UNIVERSITEIT VAN AMSTERDAM
Corporate governance and firm
performance during and after the
global financial crisis
Master Thesis
Student: Theo Dekker
Student Number: 10004598
Date: 19-05-2016
Education: MSc Accountancy and Control, Control track
Institution: University of Amsterdam, Faculty of Economics and Business
Supervisor : Prof. Alexandros Sikalidis
Content
1 Introduction 6
1.1 Background 6
1.2 Research question 10
1.3 Motivation 10
2 Literature review and hypotheses 12
2.1 Agency theory 12 2.2 Board characteristics 13 2.2.1 Board size 13 2.2.2 Board independence 15 2.2.3 CEO duality 16 2.3 Inside ownership 17 3 Research design 19 4 Evidence 22 4.1 Sample 22 4.2 Descriptive statistics 22
4.3 Board structure, inside ownership and firm performance (Tobin’s Q) 25 4.4 Board structure, inside ownership and firm performance (ROA) 27
4.5 Comparison between the two regressions 29
4.6 Sensitivity analysis 31
5 Conclusion 32
Appendix 1: Variable definitions 34
Appendix 2: Tables and figures 35
List of tables and figures
Table 1 - Observations per indstury 35
Table 2 - Descriptive statistics 35
Table 3 - Distribution of the means of the financial measurers between 36
2007-2013
Table 4 - Distribution of the means of corporate governance variables 36
between 2007-2013
Table 5 - Correlation matrix between independent and dependent variables 37
Table 6 - Test in differences in mean values 38
Table 7 - OLS regression of the financial measure Tobin's Q 40
Table 8 - OLS regression of the financial measure Tobin's Q in the crisis 41
and after the crisis
Table 9 - OLS regression of the financial measure ROA 42
Table 10 - OLS regression of the financial measure ROA in the crisis and 43
after the crisis
Table 11 - OLS regression of the financial measures Tobin's Q and ROA 44
with two-way clustered standard errors
Table 12 - Calculation of the turning points of INSIDEOWNER 45
Statement of Originality
This document is written by Theo Dekker, 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
This study uses the financial crisis (2007-2008) and the period after the crisis to examine whether and to what extent different corporate governance aspects affect the performance of firms. The focus was on four different governance aspects, the size of the board, CEO duality, the proportion independent directors in the board and inside shareholdings of executive directors. There were two measurers of firm performance the Tobin’s Q and return on assets (ROA). The sample is drawn from the S&P 500, and contains all the different industries for the years 2007 through 2013. This study finds that the size of the board had a negative
influence on firm performance during and after the crisis. Moreover, the effect was even more negative after the crisis. Another finding of this study is that the relation between the
proportion independent directors and firm performance is negative after the crisis.
Furthermore there was a negative relation between CEO duality and firm performance during the crisis. The relation after the crisis was even more significant, after the crisis CEO duality had a positive influence on corporate performance. The most robust evidence this study found was about inside ownership. There is evidence that there was a non-linear relation between inside ownership and firm performance.
1 Introduction
1.1 Background
Corporate governance practices have received heightened attention after the recent financial
crisis and there is evidence that suggests that corporate governance was ineffective during that crisis (Kirkpatrck, 2009). The general view about corporate governance is that is stands for the external rules and regulations and the internal system that is designed to minimize the agency problem between the principal and the agent (Shleifer and Vishny, 1997). The problem discussed in agency theory is the separation of ownership and control. Based on the theory different mechanisms mitigate the costs associated the conflict of interest among this separation (Jensen and Meckling, 1976). Another definition of corporate governance is given by Merchant et al. (2007). Thus, 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.
Most academic research about corporate governance focused on observable attributes of board governance such as an independent chairman, board size, % outside directors on board, % busy directors and board interlocks. When a company has some good corporate governance aspects, it is viewed as more attractive investment (Shleifer and Vishny, 1997). Interestingly, among firms during a crisis, the tight governance prescribed by the agency theory may actually be harmful to firm survival and shareholders interests (Hambrick and D’Aveni 1988,1992). However, there is relatively little research that devoted effective managing of a firm during a crisis (Daily 1994).
The financial crisis (2007-2008) can be seen as the worst financial crisis since the Great Depression (Eichengreen and O’Rouke, 2009). A financial crisis is a disturbance to financial markets, associated typically with insolvency among debtors and intermediaries and falling asset prices, which spreads through the financial system, disrupting the markets capacity to allocate capital (Eichengreen and Portes, 1987). In the crisis of 2007-2008, an unprecedented large number of financial institutions were bailed out by governments, and some large companies collapsed such as Lehman Brothers, and Merrill Lynch (Erkens et al., 2012). While the macroeconomic factors (e.g., loose monetary policies) that are at the roots of the financial crisis affected all firms (Taylor, 2009), some firms were affected more than others.
Recent studies argue that firms financing policies and risk management had a significant impact on the degree to which firms were impacted by the financial crisis
(Brunnermeier, 2009). Because firms' risk management and financing policies are the result of benefit-cost trade-offs made by shareholders and corporate boards (Kashyap et al., 2008), an important implication from these studies is that corporate governance affected firm performance during the crisis period.
There are four different stages in a business cycle (Burns and Mitchell, 1946). A cycle consists of expansions occurring at about the same time in many economic activities,
followed by similarly general contractions, recessions, and revivals which merge into the expansion phase of the next cycle (Burns and Mitchell, 1946). After a depression phase there will be a period of recovery. Therefore, there will be period of economic recovery after the financial crisis. There is a lack of literature examining firms after a period of financial crisis (Daily et al., 2003). Based on their research, there remains much to learn about governance mechanisms following a financial crisis. For example there are things to learn about the internal mechanisms such as an effectively structured board, compensation contracts and concentrated ownership holdings. To conclude this section, there is a need for more empirical research on the role of corporate governance during and after a crisis (Daily, 1994;Daily and Dalton, 1994a).
The second part of this section will cover an overview of former literature about the relationship between corporate governance and firm performance. First, it will focus on papers that looked at the relation between corporate governance and firm performance but not in a financial crisis period. Finally, the literature that focuses on the governance/performance relation during a financial/economic crisis will be described.
Bebchuk and Weisbach (2012) provide an overview of the state of corporate governance after the crisis. They discuss the areas in corporate governance and show the importance of the different areas. They focused on shareholders and shareholder activism, directors, executives and their compensation, controlling shareholders, comparative corporate governance, cross-border investments in global capital markets, and the political economy of corporate governance. All those areas could have an influence on the relationship between corporate performance and governance.
