soufyan el Hamdi 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.
Abstract
Recent corporate scandals have led to the attention of criticizing the U.S. corporate governance system. An important factor throughout several studies with regards to corporate governance is board size. This because of the fact that the board of directors plays a central role in the corporate
governance of companies. Recent studies shows mixed results with regard to board size and if a one size fit all approach will work. I propose and test hypothesis on the impact of determinant of firm complexity on board size using a sample of 1,814 observations. The first hypothesis shows results that organization complexity indeed has a significant impact on board size which means that
organizations that are more complex tend to have bigger boards and simple firms tend to have smaller boards. I also find that the mismatch between board size and firm complexity does not have a
significant impact on future firm performance. But when testing this by looking at two different aspects, undersized and oversized, I do find significant results that undersized boards do have a positive significant impact on future firm performance. These results suggest that firm consider their complexity and that undersized boards can have a positive impact on future firm performance.
Contents
1. Introduction ... 5
1.1 Background ... 5
1.2 Research question ... 7
1.3 Motivation ... 7
1.4 Summary findings ... 7
1.5 Contribution ... 8
1.6 Structure of the thesis ... 8
2 Literature review ... 9
2.1 Agency theory... 9
2.2 Corporate governance ... 9
2.3 Board structure ... 10
2.4 Board size ... 11
2.5 Organizational complexity ... 12
3. Research methodology ... 14
3.1 Sample and timeline ... 14
3.2 Board size and firm complexity variables ... 14
3.3 Model and control variables ... 15
4. Empirical analysis ... 17
4.1 Descriptive statistics ... 17
4.2 Organizational complexity analysis ... 18
4.3 Subsequent firm performance ... 21
4.4 Supplemental analysis ... 25
5. Conclusion ... 27
5
1. Introduction
This first paragraph of this chapter will discuss background information with regard to corporate governance and especially board size. The second paragraph discusses the research question. The third paragraph will give a summary of the findings. The fourth paragraph presents the motivation of this study and the last paragraph gives an overview of the structure of this study.
1.1 Background
There are multiple reasons why corporate governance can be an important tool during a financial crisis. One reason, which is mentioned in prior literature, is that companies with low quality corporate governance will lead to losing more values than firms with strong corporate governance (Erkens, Hung and Matos, 2012). During recent years big companies, like WorlCom and Enron, with low quality corporate governance have lead to big financial scandals. This scandals lead the attention to criticizing the U.S. corporate governance system. The Sarbanes–Oxley Act and the new listing requirements were a reaction to a series of dramatic corporate and accounting. Much of the blame for these scandals was put on boards of directors (Adams, 2012). It was important to have more
regulatory and more legislative changes, such as the Sarbanes Oxley Act 2002 (Holmstrom and Kaplan, 2005). A second reason is that firms that apply strong corporate governance are less likely to manage their earnings. According to prior literature high quality corporate governance leads to higher levels of earnings quality (Bekiris and Doukakis, 2011). They used in their research 55 criteria to evaluate the corporate governance quality of 427 firms. The main findings indicated that the strong corporate governance mechanisms increase the credibility of the financial reporting, which was clearly seen with large and mid-cap companies (Bekiris and Doukakis, 2011).
An important definition of corporate governance is a set of mechanisms and processes that help ensure that companies are directed and managed to create value for their owners, while fulfilling responsibilities to other stakeholders (Merchant, 2007). The Cadbury Report (1992) has a different definition of corporate governance, in this report it is defined very general as the process by which companies are directed and controlled. Another definition from prior literature is the one of Monks and Minow (2011) that suggest that corporate governance is the relationship among various participants in determining the direction and performance of corporations. The primary participants are the
shareholders, the management and the board of directors. The fact that many definitions of corporate governance exist highlights that there are different perspectives taken by different interest groups as well as different level of analysis, some of which focus on the parties involved while others focus on the context in which corporate governance is applied. If we look further in to prior literature there are a number of corporate governance variables examined in more depth. Board independence and board size have been used to measure corporate governance. Board size is seen as an important factor in determining monitoring and supervision the power of the board ( de Andres et al., 2005). These variables determine good corporate governance. Furthermore, corporate governance influences how investors view the company. Thus, companies that have good corporate governance are viewed as more attractive companies for investment (Shleifer & Vishny, 1997).
6
An important factor throughout several studies with regards to corporate governance is board size. This because of the fact that the board of directors plays a central role in the corporate
governance of companies (Guest, 2009). A lot of pressure has been put in choosing smaller boards because larger boards size suffers from coordination and communication problems. Larger boards will therefore have a negative impact on board effectiveness and this will lead to worse firm performance (Lipton and Lorsch, 1992; Jensen, 1993). The research of Guest (2009) finds empirical evidence that documents a significantly negative relation between board size and firm performance which means that lager board size causes negative firm performance and that suggests that a one size fit all approach could improve firm performance and good quality corporate governance. Furthermore, A common interpretation about the negative relation between board size and firm performance is that many boards are inefficient and that a ‘one size fit all’ approach to board size would improve performance for such firms (Jensen, 1993).
According to other studies (Coles et al., 2008; Lehn et al., 2004 and Boone et al., 2007) this conclusion and suggestion is not always the case. In their research there is documented that board size is determined by specific firm variables, such as Tobin’s Q and profitability. So a one size fits all approach would not work in all cases. Prior studies have been criticized because since firm
performance has a negative impact on board size there is not enough controlled for endogeneity problems (Wintoki, 2007). The research of Wintoki (2007) shows generalized methods of moments estimator that adjust board size to previous performance, but finds no relationship between board size and firm performance. Prior literature also shows that board size is determined by specific firm
variables which mean that the impact of board size on firm performance may differ according to these variables. In line with this is the research of Coles et al. (2008) where evidence is found that the impact of board size on firm performance is for larger firms positive and this suggest that larger boards may lead to maximizing value for those firms. In this case a one size fits all approach would not work.
The study of Faleye (2007) is also consistent with this by mentioning that a push toward a common board structure may be counterproductive because it doesn’t look at the role of specific firm characteristics in determining the appropriate board size. The research of Faleye (2007) looks at leadership structure and argues that organizations ignore the possibility that differences in specific firm characteristics determine the appropriateness of combining or separating two positions. Prior literature argues that the choice of leadership structure has to do with the board that tries to attempt to balance entrenchment, avoidance and unity of command (Finkelstein and D’Aveni, 1994). Consistent with this is the Business Roundtable (2002, p. 11) which mentions that each organization need to determine its own leadership structure given its present and anticipated circumstances. So a one size fits all approach would also not work in this case.