The relation between board composition, managerial ownership and firm performance is researched by Barnhart and Rosenstein (1998). They found that those variables are jointly determined. The variables interact, for example it might be the case that the board
structure of a firm changes in response to outside factors in a manner that is optimal for value maximization. The authors use the Tobin’s Q to measure firm performance and used the firms in S&P500 as a sample. To measure firm performance they used: 1) accounting measure (ROA), 2) firm value (stock-price). 3) Tobin’s Q. Based on former literature there is no superior measure (Love, 2011). However, Love (2011) found that the Tobin’s Q as a performance measure has the strongest association with governance.
Core et al. (1999) studied the relationship between corporate governance and firm performance but added the variable CEO compensation. They based their research on publicly traded U.S. firms. First, their results suggest that firms with weaker governance structures have greater agency problems. Second, that CEOs of firms with greater agency problems receive greater compensation. Lastly, that firm’s with greater agency problems perform worse. Agency problems play a large role during a crisis (Daily et al., 2003). The researchers used stock returns to measure financial performance. Bhagat (2008) used instead of stock returns, return on assets (ROA). His research showed that CEO duality and stock ownership of the firm has a positive effect on corporate performance. CEO duality refers to the situation when the CEO also holds the position of the chairman of the board. Klapper et al. (2004) used ROA and the Tobin’s Q to measure firm performance. They show that better corporate
governance is highly correlated with better operating performance and market valuation. Does this finding still hold during and after a crisis? They also find that there is a wide variation in firm-level governance.
There is also some literature available about the relation between corporate
governance and the likelihood of earnings management. High quality corporate governance leads according to Berkiris and Doukakis (2011) to higher level of earnings quality for large firms. This research was based on a sample of firms listed on the Athens, Madrid and Milan stock exchanges. During and after a crisis earnings management can have an influence on the reported financial statements.
In the next part this study will discuss the available research that is done about the financial crisis (2007-2008). Most literature focused only on financial firms. Financial firms and non-financial firms will have different levels of leverage. Financial firms have a high leverage compared to non-financial firms. By non-financial firms a high leverage is more likely indicates distress. It is therefore likely that the results of non-financial firms will differ from financial firms.
The conclusion from Adams (2011) was that governance of financial firms was on average, not obviously worse than in nonfinancial firms. In fact, using simple governance
scores that show that financials firms and banks generally appear to be better governed. Erkens and Hung (2012) give some more insights from financial institutions worldwide. They use data from 30 countries that show that firms with more independent boards and higher institutional ownership experienced worse stock returns during the crisis period. Their results suggest that firms with higher institutional ownership took more risk prior to the crisis, and therefore they had larger shareholder losses during the crisis.
There is also literature that focused only on the banking industry during the crisis in the U.S.. Peni and Vähämaa (2012) their results suggest that banks with stronger corporate governance mechanisms were associated with higher profitability in 2008. Hugh Grove et. al (2011) applied agency theory to the banking industry, the key variables were the level of debt and board size of banks. CEO duality was negatively associated with corporate performance. However, they argued that there are alternative theories needed to understand the performance implications of corporate governance at banks.
There are also some papers that showed the relationship between corporate
governance and firm performance during the Asian crisis (1997-1998). For example, Johnson et al. (2000) and Lemmon et al. (2003). In both papers agency theory played a role, however the papers found different roles for corporate governance during the crisis. Lemmon et al. (2003) examine the response of firms in eight different Asian countries to the crisis. They find that the Tobin's Q falls further and stock price performance is worse for those firms in which minority shareholders are potentially more subject to expropriation. Their conclusion is that ownership structure may be especially important in times when the level of investment opportunities is low. Johnson et al. (2000a) find that governance variables, for example the quality of law enforcement and investor protection indices, are powerful predictors of the extent of market declines during the crisis. These variables explain the cross-section of declines in a better way than macroeconomic variables do. The findings from the Asian crisis are hard to transfer to the situation in the U.S. First, every crisis differs in some respect to previous crises and the Asian crisis was in a different time period. There are some new aspects in the financial crisis (07-08) that were not a problem in in the previous period. Moreover, there are cultural differences between Asia and the U.S. For example, people from the U.S. have on average a higher level of overconfidence, which might influence the reasons or consequences of the crisis.
1.2 Research question
The research question that is central in this research is:
What is the relation between corporate governance and firm performance during and after the financial crisis (2007-2008)?
1.3 Motivation
The following section shows why answering the research question is relevant from a social and academic point of view. The background section demonstrated that there is a lot of prior literature that focused on the relation between corporate governance and firm performance. However, there is limited literature available for the period of financial crisis (2007-2008) that focused on this relation. There is already some research about the crisis but this was only focused on financial firms and banks. It can also be interesting to see what role corporate governance played for non-financial firms. Moreover, the sample of non-financial firms is larger.
Some firms were more affected during the crisis than others, and some company’s recovered more quickly from the crisis. Could there be some corporate governance aspects that helped to perform better during or after the crisis? Each firm is different, and for each firm some governance aspects can be more important. But it may be the case that some aspects are useful for all firms during or after a crisis. Prior literature discussed in the
background section did not focus on the period after the crisis. This papers aims to enrich our understanding about the relation between corporate governance and firm performance during and after a crisis. It is very interesting to see if some corporate governance aspects had
different effects during or after the crisis. For example was the effect of CEO duality the same during the crisis as after?
The question becomes interesting since corporate governance practices change over time, and the agency problems during a crisis may be larger (Daily et al., 2003). It is interesting to see if the theories that have an influence on corporate governance such as agency theory still holds during and after a crisis. Tight governance may have some negative consequences for the firm during a crisis. Based on agency theory, Hambrick and D’Aveni (1988, 1992) argue that corporate failures frequently unfold as a downward spiral in which executive teams are replaced so frequently and quickly that they have no time implement strategies that might, in fact, save the organization.
Kirkpatrick, G. (2009) wrote some lessons to learn from the financial crisis about the corporate governance. Nowadays there is also data available from the period after the crisis about corporate performance and governance aspects. Can we gain new insights now we have access to data after the crisis? Can we learn new lessons? If there are some new insights several social groups can benefit. Stakeholders such as investors and debt holders can for example use the new information about corporate governance practices in their decisions. The content of this paper is divided as follows, The second section will provide an overview of the related literature and described the hypotheses. The third section will explain the methodology of this study. The fourth section will give the results of this study and the fifth part will conclude after discussing these results. The conclusion includes some
2 Literature review and hypotheses:
This section will discuss the overwhelmingly dominant theory applied in corporate governance, the agency theory (Daily et al. 2003). Afterwards three aspects of corporate governance will be discussed in more detail. As discussed in the background section there are a lot of different areas in corporate governance (Bebchuk and Weisbach, 2012). This study will focus on three different corporate governance mechanisms board characteristics, inside ownership and institutional ownership. In the last part of each section the hypotheses will be formulated based on prior literature and the effect of a crisis on the different aspects.