However, the relationship between board size and firm performance may also differ by national characteristics (Guest, 2009). Looking at different countries with different institutional backgrounds, the boards’ functions are different and therefore the expected relation between board size and firm performance may differ. Looking at other countries is useful in understanding the complete relation between board size and firm performance but the few non-US empirical studies show small or cross-sectional samples.
7
1.2 Research question
In the previous section various issues have been raised about board size and if a one size fit all approach will work. Prior literature has shown mixed results between these two concepts. This relationship will be examined by looking at the complexity of firms and if that influences the size of the board. It is very interesting because of the limited available literature concerning this subject.
The aim of this paper is examining whether a one size fit all approach will work. For this paper, I will investigate what the determinants are for board size and based on those determinants empirical measure whether there is evidence for a one size fit all.
Therefore the research question will be:
- Does one board size fit all?”
1.3 Motivation
This research will have as purpose providing empirical evidence if a one board size fit all.Much has been written about the impact of corporate governance on firm performance, but the impact of board size on firm performance and if a one size fit all approach is achievable has shown mixed results. A factor that is perceived to affect the board’s ability to effectively function is the size of the board of directors. The research of Lipton and Lorsch (1992) and Jensen (1993) provide empirical evidence that larger boards are less effective than smaller board. They mention in their research that it is because of coordination problems and director free-riding. The research of Yermack (1996) and Eisenberg et al., (1998) are in line with this and provide significant evidence that smaller boards lead to higher firm value, which is measured by Tobin’s Q. The purpose of this study is to examine the reasoning and data behind the wisdom that a one board size fit all and whether bigger or smaller boards lead to better performance. I argue that complex firms such as large and leveraged ones have greater advising needs and require larger boards. According to the research of Dalton et al., (1999) larger boards bring more advising requirements, knowledge and experience and provide better advice for complex firms. So complex firms should have larger boards and simple firms should have smaller boards. This study will determine whether the complexity of a firm is associated with the size of the board and whether the mismatch between board size and firm complexity leads to higher firm value.
1.4 Summary findings
I seek to fill the gap with regard to a common board size by developing hypothesis on the
determinants of organizational complexity. This hypothesis focuses on how organizational complexity effects the relative cost and benefits of the size of the board. I test this hypothesis on a sample of 1,814 firms and find significant evidence that firms choose their board size based on the complexity of their firms. Specifically, I find evidence that complex firms have bigger boards than simple firms which suggest that a one size fit all approach would not work. After controlling for potential biased results based on principal component analysis and standardization I still find significant evidence that the size of the board is impacted by the complexity of the firm. These results raise the important question whether the mismatch between board size and organizational complexity has an impact on future firm
8
performance. When testing the mismatch on future firm performance I do not find significant evidence but when testing undersized and oversized board on the impact on future firm performance I find significant evidence that undersized board have a positive impact on future firm performance which indicates that smaller boards based on their faster decision making process and less communication gap leads to better future performance.
1.5 Contribution
This research will contribute to what is already been written about board size by looking at the composition of board size based on the complexity of a firm over a long time period. The long time frame will ensure that the results are less likely to be biased and allows using appropriate econometric methods to control for endogeneity (Guest, 2009). This research will undertake two different methods to control for potential biased results, namely principal component analysis and standardization process. This research will have a different approach compared to the research of Faleye (2007). Faleye (2007) looked at leadership structure and argued that organizations ignore the possibility that differences in specific firm characteristics determine the appropriateness of combining or separating two positions. He concludes in his research that a one size will not fit all. This study will develop further on that, by examining the size of the board and try to empirical measure if a one size fit all approach is achievable. This is also the biggest contribution because this study will challenge the uniformity that is usually set by corporate governance code, i.e., if a one size fit all? The second contribution will be on the large literature on board size by showing the relation between board size and the complexity of a firm over a long time period. But also looking at the mismatch between board size and firm complexity and whether this effects future firm performance. The research of Hermalin and Weisback (2003) mentions that the mixed results of the impact of board size is a notable finding in the literature.
1.6 Structure of the thesis
The remainder of the thesis is organized as follows. In the second chapter, literature on board size will be discussed and reviewed. Based on this chapter the researched hypotheses will be developed. The third chapter will give a description of the used methodology and also explains how to obtain the data and the different models. In the fourth chapter the empirical results of the thesis will be presented and discussed. The fifth chapter gives a conclusion and some suggestions for future research.
9
2 Literature review
The focus of this research is to investigate if a one size fit all. The issue of whether board size affects firm performance has been the subject of a large number of studies. Prior literature discussed and showed mixed results with regard to a one size fit all approach. The central concepts in this chapter will be the determinants of board size and the hypothesis development.In this section an overview and insight of board size will be discussed based on available academic literature.
2.1 Agency theory
Agency relationship is between two different parties, which consists of a principal and an agent who represents the principal. It is common that principals delegate decision rights to the agent. The agency problem arises when there is inefficiencies and information asymmetry. The agency theory is
concerned with resolving problems that occur between principals and agents. A first problem is that problems arise when the desires or goals of the principle are not in line with those of the agent. A second problem is that both agent and principal have different risk taking behaviors and because of different risk tolerances, both may take different actions. According to prior studies this might lead to having excessive perk consumptions (Jensen and Meckling, 1976), showing not enough effort in shareholders’ interest (Fama and Jensen, 1983) and making sub-optimal decisions (Farma and Jensen, 1983). An important mechanism to control and monitor agents in order to solve agency problems is the board of directors. When a diverse board of directors have the decision-making process in hands it will be much harder for agents to take actions that are not in the best interest of the principals. An argument for this is that there are different perspectives within the board of directors and this will have a positive impact on the decision-making process because of the minimizing of
informational biases (Westphal and Milton, 2000). Prior literature therefore identifies the board of directors as a mechanism that mitigates the agency problem. Board systems differ among countries, but there are two main systems: one-tier and a two tier board system (Jungmann, 2006).