2.1 Agency Theory
According to Eisenhardt (1989) agency theory is directed at the ubiquitous agency relationship, in which the principal (one party) delegates work to the agent (another), who performs that work. Agency theory attempts to describe this relationship using the metaphor of a contract. Agency theory is concerned with resolving two problems that can occur in agency relationships. The first is the agency problem that arises when the goals or desires of the principal and agent conflict and it is expensive or difficult for the principal to verify the actions of the agent. The second is the problem of risk sharing that arises when the principal and agent have different attitudes toward risk. The problem here is that the principal and the agent may prefer different actions because of they both have different risk preferences. Based on Jensen and Meckling (1976) agency theory can be seen as an explanation of how the public corporation could exist, given the assumption that managers are
self-interested, and based on a context in which those managers do not bear the full wealth effects of their decisions.
There are two different factors why agency theory is popular in governance research (Daily et al., 2003). The first reason is that it is a simple theory. In which large corporations are reduced to two different participants, shareholders and managers. The interests of each are assumed to be both consistent and clear. The second reason is based on self-interest.
Generally, managers are unwilling to sacrifice personal interests for the interests of others. Jensen and Meckling (1976) provided a rationale for how the corporation could survive and prosper despite the self-interested proclivities of managers.
2.2 Board characteristics
The board of directors is one among many elements of corporate governance structure. The role of the board is challenging and diverse. The board of directors has the power to hire, fire and compensate management at regular or irregular intervals (Baysinger and Butler, 1985). Second, the role of the board of directors in an agency framework is to resolve agency problems between shareholders and managers by setting compensation and replacing those managers that do not create shareholder value (Hirshleifer and Thakor, 1994). The alignment between the board and shareholders is presumed to be achieved by making the compensation of members of the board sensitive to corporate value. Another approach is to exploit the reputational concerns of the board members in the market for governance directors (Noe and Rebello, 1996). Third, the board provides access to resources (Hillman et al. 2000). Finally, the board also provides strategic direction for the firm (Kemp, 2006).
Prior literature mainly focused on the board characteristics like board size and the composition of the board (fraction of members that are independent of management). This study wills also focus on CEO duality. An optimal board is one with the size and composition that adjusts this tradeoff to maximize firm value. If boards are constituted optimally, they likely will differ across firms, reflecting the relative value of better information and better execution (Hermalin and Weisbach, 1998)
2.2.1 Board size:
Board size is the number of members on the board. Identifying appropriate board size that affects its ability to function effectively has been a matter of continuing debate (Jensen, 1993). The literature about board size is mixed. Most literature argue that small boards have a higher corporate performance compared to large boards (Eisenberg, et al., 1998; Yermack 1996). However, other literature argue that firms with a larger board had better performance (Adam and Mehran, 2003).
Lipton and Lorsch (1992) argue that larger groups face problems of social loafing and free riding. As board size increases, free riding increases and this reduces the efficiency of the board. According to Jensen (1993) efficiency in decision making is larger in smaller groups due to greater coordination and lesser communication problems. Consistent with this notion, Eisenberg, Sundgren, and Wells (1998) and Yermack (1996) provide evidence that smaller boards are associated with higher firm value.
Large boards were supported on the ground that they would provide greater
monitoring and advice (Klein, 1998; Adam and Mehran, 2003). Klein (1998) argues that the chief executive officer needs more advice when the complexity of the organization increases. Furthermore, firms that operate in multiple segments requite more advice (Yermack, 1996). Singh and Harianto (1989) found that larger boards improve board performance by reducing the CEO domination within board. A large board makes it difficult to adopt golden parachute contracts that might not be in the best interest of the shareholders.
Firms with more complex operations may have performed worse during the crisis (Erkens et al, 2012). However, those researchers didn’t found a relation between board size and performance during the financial crisis. The study however, was based on financial institutions. This study focused on financial and non-financial firms after and during the financial crisis. Most literature found that a smaller board size was associated with better performance, the efficiency in decision-making was larger in smaller boards. Therefore this study expects a negative relationship between board size and firm performance during and after a crisis.
A firm needs a clear strategy during and after the crisis. It is likely that a smaller board size makes it easier to set a strategy to recover from the crisis. In a smaller group there are lesser communication problems and greater coordination. This study expects that the negative relation is stronger after the crisis. During the crisis there is a chance that a larger board can add some value because of the expertise of the board members and they can give advice to the CEO. However, this can still be the case after the crisis but the effect will be smaller because their expertise is less needed.
Moreover, this study beliefs that the free riding problem amongst board members is more likely after a crisis. During a crisis board members will on average be more involved in monitoring and assisting the directors compared to a period after a crisis. Not every board member will monitor with the same effort but all get paid for their monitoring activities. When the board is too big after a crisis the board may play a more symbolic role and less part of the management process. All the individual members of the board of directors can benefit from the effort of each member and all benefit from the collective action (Lipton and Lorsch, 1992).
H1a: There is a negative relationship between board size and firm performance during a crisis.
H1b: There is a negative relationship between board size and firm performance after a crisis, and this relationship is even more negative compared to relation during the crisis.
2.2.2 Board independence:
An Independent director (also sometimes known as an outside director) is a director of a board of directors who does not have a material or pecuniary relationship with company or related persons, except sitting fees (Brudney, 1982).
Outside board members add value to firms, by providing monitoring services and expert knowledge (Fama, 1980; Fama and Jensen 1983). Outside directors are supposed to be act as guardians of the shareholders’ interests through their monitoring role. There is some evidence that suggests that outside directors are more effective monitors and a critical disciplining device for managers (Hermalin and Weisbach, 1988). Outside directors may also contribute to the value of firms through their evaluation of strategic decisions (Brickley et al., 1987) and through their role in dismissing inefficient and poorly performing managers (Hermalin and Weisbach, 1988).
However, it is not clear if board independence is always beneficial because
independent directors lack information (Adams and Ferreira, 2007). Independent directors are less likely to have an in-depth knowledge of the internal workings of the firm on whose board they sit. Sometimes the role of independence is been overemphasized, and the monitoring role can be more difficult if a director does not have the expertise.
Prior research however does not find a clear relation between corporate performance and board independency (Baysinger and Butler, 1985; Hermalin and Weisbach, 1991). Firms with more independent boards do not perform better than other firms. This finding suggests that firms choosing board composition optimally and this differs for each firm.