2.2 Corporate governance
Prior literature has focused on the relationship between corporate governance and firm performance by examining specific corporate governance variables (e.g., board structure, audit committee characteristics) (Bekirisa and Doukakisb, 2011). An important definition of corporate governance is a set of mechanisms and processes that help ensure that companies are directed and managed to create value for their owners, while fulfilling responsibilities to other stakeholders (Merchant, 2007). If we look at prior literature we can say that the majority of previous studies has concentrated on board and audit committee as proxy for corporate governance mechanism (Abed, Attar and Suwaidan, 2011). According to prior literature there are a couple of variables that are examined in more depth with regard to corporate governance. Board independence and board size have been used to measure corporate governance. According to the research of de Andres et al., (2005) board size is seen as an important factor in determining monitoring and supervision the power of the board. Furthermore, corporate governance influences how investors view the company. Thus, companies that have good
10
corporate governance are viewed as more attractive companies for investment (Shleifer & Vishny, 1997). According to prior literature the board of directors plays a central role when looking at the relation between corporate governance and firm performance. The next paragraph will discuss the structures of the board.
2.3 Board structure
Prior literature mentioned the task of the board of directors in different ways. The major role of the board of directors is to monitor and advice top management. The board of directors is supposed to acts as an agent for shareholders by representing the interest of the shareholders (Barkema and Gomez-Mejia, 1998). Jensen (1993) mentioned that the board of directors have a final responsibility for the functioning of the firm. Two main models to organize the board of directors are the one-tier board and the two-tier board. The one-tier board is mostly used in Canada, the United Kingdom and the United States. The two-tier board is mostly used in European countries like Germany and the Netherlands. According to Coles et al. (2008) the board of directors in a one-tier structure monitors and advices top management. In a two-tier board structure the daily management and monitoring of the management is separated between the management board and the supervisory board. The management board does the day-to-day operations of the firm and the supervisory board appoints, monitors and advises members of the management board. The main difference between these two structures is between the formal board roles for the executives and non-executives board members and that the role of the CEO and chairman are always split (Maassen, 2002). The one-tier board has as advantage that it increases information symmetry between executives and non-executives. This will have a positive impact on the knowledge and decision making of the non-executives. According to the study of Cheffins (1997) information symmetry also occurs because of the frequent meetings of the one-tier board compared to the two-tier board. The main disadvantage of a one-tier board is that decisions and monitoring of these decisions are done by the same board members (executive and non executive). According to the study of Cheffins (1997) this will result in a difficulty with regard to
objective monitoring the decisions. The two-tier model is developed to ensure the protection of shareholders and the public’s interest by separating the monitoring and decision making activities between a management board and a supervisory board. A disadvantage of the two-tier model is that non-executive board members not present are at meetings of the executive board members. This will have as consequence that information asymmetry arises between the two boards and this will have negative impact on the quality of their monitoring role.
The one-tier structure is mostly used in countries like the United Kingdom and the United States. A chief executive officer is the position of the most senior corporate executive responsible for initiating and implementing the plans and policies of a company. On the other hand, there is the chairman who is responsible for ensuring that the board of directors provides good quality advises and monitoring of the CEO. The chairman has an important control function and that is why it is often said that the CEO and the chairman should be separated (Faleye, 2007). According to the research of Fama and Jensen (1983) where they mention that when the CEO also serves as a chairman (CEO duality) the principle of separating decision-making and control will be violated and this will affect the
11
board’s function to perform its monitoring function. The research of Jensen (1993) also argues that separating those two positions is essential for the effectiveness of the board. According to
shareholders activists, institutional investors and regulators there appears to be more understanding that the CEO should not serve as a board chairman (Faleye, 2007). The research of Dahya (2004) shows that 15 other countries next to the United Kingdom issued reports where they recommend that the CEO and chairman duties need to separated. The shareholder proposals calling for separating the CEO and chairman position significantly increased between 2001 and 2004. According to shareholders activists and regulators firms will benefit from duality. But because of the non-conclusive evidence a choice system in which economic consideration leads to choosing the right leadership structure is also realistic (Faleye, 2007). The research of Brickley et al. (1997) is also consistent with this where they recognize that there are costs and benefits with regard to separating the CEO and chairman duties. Rechner and Dalton (1991) find in their research that firms where the position of CEO and chairman are separated perform better on a number of accounting measures. They find in their research that the separate positions consistently perform better than firms where the positions are combined. They conclude in their study that their results provide empirical support for the warning that a board structure where the CEO serves as the chairman of the board has a negative impact on the performance of the company. However, Faleye (2007) suggest that separating CEO and the responsibilities of the chairman will be appropriate for some firms, but should not be seen as a standard that can be used by all firms. A one hat may not fit all, but what about a standard board size? The research of Faleye (2007) looked especially at CEO duality, while this research will further
examine this by looking at board size. In the next paragraph the variable board size will be discussed more in detail.
2.4 Board size
Board size can be seen as one of the mechanisms for the effectiveness of the board of directors (Beiner et al., 2004). According to prior literature large boards can result in poor communication and worse decision making (Jensen, 1993; Guest, 2009). Irrespective of other characteristics of board it is argues that board size plays an important role in corporate governance (Jensen, 1993). Guest (2009) finds evidence that support the argument that problems of poor communication and decision-making undermine the effectiveness of large boards. A common interpretation about the negative relation between board size and firm performance is that many boards are inefficient and that a ‘one size fits all’ approach to board size would improve performance for such firms (Jensen, 1993). Yermack (1996) also provides evidence that firm value and performance has a decreasing function on board size. The research of Coles (2008) finds evidence that the relation between firm performance, measured by Tobin’s Q, and board size has a U-shaped model. This suggests that either very small or very large boards are optimal and that this relation arises from differences between complex and simple firms (Coles, 2008). Many literature suggest that firms should have smaller boards. The research of Lipton et al. (1992) finds evidence that a board which consists of eight or nine members are most effective. They mention that a small board will create more cohesion, more productivity and is effective in controlling and monitoring of the organization which is consistent with the research of Jensen (1993). Too large boards will have disadvantages in the form of coordination costs and free rider problems
12
(Guest, 2009). Furthermore, coordination and communication problems will occur because it is too hard to set board meetings which will lead less efficient decision-making (Jensen, 1993). This makes it easy for the directors to free-ride on the effort and take decisions based on the opinions of other directors (Jensen, 1993). A second problem is the fact that board members will not share a common purpose which will result in not clearly communicating between the members of the board. According to the research of Wintoki (2007) a significant problem with regard to the examining of the impact of board size on firm performance is that of endogeneity. This arises because both firm performance and board size are together determined by unobservable firm specific variables. The research of Yermack (1996) addresses this problem with a model and finds that the negative influence of board size still occurs.