But research based on the financial crisis found a negative relation between board independency and firm performance (Erkens et al., 2012). They conclude that firms with more independent boards raised more equity capital during the crisis, which led to a wealth transfer from existing shareholders to debt holders. This study also expects a negative relation
between the percentage of independent directors and firm performance during the crisis. Inside board members may have more in depth knowledge of the firm (Adams and Ferreira, 2007), and may be more loyal. When a board member has more knowledge about the firm, he or she can give better advice to the managers how to react on the crisis. Moreover, the
monitoring role can be performed better with more knowledge about the firm. This will be the same in the period after the financial crisis.
H2a: There is a negative relationship between board independence and firm performance during a crisis
H2b: There is a negative relationship between board independence and firm performance after a crisis
2.2.3 CEO duality:
The position of board chairman wields influence, responsibility and power (Lechem, 2002). The chairman ensures that different ideas are heard and brought to table for discussion to enable an effective and harmonious decision making.
Leighton and Thain (1993) stated that the effectiveness of the board of directors is in fact largely depending and decided on, the efficacy of position of the CEO. An important board control structure mechanism is CEO duality. It refers to a situation where the firm’s CEO also serves as chairman of the board of directors. There are two competing views about CEO duality based on the perception of whether a firm is best served by monitoring
effectively (agency theory), or by strong leadership (stewardship theory).
Based on agency theory the board without an independent chair, the monitoring role can be particularly difficult. With CEO duality there is a conflict of interest. The CEO is responsible for the overall strategy of the company but he is also in the position to evaluate the effectiveness and efficiency of the strategy. The CEO has to evaluate his own actions. The CEO has in that situation combined power, which can lead to behavior of self-interest (Jensen and Meckling, 1976). The question is, can managers be trusted to act without self-interest and in the best interest of the shareholders. According to the agency theory, managers cannot be trusted and therefore must be controlled by the board.
It is preferable to have an outside chairman because, outside directors focus more on financial performance (Fama and Jensen, 1983; Johnson et al., 1993). Second argument for an outside director is that they are more likely than insiders to dismiss managers after poor performance (Coughlan and Schmidt 1985; Warner, Watts, and Wruck, 1988). Thirdly, an outside director will protect his own personal reputation as director, thus giving him a
stronger incentive to monitor (Fama and Jensen, 1983). A fourth argument is that CEO duality can lead to lower shareholder wealth, because of their possible inefficient and opportunistic behavior (Jensen and Meckling, 1976).
However, proponents of the stewardship theory argue that an organization should be guided by a strong unambiguous CEO who sets the strategy, issues commands and relays orders through the organization. In this way strategies are implemented as directed (Andrews,
1971; Barnard, 1938). Other studies support the idea that strong leadership can help a company to adapt environmental demands faster (Chandler, 1962; Mintzberg and Waters, 1982). Eisenhardt (1989) described that the speed of decision making is critical, and can lead to greater corporate performance. Especially in highly competitive or technology based Industries (Bourgeois and Eisenhardt, 1988). They argue that a shorter time frame for decision-making will lead to better corporate performance. This could have a positive effect for shareholder outcomes.
Lorsch (1989) suggested that providing a leader of the board of directors separate from the CEO could significantly help directors prevent crises, as well as to act swiftly when a crisis occurs. Moreover, positive changes are unlikely under crisis conditions in bankrupt and failing firms (Hambrick and D’Aveni, 1988). Therefore the expectation is that there will be a negative relationship between CEO duality and firm performance during the crisis.
The expectation of this study after the crisis is that CEO duality will be positively related to company performance. Duality offers the clear direction of a single leader, and a concomitantly faster response after a crisis. As with other inside-directors, the CEO-chair would be expected to have a greater knowledge of the company and its industry. With CEO duality there is a shorter time frame for decision-making and this will lead to better corporate performance.
H3a: There is a negative relationship between CEO duality and firm performance during a crisis
H3b: There is a positive relationship between CEO duality and firm performance after a crisis
2.3 Inside ownership
Prior literature has examined the relationship between the proportion of shared owned in the firm by board members and insiders (or insider ownership) and firm value. Jensen (1993) argued that there is a convergence of interest when there is inside ownership. Managerial shareholders will help to align the interest of managers and shareholders. Based on this theory corporate performance will increase.
However, Morck et al. (1988) consider non-linearity in the relationship between corporate performance and inside ownership. They regressed the Tobin’s Q on the fraction of the firm owned by the board for large U.S. firms in 1980. The conclusion was that there is a positive relationship between managerial ownership and Tobin’s Q for ownership levels between 0% and 5% and for levels beyond 25%. However the effects for levels greater than
25% were small. But inside ownership between 5 and 25% increasing ownership lead to poorer corporate performance. Morck et al. (1988) argue that in the range from 5-25% managerial ownership can be seen as entrenchment. Private benefits of agency driven decisions outweigh costs to managers in terms of value loss from choices that were
suboptimal. The convergence of interest theory of Jensen (1993) holds for the other ranges. McConnell and Servaes (1990) extended the analysis for eight additional years (1976-1988) with a larger sample and found a similar result.
Morck, Shleifer and Vishny (1988) have some arguments for the non-monotonic relation between firm performance and inside ownership. Inside ownership gives managers a claim to the firm profits, in addition managers also gain more voting power. When the power of a manager increases, the manager will be more difficult to displace and so becomes more entrenched. Therefore, there are multiple mechanisms at work that act together and have all an influence on the manager.
There is also some more recent literature about the relationship between inside
ownership and corporate performance. For example, Fahlenbrach and Stulz (2007) find some evidence that increases in inside ownership are associated with a higher Tobin’s Q. But they find no relationship when the level of managerial ownership decreased. Coles et al. (2012) made a new model in which corporate performance and inside ownership are jointly
determined. They use data during the period from 1993-2000 based on U.S. firms. They find results that are in line with those of Morck et al. (1988). As inside ownership increases, Tobin’s q at first rises and then falls.
The crisis played a significant role in the failure of key businesses, declines in
consumer wealth estimated in trillions of U.S. dollars, and a downturn in economic activity leading to a global recession (2008-2012) (Erkens et al., 2012). Shares prices decreased, so the shares of the managerial shareholders decreased in value. The wealth of those managers decreased and it is likely that they will be more uncertain about their job. It will be harder to entrench. Because of the lower wealth, you could expect that the managers will have a higher incentive to perform better, to increase the share price. However, this study expects a non-linear relationship between inside ownership and firm performance after and during the crisis. This is in line with former literature that showed a non-linear relationship. There are multiple
mechanisms (Morck, Shleifer and Vishny, 1988) that together and have all an influence on the
manager, and the crisis period did not change that fact.