2.5 Organizational complexity
According to recent empirical studies (Boone et al., 2007; Coles et al., 2008 and Linck et al., 2008) board size seems to be greater when the information need and advice from the board is high. In these studies there is a positive relation between board size and firm size. Proxies for complexity shown to have a positive impact on board size are financial leverage, firm size and industrial diversification (Coles et al., 2008). This study shows that board size is strongly influenced by firm specific variables. This suggests that the impact of board size on firm performance will differ among different types of firms. In this research complexity will be distinguished between diversify, size and leveraged. Diversified firms are more complex and operate in multiple segments (Rose and Shephard, 1997). According to the research of Hermalin and Weisbach (1988) the need for advice for CEO’s of diversified companies are greater because of the fact that they require a lot of expertise of a great number of industries. The research of Klein (1988) finds evidence that advisory needs of the CEO will increase when the firm depends more on the environment for resources. In line with this is that larger firms are more likely to have more external relations and require larger boards (Pfeffer, 1972). Pfeffer (1972) argues that the size of the board is chosen to lead to maximizing the procuring of important resources of the firm. High leveraged firms focus more on external resources and because of these high leveraged firms have greater advising needs (Pfeffer, 1972). Firm size, nature of organizational activities, diversification and leverage can be seen as proxies for the complexity of a firm and also the need for advice for the CEO. The need for a bigger board will increase if a firms’ complexity increases along one of these aspects (Coles, 2008).
The three proxies, as used in before conducted research, for organization complexity will be firm size, organization’s operation and leverage (Coles et al., 2008; Faleye, 2007 and Pfeffer, 1972). Firm size is based on the fact that large organizations are more complex than small organizations. However, according to Faleye (2007) complexity may not depend only on scale but also on the nature of organization’s operations. In order to examine this, the ratio of net property, plant and equipment to total assets will be used as a measure to determine the nature of organization’s activities. I will follow the research of Coles et al., (2008) and also expect that this ratio will be much lower for complex firms because the more tangible the firm’s assets are, the less complex the operations of an organizations are. It is also expected that high leveraged firms have greater advising needs because these firms rely
13
more on external resources(Klein, 1998). Klein (1998) suggests that advisory needs increases with the extent to which the firm has to depend on the environment for resources. The research of Booth and Deli (1999) finds evidence that bankers in corporate boards provide bank-debt-market expertise. Anderson, Mansi and Reeb (2003) found evidence that bigger boards leads to lower cost of debt. Based on the above mentioned expectations I formulated the next hypothesis:
Hypothesis 1: Complex firms have larger boards than simple firms.
Table 1 Summary of hypothesis
Hypothesis Proxies Variables Prediction with regard to board size
Organization complexity Firm size
Asset characteristics
Leverage
Total assets
Ratio of net property, plant, and equipment to total assets
Ratio of book value of total debt to book value of assets
+
-
+
At this point I will summarize the discussed hypothesis in Table 1. The above mentioned proxies will be used in order to analyze whether board size differs between complex and simple firms.
The first hypothesis will demonstrate the role of organizational complexity in the choice of the right board size. An important related question is whether these considerations have any impact on firm performance, since the fact that a phenomenon that has been observed must be differentiated from the question whether it is preferable or not (Faleye, 2007). It is interesting for example to show that complex organizations tend to have larger boards because of the multiple advising needs, but it is equally important to understand if there is a mismatch between board size and organizational complexity and if that is related to subsequent firm performance. The research of Coles et al., (2007) suggest and argued that firm performance which is calculated as Tobin’s Q should be negatively related to board size for firms which is derived from the long-lived deviations from optimal board size. They state in their research that transaction costs are a possibility that could hinder simple and complex firms from downsizing or upsizing their board sizes. Jensen (1993) and Yermack (1996) find evidence that undersized boards enhance the ability of the board in monitoring functions and
oversized boards are less likely to function effectively and are easier to control for the CEO. The research of Imam and Malik (2007) find empirical evidence that boards that are oversized tend to give the rationality that the coordination between the board members will be negatively impacted and if boards are undersized, the provisions of different opinions will be limited so having a negative impact.
14
So based on this I expect a negative impact of the mismatch between board size and organizational complexity on subsequent firm performance. I formulated the next hypothesis:
Hypothesis 2: There is a negative impact of the mismatch between board size and organizational complexity on subsequent firm performance.
3. Research methodology
In this chapter the data that will be used will be explained. The first paragraph will discuss the sample of the firms and the timeline. The second paragraph is about the used variables. And the last
paragraph will discuss the model and the control variables.
3.1 Sample and timeline
A quantitative research is in my opinion is most suitable because the purpose of this research is to provide empirical evidence if one size fit all. This research will look at the years from 2002 until 2006. The period after 2006 is less suitable because of the financial crisis which led to unstable information about board conditions. So the data will demonstrate the most contemporary results about the issue whether one board size fit all. The data needed will be gathered from the companies’ annual reports.
The data that will be used shall be conducted from the Standard & Poor’s 1500. The S&P 1500 included the indices from the S&P 500, S&P MidCap 400 and the S&P SmallCap 600. In total these indices cover up 90% of the total U.S. market capitalization. The information of the firms will be conducted from Wharton Research Data Service (WRDS), from the database Compustat North America. The firms that are just listed once will be excluded to increase the reliability of this research. Furthermore, the data revealed will be more trustworthy because of the fact that US companies have better information disclosure than companies in other countries.
3.2 Board size and firm complexity variables
The numbers needed will be gathered from the Thomson Reuter database. Board size will be calculated by the number of board of directors. The financial data will be conducted from the COMPUSTAT database. These are data about firm size, asset characteristics, leverage and diversification. The measure of firm size will be the natural logarithm of total assets in dollars. The measure of asset characteristics will be the ratio of net property, plant and equipment. Leverage will be measured as the ratio of book value of total debt to book value of assets. After gathering these financial numbers I will use these four components to examine it in more depth. First I will address the dependency among these proxies by extracting one common factor from these variables. This will be done first by principal component analysis, where I will extract an organizational complexity factor from firm size, asset characteristics, and leverage. I will also use a standardization analysis in order to compute a single factor which will also solve the problem of the dependency among these proxies.
15
Based on these numbers I will run an OLS regression and capture the residual term of each of these regressions to use it as a new variable for my second hypothesis.
Based on the second hypothesis the residual terms which are gathered from the first hypothesis will be the independent variables of the model. In order to analyze whether the mismatch between board size and firm complexity has an impact on subsequent firm performance I will run a regression where the residual terms will be the independent variables and future firm performance the dependent variable. The measure of firm value will be Tobin’s Q which will be calculated as book assets minus book equity plus market value of equity and divided by book assets.