3 Research Methodology
In this chapter the research methodology will be described. The first paragraph is about the financial performance variables. The second paragraph will discuss the model. Lastly, the control variables that are used in the model will be discussed.
The research method that will be used in this study is quantitative research. To address the research question and developed hypotheses, a regression model will be used. In a later section the model will be explained. But to use the model it should be clear what the financial crisis period was. The global financial crisis started in 2007 and end of the crisis was in 2008 according to literature (Erkens et al., 2012). After 2008 there were some recovery years, whereby the crisis still had some impact on corporate performance. Moreover, there may be some signals on the stock market before 2007 that indicated some pre-crisis effects. For example Erkens et al. (2012) control already for this fact. For that reason this research will look to a longer time period. Therefore this year was also added to the crisis period.
This research will look at the years 2007 until 2013. This time period will be divided in two different time periods. The first period is during the financial crisis (2007-2009) and the third period consists of years after the crisis (2010-2013). However, you can still argue about how long each period was, and when the crisis ended. But based on Figure 1, this study used 2009 also as a crisis year. The figure shows the movement of the S&P 500 from January 2007 till December 2010. It indicates still a decline in the first three months of 2009.
In this study there will be two different dependent variables for corporate
performance. Return on assets (ROA) and the Tobin’s Q. Based on former literature the Tobin’s Q has the strongest association as a performance measure with governance, followed by the operating performance such as ROA (Love, 2011). The main argument for using Tobin’s Q ratio is that it takes risk and return into consideration and is a long-term measure (Manuel et al., 1996). Other arguments for the use of the Tobin’s Q ratio are that it recognizes the present value of future profits, and reflects the company’s ability to improve performance over time (Salinger, 1984; Caton et al., 2001). The Tobin’s Q can be defined by the ratio of the market value to the replacement cost of its assets.
The second dependent variable for corporate performance is return on assets (ROA), an accounting measure that is more backward looking. The choice of return of assets is based on main operating results instead of capital structure decisions. The return on assets is
performance measure.
Martin (1993) argued that both measures should be used. He argued that ROA and the Tobin’s Q approaches could be seen as complements instead of substitutes. Both measures contain information about market power, and there is no compelling reason to think that one of them is better than the other. Klapper et al (2004) also uses both measurers to measure corporate performance across 14 emerging markets.
Stock returns were a third option to measure firm performance. This is the relative change in stock price over time. Former literature often used stock return in a crisis period (Erkens et al., 2012;Adams, 2012; Lemmon and Lins 2003). However, based on theory better governed firms should not be associated with future stock returns. Because in an efficient market, differences in governance will be taken into stock prices and hence have no additional impact on subsequent stock returns after controlling for risk (Love, 2011). Therefore the relationship between corporate governance and stock returns can be less strong. For this reason this study does not use stock returns as a performance measure.
The model that this study uses is based on the model that Mak and Kusnadi (2005). However, those researchers did research about board size in relation with firm performance. The model they used is also usable for this study, because they use for several corporate governance aspects. However, this study added an additional performance measure return on assets. The variables are described in the Appendix 1: Variable definitions.
The following research models will be used in this study:
Tobin’s Q = β0 + β 1 LNBOARDIZE+ β2 BCHAIR+ β3PROINDEP + β4 INSIDEOWNER +
β5 INSIDEOWNER2 + β6 LNASSET + β7 LEVERAGE+ β8 TFARATIO + β9 CAPEX +
β10 LNAUDITSIZE + β11 ACHAIR + β12 CRISIS + β13 (Industry dummies)
ROA = β0 + β 1 LNBOARDIZE+ β2 BCHAIR+ β3PROINDEP + β4 INSIDEOWNER +
β5 INSIDEOWNER2 + β6 LNASSET + β7 LEVERAGE+ β8 TFARATIO + β9 CAPEX +
β10 LNAUDITSIZE + β11 ACHAIR + β12 CRISIS + β13 (Industry dummies)
Whereby the Tobin’s Q and ROA are the two different dependent variables. And the other variables can be seen as different corporate governance aspects and control variables. The other financial control variables are the book value of total assets (which is a proxy for firm size (ASSET)); changes in fixed asset plus depreciation divided by total assets
(CAPEX); total liabilities divided by total assets (LEVERAGE); total fixed assets divided by the total assets (TFARATIO); changes in total sales divided by the total sales in the previous year (SALESGROWTH). The variables CAPEX and SALESGROWTH are to control for growth and investment opportunities.
The board variables are the board size (BSIZE); the proportion executive and
independent directors (PEXEC and PINDEP); a dummy variable representing board structure which equals 1 if the CEO also held the position as chairman; the audit committee size (ASIZE); a dummy variable that represents the status of the audit committee chairman which equals 1 if the chairman is an independent director (ACHAIR), otherwise zero.
There is one ownership variable (INSIDEOWNER) this variable is the percentage of inside ownership of an executive director. The four independent variables where this study will focus mainly on are the board size, CEO duality, proportion independent directors and inside ownership. In the regression models there is also the square term of inside ownership (INSIDEOWNER2). The reason here fore is a possible non-linear relation that previous studies had documented (Mak and Kusnadi, 2005).
In this study there is controlled for industry fixed effects. The reason for this is that the level of differentiation of products may depend on the industry and markets itself to which it belongs. For instance, environmental performance can be particularly important in a certain industry. Therefore the SIC code should be used as a dummy variable. Corporate performance may alter significantly over time during the study period as especially because of the crisis years. Therefore a crisis dummy was added to the model.
To control for the company size in the regression models, this study will look to the total assets, board size and audit committee size. But it is unlikely that those variables are evenly distributed. For this reason the natural logarithm will be taken of the each variable. Mak and Kusnadi (2005) had some additional variables about blockholders in their model, however this study didn’t had access to the Thomas Reuters database for blockholders. To the lack of data of stakeholders, there is no way to control for people or institutions with a high amount of shares. Large blockholders can have an important role in company
governance. It is plausible that some blockholders are active institutional investors. They will have the capability to monitor their investment, and the magnitude of their investment can affect managerial behavior. There is evidence in the literature that shows that blockholders and institutional investors do have an impact on managerial behavior (Brickley et al., 1988; Van Nuys, 1993).
4 Evidence
4.1 Sample
The sample is drawn from the S&P 500, and contains all the different industries for the years 2007 through 2013. The industries will be distinguished on the basis of their SIC (Standard Industrial Classification). The S&P 500 is an American stock market index based on market capitalizations, it consists of the 500 large companies having stock listed on the NYSE or NASDAQ. The information about corporate governance is gathered from the database
Riskmetrics. There was only data from 2007 in the Riskmetrics database, in the ideal situation the research period was from 2004 or 2005 till 2013. In this situation you also had some data from before the crisis. However, this was not possible because of the lack of data. The
financial information is covered by the Compustat North America. This study excluded firm-years for which either the governance data or financial data were unavailable. Moreover, this study excluded some unrealistic values by winsorizing the top and bottom 1% of the Tobin’s Q measure and ROA measure. The total sample consisted of 2995 observations.