3.3 Model and control variables
For the first hypothesis I will follow the method of Faleye (2007) where I will use an OLS regression relating board size to the measures of organizational complexity. The dependent variable will be a continuous variable. The model will also include a separate intercept term for each two-digit SIC code in order to control for unobservable industry effects. I will also include control variables such as growth which will be calculated as the market to book ratio and cash which will be calculated as the current ratio. If we look at prior literature current ratio is often used in operating performance regression models. In addition to above mentioned controls, I will also include controls that capture corporate governance. Prior literature mentions board independence, CEO duality and institutional holdings as potential control variables in order to capture corporate governance. I will follow the research of Faleye (2007) and include board independency and CEO duality as control variables in the model. Board independence will equal one if at least two-third of directors are independent. CEO duality will also be a dummy variable, which will equal one when the CEO serves as the chairman of the board and zero otherwise. The regression model for the first hypothesis will be as followed:
Board size = α + β1FirmSize + β2Asset characteristics + β3Leverage + β4Growth + β5CurrentRatio +
β6BoardIndependency + β7CEODuality + ∑βjIndustryDummyj + ε (1)
The second part of the first part of the analysis will consist of determining whether the complexity proxies are all positively related to each other which implies that these variables are more likely used together rather than as substitutes. I will follow (Keeling (2000) and approximate the expected eigenvalues for a random sample which uses the same sample size and same observation items. Having established that all three proxies load to one factor I will use principal component analysis to create a factor score that weighs each of the observed variables (FirmComplexity1), which is in line with the research of Abernethy, Kuang and Qin (2015). I will also use the standardization process and follow Carter et al. (2009) and Abernethy et al. (2015) and construct an alternative measure (FirmComplexity2) that combines equally weighted standardized values of each individual variable.This means that all individual variables will be standardized based on making the mean zero and standard deviation 1. After computing the PCA and standardization analysis the regression model will include two new independent variables, namely FirmComplexity1 and FirmComplexity2:
Board size = α + β1FirmComplexity+ β2Growth + β3CurrentRatio + β4BoardIndependency +
16
The analysis of the second hypothesis will consist of an OLS regression analysis. Additional to this there should also be control variables in order to reduce factors that are unrelated to board size but are actually related to firm performance which can result in affecting both concepts. Control variables which are mentioned in prior literature are current ratio and intangible assets, as in Lehn et al., (2004) and Baker and Gompers (2003). Intangible assets will be calculated as one minus the ratio of net PPE to book value of assets. Control variables which capture corporate governance will also be added. The information needed for the control variables will be conducted from the WRDS database. The
regression models will be as followed:
Tobin’s Qt+1= α + β1Oversized + β2Intangible assets + β3CurrentRatio + β4BoardIndependency +
β5CEODuality + ∑βjIndustryDummyj + ε (4)
Tobin’s Qt+1= α + β1Undersize + β2Intangible assets + β3CurrentRatio + β4BoardIndependency +
β5CEODuality + ∑βjIndustryDummyj + ε (5)
Tobin’s Qt+1= α + β1Oversized + β2Undersized + β3Intangible assets + β4CurrentRatio +
β5BoardIndependency + β6CEODuality + ∑βjIndustryDummyj + ε (6)
Tobin’s Qt+1= α + β1Deviation + β2Intangible assets + β3CurrentRatio + β4BoardIndependency +
β5CEODuality + ∑βjIndustryDummyj + ε (7)
The dependent variable will be future (t+1) firm performance which can be explained in many different
ways. In this study firm performance will only refer to financial performance and will be calculated as Tobin’s Q. The independent variable will be the mismatch (t0) which is derived from the residual term
from Model 1, 2 and 3, as a robustness test I will use the original residual because I can distinguish oversized boards from undersized boards which is based on the sign of the residual, i.e., positive residual means oversized board, and negative residual points at undersized board and the absolute value of the residual. This means that I will gather two different variables(Undersized and Oversized) from the first hypothesis which will represent the mismatch between board size and firm complexity. Oversized will be tested with Model 4, Undersized will be tested with Model 5, the absolute value of Oversized and Undersized will be tested with Model 6 and the deviation will be tested with Model 7. The regression model will measure how the mismatch between board size and firm complexity influence the changes of future firm performance.
17
4. Empirical analysis
This chapter will discuss the results of the study. This empirical research will be done by using an OLS regression model. The first paragraph will show the descriptive statistics of all variables. The second paragraph will discuss the analysis of model 1 and 2. And the last paragraph will present the analysis of the second hypothesis.
4.1 Descriptive statistics
Table 1 provides the descriptive statistics of all variables used in this study. There are in total 1814 observations between the years 2002 and 2006 which is almost the same as the observations of Faleye (2007) where his sample consisted of 1883 firm observations. The mean of board size in this study is almost the same as the study of Faleye (2007) and Cheng (2008) where they reported a mean of 9 board members. The mean of the variable Tobin’s Q is 3.63 with a standard deviation of 3.21. Table 1 also shows diversity among the sample firms. The average and maximal value of firm size are 8.20 and 12.51, respectively. On average, asset characteristics (Net PPE) constitute 26.83% of total assets and have a standard deviation of 19.07% which is almost in line with the results of the study of Faleye (2007). Over the period 2002-2006, growth average is around 2.66 with a standard deviation of 6.68. CEO duality has a mean of 0.64 and Board independence has a mean of 0.65 which is also almost in line with the research of Faleye (2007) and Holderness et al. (1999).
Table 1 Discriptive statistics Variables N Mean STD. Dev. First quartile Median Third quartile Independent variables Board size 1814 9.85 2.32 8 10 11 Tobin’s Q ROA 1814 1814 3.63 0.04 3.21 0.08 1.88 0.02 2.59 0.05 4.07 0.08 Dependent variables
Firm size (log) 1814 8.20 1.49 7.03 8.11 9.37 Asset characteristics 1814 26.83% 19.07% 12.67% 21.63% 37.43% Leverage 1814 1.00 1.24 0.32 0.58 1.17 Control variables Current Ratio 1814 0.57 0.29 0.36 0.51 0.69 Growth Intangible assets Firm size total Firm size (log) Board independence CEO duality 1814 1814 1814 1814 1814 1814 2.66 0.73 10464.14 8.20 0.65 0.64 6.68 0.19 18896.45 1.49 0.47 0.48 1.54 0.63 1131.83 7.03 0 0 2.14 0.78 3316.32 8.11 1 1 3.25 0.87 11731 9.37 1 1
18
Board size is a continuous variable and will be calculated by the number of board of directors. Tobin’s Q will be calculated as book assets minus book equity plus market value of equity and divided by book assets. The dependent variable firm size will be calculated as the logarithm of total assets. ROA will be calculated as the ratio of
earnings before interest, taxes, depreciation, and amortization to book value of assets. Asset characteristics will be calculated as the ratio of net property, plant and equipment. Leverage is the ratio of book value of total debt to book value of assets. Current ratio will be calculated as percentage. Growth is calculated as the market to book ratio. Intangible assets will be calculated as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are independent. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of 1,814 firms.
Table two gives an overview of the Pearson correlations. This analysis shows the relation between the variables. A thumb rule is that a correlation which equals 1 means that the two variables perfectly correlate with each other and -1 means that the variables perfectly negative correlate with each other. Table two shows that most of the variables are correlated with each other. The correlations that are significant at the 5% level are the bold numbers. The complexity proxies are all positively related to each other which implies that these variables are more likely used together rather than as substitutes. This leads to the question whether these variables should not be treated separately but actually as one variable that capture the complexity of firms as a whole. This will be further examined later by using principal component analysis and standardization.
Table 2
Pearson correlations
4.2 Organizational complexity analysis
Table 3 presents the results of the first hypothesis which analysis whether firm complexity has an impact on board size. The results indicate that firm size, and leveraged firms indeed have an impact on board size. This is based on the P-values which are significant at respectively 10%, 5% and 1%. The coefficient of firm size is positive and significant at the 1% level, which implies that the probability of board size increases significantly with firm size. The coefficient also suggests that an increase of one standard deviation of firm size while other variables stay the same will have a significant impact on board size. Similarly, the coefficient of leverage is negative and significant which indicates that leverage firms consider their size of the board. The P-value of asset characteristics of firms indicates that the ratio of Net PPE has no impact on board size. The results present that firm complexity proxies
BoardSize TobinsQ ROA FirmSize Assetcharac. Leverage BoardIndep. CEOduality Cash Growth Intangible
BoardSize 1 TobinsQ -0.0327 1 ROA 0.0168 0.3428 1 FirmSize 0.1027 -0.1581 0.1189 1 Assetcharac. 0.0434 -0.2107 -0.1170 0.1579 1 Leverage -0.0006 -0.3039 -0.1090 0.2658 0.4510 1 BoardIndep. 0.1124 -0.0159 0.0112 0.0161 0.0121 0.0164 1 CEOduality -0.1808 -0.0177 -0.0092 -0.0000 0.0077 -0.0341 0.1089 1 Cash -0.0874 0.1982 -0.0246 -0.3574 -0.3756 -0.4553 -0.0444 0.0410 1 Growth -0.0001 0.1670 0.2211 0.0158 0.0043 -0.0713 0.0288 0.0026 0.0479 1 Intangible -0.0434 0.2107 0.1170 -0.1579 -1 -0.4510 -0.0121 -0.0077 0.3756 -0.004 1
19
indeed have an impact on board size of companies and thus that firms consider the complexity in choosing their board size. But the results indicate that Board independence and CEO duality do have a significant impact on board size. Board independence has a positive coefficient which means that bigger boards tend to be more independent. CEO duality has a negative coefficient which is in line with the argumentation that powerful CEO’s want less board members. So based on the significant P-values which are presented in table 3 the null hypothesis is rejected.
Table 3
Board size and firm complexity (individual measures)
Variable Coefficient Standard error P-value
Firm size .139 .0389 0.000*** Asset characteristics .473 .320 0.140 Leverage -.154 .052 0.003*** Current ratio -.510 .217 0.019** Growth -.002 .008 0.822 Board independence .628 .112 0.000*** CEO duality -.939 .111 0.000*** Adjusted R2 0.070 Sample size Highest VIF 1814 2.99
Board size is a continuous variable and will be calculated by the number of board of directors. The dependent variable firm size will be calculated as the logarithm of total assets. Asset characteristics will be calculated as the ratio of net property, plant and equipment. Leverage is the ratio of book value of total debt to book value of assets. Current ratio will be calculated as percentage. Growth is calculated as the market to book ratio. Intangible assets will be calculated as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are
independent. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of
1,814 firms.
*, **, *** indicates significance level at respectively 10%, 5% and 1%
A concern with regard to above results is whether the measures of organizational complexity are jointly independent. If firm size, asset characteristics and leverage capture the same dimension of organizational complexity, the results may be biased. The Pearson correlation coefficients which are presented in the first paragraph of this chapter indicate that the complexity variables are correlated and significant at the 1% level. This indicates that these proxies are significantly collinear so including these variables in the same regression will bias the coefficient estimates and test statistics. I will address this in two different ways. First by using principal component analysis to capture one common factor from the variables to extract an organizational complexity factor (FirmComplexity1). The
components will be presented in Table 5. Second by using standardization analysis, which means that all variables will have a mean of 0 and a standard deviation of 1, where the output will be captured and transformed into a new variable (FirmComplexity2).
20
Table 5 Principal components Variables Organizational complexity factor Firm size0.448
Asset characteristics 0.608 Leverage 0.654Factors are created by using principal component analysis. Financial data are obtained from COMPUSTAT, averaged over 2002–2006. Sample size is 1,814 firms
The complexity factor assigns 0.448 to firm size which is measured as the natural logarithm of total assets. Factor loading of 0.608 is assigned to asset characteristics which are measured as the ratio of net property, plant and equipment to total assets and 0.654 to leverage which is measured as the ratio of book value of total debt to book value of assets. Based on above analysis where I tried to establish one common variable for organizational complexity rather than substitutes. Having established that all three proxies load to one factor I created two new variables, one for the PCS analysis and one for the standardization analysis, namely FirmComplexity1 and FirmComplexity2. I will use these two new variables as independent variables where I will regress board size on these two variables and include several control variables. The results of these two regressions are presented in Table 7 and Table 8.
Table 7
Board size and FirmComplexity 1 (composite measure)
Variable Coefficient Standard error P-value
FirmComplexity1 .265 .080 0.001*** Board independence .625 .112 0.000*** CEO duality -.941 .111 0.000*** Current ratio -.554 .216 0.021** Growth -.305 .076 0.000*** Adjusted R2 0.061 Sample size Highest VIF 1814 3.68
Board size is a continuous variable and will be calculated by the number of board of directors. The dependent variable Firmcomplexity1 which is measured by using principal component analysis to capture one common factor from the proxies of organization complexity to extract an organizational complexity factor. Current ratio will be calculated as percentage. Growth is calculated as the market to book ratio. Intangible assets will be calculated as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are independent. The model will also include a separate intercept term for each two-digit SIC code in order to control for unobservable industry effects. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of
1,814 firms.