4.2 Descriptive statistics
Table one describes the number of observations per industry. The industries are based on
their SIC codes. The manufacturing industry represents 41% of the total firms in the sample. The financial industry and services industry are the second and third largest industry in this sample. The coefficients of the industry dummies had significant influence on the different models. The industry dummies are not tabulated but in all the models there is controlled for industry fixed effects.
Descriptive statistics for the sample are showed in Table 2. The mean value of the TOBINQ was 1,5494, which is a bit lower than the Tobin’s Q of 1,971 based on the S&P1500 (Adams, 2012). However, those researchers focused on the period before the crisis 1996-2007. However, Adams found a ROA of 3,376 %, this study found a mean value of 6%. An average board consisted of 11 people. Most of the board members were independent directors (81,75%). This is quite in increase compared to Adams, in their research they found that about 65% of the board members were independent. The size of the audit committee was around 4 people, and almost in every company the chairman of the audit committee was an
independent board member (96,39%). In 58,43% of the observations the CEO also held the position of chairman of the board. Moreover, the average director held 2,27% of the company shares.
The distribution of the means of the dependent variables TOBINQ and ROA are in Table 3. When you compare the years 2007 and 2008, there is a large decrease in performance because of the financial crisis. In the year 2008 the ROA and TOBINQ are have their lowest mean value. In 2009 the performance is still quite lower compared to their average mean values. It is noticeable that there is an upward trend since the year 2009 for the TOBINQ. The highest mean value of the ROA was 68,78% in the fiscal year 2011.
Table 4 describes the distribution of the mean values of the corporate governance variables. The control variables are not showed. Because this study focuses mainly on the size of the board, CEO duality, proportion independent directors and inside ownership. The board size is quite constant over the years. Inside ownership increased during the crisis, in the years 2008 and 2009 board members held more shares. In the same years CEO duality
increased, during the crisis companies choose more often to the situation whereby the CEO was also chairman of the board. The proportion independent directors increased almost every year. In 2007 80,35% of the board members were independent, this percentage was 82,84% in 2013.
In Table 5 the correlations of all the independent and dependent variable are shown. There was a high correlation (0,5206) between the two dependent variables. These variables are the two corporate performance measures and will not be tested in one model. The
correlation between TFARATIO and CAPEX is very high (0,7183) but this was also expected based on (Mak and Kusnadi, 2005). Both control variables depend on fixed assets and total assets of a company. Moreover, Table 5 suggests that a larger company (based on total assets) will have a larger board and audit committee. This seems legit, larger companies will have on average more assets, a larger board and a larger audit committee.
Furthermore, the correlation table showed a negative significant relation between inside ownership and the proportion independent directors (-0,4275). This finding means that when a board member is independent he or she has lower inside ownership. This finding can be explained by how the variables are formulated in this study. Inside ownership only focused on the shares of executive directors, therefore only executive directors have inside ownership. When you are an independent director the variable INSIDEOWNER will have a value of zero. It is noticeable that there are a lot of statistical significant correlations.
Reverse causality was a concern based on prior literature of Barnhart and Rosenstein (1998), who argued that the relation between board composition, managerial ownership and firm performance is jointly determined. The variables interact, and it is possible that for example the number of board members has an effect on firm performance. But firm
performance also has an effect on the number of board members of a firm. But the results of table 5 do not show a high correlation between those variables. But potential reverse causality should be kept in mind for interpreting the findings.
Next this study examines whether differences in mean values of the different variables are significant under a different period. Whereby 2007-2009 represents the crisis years and 2010-2013 the non-crisis years. In addition, test of the mean values of companies in the financial industry differ from the other industries. The third test will test if the mean values of companies differ if the company as combined the function of CEO and chairman. In the fourth test this study focused on the differences between companies with or without leverage. Finally, the fifth test if there is a significant change in the mean values of companies with a higher than average Tobin’s Q and companies with a lower than average Tobin’s Q. The mean value of the Tobin’s Q was 1,5494.
All the results of those tests are described in Table 6. Panel A shows that the financial measures the Tobin’s Q and ROA are higher in the period after the crisis. Moreover, the variables BCHAIR and PROINDEPENT are significantly higher in the period after the crisis.
Panel B shows the mean values of the financial and non-financial firms. All the variables expect inside ownership and CEO duality are significantly different. Financial firms had a lower ROA and Tobin’s Q and had on average a lower percentage leverage.
Panel C described a test based on CEO duality. Firms without CEO duality performed on average in the whole time period better compared to firms with CEO duality. However, this finding does only hold for the Tobin’s Q. The t-test for ROA is not significant.
Furthermore, there are some differences in the proportion independent directors, amount of leverage and inside ownership between companies with and without CEO duality.
The most significant differences in mean values are based on the amount of leverage in a company. Panel D describes the differences in mean values for companies who had no leverage, and the second group had leverage higher than zero. Firms without leverage performed much better, and had on average a smaller board. CEO duality was less likely when the firm had no leverage, but the board members are likely to have a higher percentage company shares.
Panel E shows the results of the test in difference between sample firms with a higher value than the median Tobin's Q and sample firms with a lower value than the median Tobin's Q. When the Tobin’s Q had a higher value than the average value there are some significant mean values for the different groups. The amount of leverage is lower when the Tobin’s Q was higher. Moreover, inside ownership is also higher, when the Tobin’s Q is higher than
average. But the proportion independent directors is lower in the group with a higher than average Tobin’s Q.
4.3 Board structure, inside ownership and firm performance (Tobin’s Q)
In the next sub-sections the results of the multivariate models will be showed, followed by an analysis. The standard model will be used to show the relations between the dependent
variable and de independent variables. After that, there will be two different models with each another time period. The first model is during the crisis, and the second model is after the crisis. This means that there will be three models for each performance measure. First this study will focus on the results with the Tobin’s Q as a firm performance measure. Secondly, the model will be analyzed with ROA as dependent variable. Then a comparison between the two models will be discussed. Finally there will be a sensitivity analysis.
Results of the multivariate model of the Tobin’s Q are reported in Table 7. Column (I) represents an OLS regression without the control variables, column (II) represents an OLS regression with control variables, and column (III) represents an OLS regression with control variables and the t-statistics are corrected for heteroskedasticity. The R-squared is a statistical measure of how close the data are fitted to the regression line. The R-squared for the three models is 0.0727, 0.2342 and 0,2342. This means that the second and third model have a stronger relation with Tobin’s Q than model (I).