21
Table 8Board size and FirmComplexity 2 (composite measure)
Variable Coefficient Standard error P-value
FirmComplexity2 .325 .127 0.011** Board independence .624 .112 0.000*** CEO duality -.926 .111 0.000*** Current ratio -.606 .216 0.025** Growth -.231 .2074 0.001*** Adjusted R2 0.066 Sample size VIF 1814 3.01
Board size is a continuous variable and will be calculated by the number of board of directors. The dependent variable Firmcomplexity2 which is measured by using the standardization process. Current ratio will be calculated as percentage. Growth is calculated as the market to book ratio. Intangible assets will be calculated as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are independent. The model will also include a separate intercept term for each two-digit SIC code in order to control for unobservable industry effects. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of
1,814 firms.
*, **, *** indicates significance level at respectively 10%, 5% and 1%
The dependent variable in the regressions is a continuous variable which indicated the size of the board. Factor loadings for organizational complexity are 0.448 for the natural logarithm of total assets, 0.608 on the asset characteristics and 0.654 on leverage. The governance data are obtained from ISS (formerly Risk metrics). Financial data are conducted from COMPUSTAT, averaged over the period 2002-2006. I included also in each regression industry dummies. *, **, *** indicates significance level at respectively 10%, 5% and 1%. Based on the result which are presented in Table 7 and 8 where I avoid potential biased results with regard to the coefficient estimates and test statistics I can state that results are virtually identical to those of Table 3. The results indicate that FirmComplexity1 is
respectively significant at 10%, 5% and 1% and FirmComplexity2 is significant at 10% and 5% which indicates that the complexity of a firm has significant impact on board size. This is in line with the research of Faleye (2007) and Coles et al. (2008) where they also found significant impact of the complexity of firms on board size. The research of Boone et al. (2007) and Lehn et al. (2004) also find similar results and mention that complex firms are in fact associated with bigger boards which is derived from the need for greater advice. So this research is in line with prior literature and finds evidence that firms indeed take into consideration the complexity of their operations when composing a board of directors.
4.3 Subsequent firm performance
An important related question is whether the consideration of the size of the board has any impact on firm performance, since the fact that a phenomenon that has been observed must be differentiated
22
from the question whether it is preferable or not (Faleye, 2007). It is interesting for example to show that complex organizations tend to have larger boards because of the multiple advising needs, but it is equally important to understand if the mismatch between board size and organizational complexity occurs and if that is related to subsequent firm performance. The dependent variable will be Tobin’s Q (t+1) firm performance which will capture future firm performance. I will follow Faleye (2007) and
estimate Tobin’s Q as book assets minus book equity plus market value of equity and divided by book assets. The independent variables will be the mismatch (t0) which are derived from the residual term
from the first hypothesis, as a robustness test I will use the original residual because I can distinguish oversized boards from undersized boards which is based on the sign of the residual, i.e., positive residual means oversized board, and negative residual points at undersized board. After capturing three different residuals from the Model 1, 2 and 3 I will divide each residual into two groups, namely “Oversized” and “Undersized”. Oversized is the residual if residual is higher than 0 and if the residual is lower than 0 it becomes 0. Undersize is the absolute value of the residual, if residual is lower than 0 and 0 if the residual is higher than 0. Based on the values of Undersized and Oversized I will add another variable which will be defined as “Deviation” that represents the absolute value of the total residual of each model. The regression model will measure how undersized, oversized and the mismatch between board size and firm complexity influence the changes of future firm performance. The results of the regressions will be presented in Table 9, 10, 11 and12.
Table 9
Oversized board and subsequent firm performance
Variable Model1 (Proxies) Model2 (Firmcomplexity1) Model3 (Firmcomplexity2) Oversized -.073 (.054) -.073 (.055) -.071 (.057) Board independence -.099 (.167) -.099 (.167) -.099 (.167) CEO duality -.225 (.164) -.225 (.165) -.225 (.165) Intangible assets .538 (.449) .538 (.449) .538 (.449) Current ratio .589** (.292) .589** (.292) .589** (.292) Adjusted R2 0.009 0.009 0.009 Sample size VIF 1814 2.92 1814 2.92 1814 2.92
Tobin’s Q is the dependent variable where this variable is used to capture future firm performance(t+1). Oversized is the independent variable which is derived from the residual term from the first hypothesis. Oversize is the residual if the residual is higher than 0 and what remains which is lower than 0 will be 0. Intangible assets is measured as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are independent. The model will also include a separate intercept term for each two-digit SIC code in order to control for unobservable industry effects. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of 1,814 firms. *, **, *** indicates significance level at respectively 10%, 5% and 1%.
23
Table 10Undersized board and subsequent firm performance
Variable Model1 (Proxies) Model2 (Firmcomplexity1) Model3 (Firmcomplexity2) Undersized .123** (.064) .137** (.064) 0.134** (.064) Board independence -.095 (.166) -.095 (.166) -.096 (.166) CEO duality -.224 (.165) -.224 (.165) -.225 (.166) Intangible assets .507 (.448) .531 (.448) .503 (.448) Current ratio .617** (.292) .611** (.292) 0.619** (.292) Adjusted R2 0.010 0.010 0.010 Sample size Highest VIF 1814 2.92 1814 2.92 1814 2.92
Tobin’s Q is the dependent variable where this variable is used to capture future firm performance(t+1). Undersized is the independent variable which is derived from the residual term from the first hypothesis. Undersized is the absolute value if residual is lower than 0 and what remains which is higher than 0 will be 0. Intangible assets is measured as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are independent. The model will also include a separate intercept term for each two-digit SIC code in order to control for
unobservable industry effects. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of 1,813 firms. *, **, *** indicates significance level at respectively 10%, 5% and 1%.
Table 11
Undersized and Oversized boards and subsequent firm performance
Variable Model1 (Proxies) Model2 (Firmcomplexity1) Model3 (Firmcomplexity2) Undersized Oversized .114* (.071) -.031 (.060) .123* (.071) -.029 (.060) .121* (.071) -.028 (.060) Board independence -.097 (.167) -.098 (.166) -.097 (.167) CEO duality -.225 (.165) -.225 (.165) -.224 (.165) Intangible assets .514 (.449) .540 (.449) .509 (.449) Current ratio .610** (.293) .605** (.292) .613** (.292) Adjusted R2 0.010 0.010 0.010 Sample size 1814 1814 1814 Highest VIF 2.92 2.92 2.92
24
Tobin’s Q is the dependent variable where this variable is used to capture future firm performance(t+1). Oversized and Undersized are the independent variables which are derived from the absolute value of the residual term from the first hypothesis. Undersized is the absolute value if residual is lower than 0 and what remains which is higher than 0 will be 0. Oversize is the residual if the residual is higher than 0 and what remains which is lower than 0 will be 0. Intangible assets is measured as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are independent. The model will also include a separate intercept term for each two-digit SIC code in order to control for unobservable industry effects. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of 1,814 firms.