The coefficients for LNBOARDSIZE are negative in all three models and significant at the 0.01 level. This implies that a larger board size had a negative influence on the firm performance. The variable BCHAIR is not significant in any model, and the sign is different for the models. In the first model BCHAIR has a negative value, and in the second and third model it is positive. Furthermore, the proportion independent directors had a negative
influence on firm performance according to the three models in Table 7. In all the models the results are significant however the level differs.
Inside ownership showed a non-linear relationship, the variable INSIDEOWNER is negative, however INSIDEOWNER2 is positive. This suggests that when you plot a graph of the variable inside ownership you will get parabola. However, there is no significant
evidence. In model (I) is variable INSIDEOWNER only significant, and in model (II) and (III) is INSIDEOWNER2 significant. In none of the models they are both significant.
With respect to the control variables they all have a significant influence on the model expect the variable LEVERAGE. The CRISIS variable is also significant, this was dummy variable that was 1 during the crisis and zero after the crisis. Almost all control variables had
a negative influence on firm performance expect CAPEX and LN AUDITSIZE.
The main topic in this study is what is the relation between corporate governance and firm performance during and after the financial crisis? To answer this question the model was adjusted. The crisis variable was dropped, and the sample was split in two subgroups. The first sample consists data of companies during the crisis, and the second sample consist data after the crisis. The results of the OLS regressions of the two groups are tabled in Table 8. Column (I, II and III) of Table 8 are based on data from during the crisis. Column (IV), (V) and (VI) represent the same as the first three columns but these results are from after the crisis.
The first hypothesis expected a negative relationship between board size and firm performance during and after the crisis. Moreover, the relation was even more negative after the crisis. LNBOARDSIZE is negative in all the six models, however it was not significant in model (II) and (III). Therefore it is hard to say something about the relation between board size and firm performance, there is only a significant evidence of a negative relation in the model without control variables. But there is weak evidence that there was a negative relation between board size and corporate performance during the crisis. When you compare the results of after the crisis with those before the crisis, there is an increase in magnitude. Based on this finding there is support for hypothesis H1b.
The variable PROINDEP is only significant in models after the crisis. In all the models the proportion independent directors had a negative influence on the Tobin’s Q. But only after the crisis this variable had a significance level of 0,01. For this reason there is only support for H2b. There is a negative relationship between board independence and firm performance after the crisis.
CEO duality had a negative influence on firm performance during the crisis based on the significant values of the coefficient BCHAIR. During the crisis this variable was negative and significant, however after the crisis BCHAIR had a positive value. But only significant in the model with control variables. Therefore there is evidence that supports both hypothesizes about CEO duality H3a and H3b.
The final hypothesis was about inside ownership, the first model (Table 7) already showed a non-linear relation of the coefficient INSIDEOWNER however in this model there was no level of significance. In the models (II and III) in Table 8 their in is proof that there is a non linear relation. The variable INSIDEOWNER is negative (-4,7363) and the variable INSIDEOWNER2 positive (13,3513). Both variables had a large influence on the dependent variable. In Table 12 the turning point is calculated. The turning point is after 17,73 %, this
means that after this point the variable is positive and before this point negative. The
coefficients in model IV, V and VI are not significant. In model IV only INSIDEOWNER is significant, and in MODEL V is only INSIDEOWNER2 significant. However there is evidence that suggests that there is a non-linear relation between inside ownership and firm performance during the crisis. It is worth mentioning that this study used OLS regressions, but this finding about inside ownership suggests nonlinearity, this might lead to a
misspecification of the estimated model.
With respect to the control variables, LNASSET, TFARATIO and CAPEX had in all the models a significant influence. LNAUDITSIZE and ACHAIR only had an influence in the models after the crisis. The variable LEVERAGE was negative, however it was never
significant in any model.
To summarize the results in Table 8, there is weak evidence that support the first hypothesis 1a that there is a negative relationship between board size and firm performance during the crisis. Moreover, there is strong evidence that supports the even stronger negative relation after the crisis. Furthermore, there is no evidence for hypothesis h2a. There were no significant results of a negative relation between board independence and firm performance during the crisis, but there was proof that there is a negative relation after the crisis (h2b). There was also proof of the hypothesizes h3a and h3b, there was a negative relation between CEO duality and firm performance during the crisis, and a positive relation after the crisis. Finally, there was some proof that there is a non-linear relationship between inside ownership and firm performance during the crisis, but this relation was only seen in the model during the crisis.
4.4 Board structure, inside ownership and firm performance (ROA)
The other dependent variable of this study was return on assets (ROA). The OLS regression results are reported in in Table 9. The coefficient LNBOARDSIZE has in all the three models a negative significant influence. However, the level of significance was higher in the first model without control variables. The variable BCHAIR is however in no model significant. The model showed a negative relation between proportion independent directors and firm performance. Furthermore, the model showed a non-linear relation of the coefficient INSIDEOWNER. This relation was significant in all the three models. The turning point of the slope in model II and III is at 16,13% (Table 12).
All the control variable are significant, but it the variable CAPEX had the largest influence on firm performance. The models showed a positive relation between CAPEX and
firm performance. This means when the value of the change in fixed assets plus depreciation divided by total assets was positive this had a positive on corporate performance. The audit size also had a positive influence on performance, but all the other control variables had a significant negative influence.
The coefficients for LNBOARDSIZE are negative in all six columns. However, these are not all significant. In model II and III the coefficients are not significant. In line with the Tobin’s Q model, the coefficients in the models after the crisis are more negative. Based on Table 10 there is weak evidence that supports H1a, because only model (I) shows a significant negative relationship between board size and corporate performance. But there is strong evidence of a negative relation in the models after the crisis, therefore there is some support for H1b. When you compare models I and IV the relationship is more negative after the crisis. The variable PROINDEP is not significant in any of the models during the crisis. The sign of the coefficient is negative, but there is no support for hypothesis 2a. After the crisis the variable had a 0.01 significance level, and supports H2b that there was a negative relationship between board independence and firm performance after the crisis. Furthermore, based on the results in Table 10 there is no evidence of a negative relation between CEO duality and corporate performance during the crisis. The coefficient BCHAIR is negative in the first three models but not significant. Therefore there is no support for H3a. But in the period after the crisis in the models with control variables the coefficient BCHAIR is positive and significant.