*, **, *** indicates significance level at respectively 10%, 5% and 1%.
Table 12
Mismatch and subsequent firm performance
Variable Model1 (Proxies) Model2 (Firmcomplexity1) Model3 (Firmcomplexity2) Residual .022 (.055) .026 (.055) .026 (.055) Board independence -.098 (.167) -.098 (.167) -.098 (.167) CEO duality -.225 (.165) -.225 (.165) -.225 (.165) Intangible assets .520 (.449) .519 (.449) .519 (.449) Current ratio .605** (.293) .606** (.293) .607** (.293) Adjusted R2 0.008 0.008 0.008 Sample size 1814 1814 1814 Highest VIF 2.92 2.92 2.92
Tobin’s Q is the dependent variable where this variable is used to capture future firm performance(t+1). Residual is the independent variable which is derived from the absolute value of the residual term from the first hypothesis (Model 1, 2 and 3). Intangible assets is measured as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are independent. The model will also include a separate intercept term for each two-digit SIC code in order to control for unobservable industry effects. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of 1,814 firms.
*, **, *** indicates significance level at respectively 10%, 5% and 1%.
The results in Table 9 indicate that oversized boards do not have significant impact on future firm performance. The coefficient of oversized from model 1 is -0.073 and has a standard deviation of 0.054. The other oversized results which are derived from model 2 and 3 are close to these numbers. All three are not significant at the 10%, 5% and 1% level. The P-values are around 0.192 which indicates that there is a small impact but not significant enough. The model shows that current ratio is significant at a 5% and 1% level. This has to do with the fact that this variable is connected to performance. According to the results of Table 9 I can indicate that oversized boards do not have a significant impact on future firm performance. This is in line with the research of Jensen (1993) and Yermack (1996) where they mention that companies with oversized boards tend to become less effective. But their research goes beyond that and looks at undersized boards and their impact on
25
future firm performance. They find in their research that smaller boards enhance company
performance and influence the behavior of the investors and company value which is in line with the results of Table 10. According to this research undersized boards have a positive and significant impact on future firm performance. This is in line with the research of Jensen (1993) and
Yermack(1996) and their argumentation is that undersized boards tend to lead to prompt decision making, closer coordination and lesser communication gap. The results in Table 10 show that undersized boards indeed have an impact on future firm performance. The coefficient of undersized from model 1 is 0.123 with a standard deviation of 0.064. The results of model 2 and 3 are close to these numbers and are all significant at the 5% and 10% level. This means that undersized boards do have a significant impact on future firm performance because keeping boards small can help improve the performance of a company. When boards get beyond seven or eight people they tend to become less effective and are easier to control for the CEO (Jensen, 1993). Smaller boards also tend to dismiss CEOs following periods of poor performance much easier and evidence shows that CEO compensation has greater sensitivity to performance in companies with smaller boards (Yermack, 1996). The results of Table 11 ,where the mismatch (Undersize and Oversize) between board size and firm complexity has been taken as the independent variable, suggest where the complexity of firms point to a significant impact on the size of the board, oversized boards do not have significant impact on subsequent firm performance but undersized boards do have a significant positive impact on future firm performance. The results of Table 12, where I took the absolute values of Undersized and Oversized to look at the mismatch and regress that against future firm performance I do not find significant evidence. Based on this I found no evidence for the hypothesis that the mismatch between board size and firm complexity leads to a negative impact on future firm performance. But I find significant evidence based on the results of Table 10 and 11 that indicates that undersized boards have a positive impact on the future performance of companies.
4.4 Supplemental analysis
Additional analysis will be done to assess the robustness of the results of the previous paragraph with regard to mismatch and subsequent firm performance. This will be done by using an alternative measure of firm performance, which will be ROA. As another robustness test, I will add year dummies. The results of the additional analyses will be presented in Table 13.
Table 13
Mismatch and subsequent firm performance
Variable Model1 (Proxies) Model2 (Firmcomplexity1) Model3 (Firmcomplexity2) Residual -.006 (.001) -.006 (.001) -.005 (.001) Board independence -.005 (.004) -.005 (.004) -.005 (.004) CEO duality .006 .006 -.006
26
(.004) (.004) (.004) Intangible assets .021* (.111) .021* (.111) .020* (.111) Current ratio -.002 (.007) -.002 (.007) -.002 (.007) Adjusted R2 0.014 0.014 0.015 Sample size 1814 1814 1814 Highest VIF 2.92 2.92 2.92ROA is the dependent variable where this variable is used to capture future firm performance(t+1). Residual is the independent variable which is derived from the absolute value of the residual term from the first hypothesis (Model 1, 2 and 3). Intangible assets is measured as one minus the ratio of net PPE to book value of assets. CEO Duality is a dummy variable which equals one when the CEO also serves as board chairman, zero otherwise. Board independence is also a dummy variable which equals one if at least two-third of directors are independent. As a robustness test I will also include year dummies. The model will also include a separate intercept term for each two-digit SIC code in order to control for unobservable industry effects. Financial data are averages over the period 2002–2006, gathered from the database COMPUSTAT. The governance data are gathered from ISS (formerly Risk metrics). The sample consists of 1,814 firms.
*, **, *** indicates significance level at respectively 10%, 5% and 1
The results of Table 13 where I took the absolute values of Undersized and Oversized to look at the mismatch and regress that against an alternative measure of future firm performance show that I could also not find significant evidence for the impact of the mismatch between board size and firm
complexity on subsequent firm performance. The research of Love (2011) argues that the strongest relation between corporate governance variables, such as board size, is the strongest for valuation. Tobin’s Q is the strongest measure to use for valuation and other alternative measures such as ROA and ROE seem to be less strong in measuring the relation between corporate governance and operating performance (Love, 2011). Consistent with the findings in the previous paragraph, ROA has not a significant relationship with the mismatch between firm complexity and board size. So I still found no evidence for the hypothesis that the mismatch between board size and firm complexity leads to a negative impact on subsequent firm performance.