With respect to insider ownership this study found evidence of a non-linear relation with firm performance. In all the models the coefficients were significant expect for model IV. Therefore there is evidence that supports H4. In the models II and III the slope of the coefficient changed after 17,03%, this was at 14,90% in the models V and VI (Table 12). This finding was in line with the results of the Tobin’s Q model. Whereby the coefficient of
INSIDEOWNER was negative, but the coefficient of INSIDEOWNER2 was positive. Both variables had a large influence on firm performance.
With respect to the control variables, the coefficients for LNASSET, LEVERAGE, TFARATIO, CAPEX are negative, negative, negative and positive and all significant at 0.01 in the models after the crisis. During the crisis only the coefficients for LNASSET and LEVERAGE were significant. The sign of the coefficients does not change, but there are some different levels of significance of the other control variables. A possible explanation of less significant results in the period during the crisis is the smaller sample size.
hypotheses 1a and 1b. There was a negative relationship between board size and firm
performance during the crisis, and this relation was stronger after the crisis. This study found no evidence for hypothesis 2a and 3a. In the models during the crisis there was no significant relation between board independence/CEO duality with firm performance. However, there was evidence that supports a negative relation between board independence and firm performance after the crisis (H2b). Moreover, there was also proof that CEO duality had a positive influence on corporate performance (H3b). Finally, the models in Table 10 indicate that the relation between inside ownership and firm performance is not linear.
4.5 Comparison between the two regressions
In this section the regression results of Table 8 and Table 10 will be elaborated. First of all
the R-Squared of the models based on the Tobin’s Q as performance measure is higher
compared to the ROA model. This means that the variables of the model can better predict the dependent variable Tobin’s Q.
The first hypothesis was based on the size of the board, and what will be the influence of a larger or smaller board on firm performance. This study expected a negative influence of board size on performance during and after the crisis (Eisenberg et al., 1998; Yermack, 1996). Moreover, the influence was expected to be larger after the crisis. Both models only found weak evidence of a negative relation during the crisis. Only the models without control variables gave significant results. After the crisis there was proof of the negative relation, but only the models without control variables can be compared to see if the relation was more negative after the crisis. Therefore, there was only weak evidence that supported that the relation was even more negative after the crisis, but it was clear that after the crisis the relation was negative. But these findings were in line with the findings of Eisenberg et al. (1998) and Yermack (1996) that small boards have a higher corporate performance compared to large boards. Larger groups faced during and after the crisis problems of social loafing and free riding. Moreover, the efficiency in decision making was larger in smaller groups (Jensen, 1993) due the greater coordination and lesser communication problems.
The second variable of interest in this study was board independence, this study expected a negative relation of board independence and firm performance during and after the crisis. However, in both models there was no proof of a negative relation during the crisis. The coefficient was negative, but never significant. Erkens et al. (2012) did find a negative relation between board independence and firm performance at financial institutions during the
crisis. But this study did not found no relation and this is in line some previous literature (Hermalin and Weisback, 1991;Baysing and Butler, 1985).
But in the Tobin’s Q model and ROA model after the crisis there was a negative relation between board independence and firm performance. This finding suggests that independent board members may have had less in depth knowledge of the firm (Adams and Ferreira, 2007), and may be less loyal.
CEO duality was third corporate governance aspect in the models. During the crisis there was a negative relation between CEO duality and corporate performance. However, there was only proof of the models based on the Tobin’s Q. But these findings support the agency theory. A board that does not have an independent chairman can monitor more difficult. With CEO duality there may be a conflict of interest. These results support Lorsch (1989) and Hambrick and D’Aveni (1988) that providing a leader of the board of directors separate from the CEO could significantly help directors to act swiftly when a crisis occurs, and a firm is less likely to go bankrupt when the positions are separated.
After the crisis this study expected a positive relation between CEO duality and firm performance, and in the models with control variables there was proof of this relation. This suggests that the speed of decision-making is critical and can lead to greater performance (Eisenhardt, 1989). With CEO duality after a crisis there is a shorter time frame and this could lead to better performance. After the crisis there is support for the stewardship theory
(Donaldson and Davis, 1991), an organization should be guided by a strong and unambiguous CEO who sets strategy.
The last variable of interest was inside ownership, the number of shares held by executive directors. This study found strong evidence that there was a non-linear relation between inside ownership and firm performance. There were significant results in the models based on return on assets. However, in the models based on the Tobin’s Q the relation was only significant during the crisis. The findings are in line with that of Morck et al. (1988) and Fahlenbrach and Stulz (2007), those researchers also found a non-linear relation for inside ownership. There are multiple mechanisms at work that act together and all have an influence on the manager. Inside ownership gives managers a claim to the firm profits, in addition managers also gain more voting power. When the power of a manager increases, the manager will be more difficult to displace and so becomes more entrenched. Morck, Shleifer and Vishny (1988) found a positive relation of inside share holdings after 25%, however this current study found a positive relation after 18%. However, those researchers did research in 1980 that is a different time period.
4.6 Sensitivity analysis
In the previous sections, this study demonstrated the relationship between different corporate governance aspects and firm performance during, and after the crisis. In this section the sensitivity of the results will be examined by investigating whether the results are attributable to correlations between observations within the sample of this study. Almost all firms appear more than once in the sample. This study preformed a sensitivity analysis because Gow et al. (2010) report that time series and cross-sectional dependence could lead to miss specified test statistics. A regression with two-way clustered standard errors can reduce the risk on miss specified test statistics due the dependence between observations in the whole sample (Gow et al., 2010). The results of the sensitivity analysis are reported in Table 11.
The standard errors were clustered by year and by company. The R-Squared of the clustered standard error models are respectively 0,2324 and 0,1274. These R-Square values are consistent with the previous models. The total number of clusters of Tobin’s Q model was 476, and there were 521 clusters in the ROA model. This means that there were for example 476 unique companies in the sample of the Tobin’s Q model.
The results in Table 11 show less strong results than the original models. The
coefficient LNBOARDSIZE is less significant in the Tobin’s Q model, and is not significant anymore in the ROA model. Moreover, the proportion independent directors still indicate a negative relation with firm performance but this is only significant in the ROA model. Furthermore, there is still no significance in the coefficient BCHAIR, but the coefficient is positive in both models. The non-linear relation of INSIDEOWNER is still significant but lower in the ROA model.
A lot of the control variables are not significant anymore when the standard errors were clustered. LNAUDITSIZE, ACHAIR and CRISIS are not significant anymore in the ROA model. The CRISIS is also not significant anymore in the Tobin’s Q model. The coefficient LNAUDITSIZE lose some significant value in the Tobin’s Q model, and the coefficient LEVERAGE also lose some significant value but in the ROA model.
Overall, not all the results hold after the regression with clustered standard errors. However, this study only applied this to the original models, and not for the different time periods. But this implies that the regression results are affected by correlations between the firm-year observations.
