1
Master’s Thesis in Finance
Luis Alberto Peña Cabral,
10825312Msc Finance
luis.pcabral89@gmail.com
December 2015
The importance of having corporate
governance practices
2
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.
3 Abstract
This thesis is focused on the importance of having good corporate governance practices and its impact on the performance of the company. This research also attempts to test the impact of having a new CEO in the office and also the reason of the departure of the previous CEO how affects the company. The model used in this thesis is OLS regression using Tobins Q ratio as a measure for the performance of the company. The corporate governance proxies used in this thesis are: the E-Index, the board size and the fraction of independent members in the board.
Using this model, this thesis shows that companies with good corporate governance level, given by E-Index, outperform companies with bad corporate governance level. In this research we can also appreciate the importance of having a big board with a certain amount of balance between independent members and insiders in the board. In fact, when a firm has too many independent members, it does not benefit the firm, but only decreases its performance.
Regarding the impact of the reason why the previous CEO left the company, this thesis shows that when the previous CEO did not leave the company voluntarily, there is a positive impact on the performance of the company.
4
Table of contents
1. Introduction ……….5 2. Literature review………..9 2.1 Hypotheses ………....16 3. Methodology ………...174. Data descriptive and statistics ……….20
4.1 Data ………...20
4.2 Descriptive statistics ……….21
5. Results ……….23
6. Robustness Checks ………...30
5 1. Introduction
Corporate Governance, as defined by Shleifer and Vishny (1997), is the way that investors insure their return on investment. Like Bebchuk and Weisbach (2009) mentioned, it has become a popular field of study for many researchers from many fields in the past years, such as management, law, finance and even accounting.
Some other scholars like Bebchuk et al., (2008) have focused on the importance of corporate governance practices and how they relate to the value of the firm. They made an analysis between corporate governance and firm performance. This analysis resulted in an Index which contains six provisions, four of them are related to the limits of the voting power of shareholders and the other two to measures performed in case of a hostile offer. These six provisions are negatively correlated with firm valuation. Moreover, this Index can be used as a measure of corporate governance practices for each company.
Furthermore, other researchers focused more on one of the most relevant tools of corporate governance: the board of directors. Generally, the interest of the members of the board is not necessarily aligned with the interests of the company, so the ongoing question of which type of board of directors is better to implement, regarding size and independence, has been looked at by different researchers (Yermack, 1996; Klein, 1998; Coles et al., 2008; Bhagat and Black 2002). Prior research has tried to find if there is any relationship between board characteristics, such as size of the board and degree of independence with firm value; but have failed to find any relationship between them. Boone et al. (2007) intended to test the relationship between the complexity of a company given by the number of business segments and the size and percentage of independent members in the board of directors. They were inspired by the work of Fama and Jensen (1983), implying that the board structure is conditioned by the complexity of the firm.
6
Due to the importance of corporate governance in recent years, and its concern to protect shareholders, throughout this paper, I intend to test the importance of having good corporate governance practices and how it affects the performance of the company and also testing these practices when there is a CEO turnover in the company. The reason for choosing to test corporate governance efficiency when there is a new CEO is that the CEO, as the leader of the organization, can bring instability when there is a shakeup, and shareholder value can be affected because of this. And more importantly if a company has good corporate governance practices, their performance should not be affected in a negative way because of the CEO turnover. Most of the previous research on this subject was focused on the CEO turnover originated form the poor performance of the company, the impact on the performance of the company and the many corporate governance measures that could also affect the performance of the company, like the board of directors (Coles et al. 2008). However there is not many research regarding the impact of the turnover given that the previous CEO was fired, and the importance of different corporate governance measures that could help the company during this adaptation process. Previous researchers also tried to find the relationship between corporate governance and firm performance, concluding that corporate governance is strongly correlated with a better operating performance (Bhagat and Bolton, 2008). They conducted their research by using the GIM and BCF indices as a measure for corporate governance. This is one of the main differences between their research and my thesis; the different methods to measure the corporate governance practices and the importance in different situations of the company that in the case of my thesis is after a CEO turnover.
The method for testing this relationship will be an OLS regression using Tobins Q as the dependent variable in order to measure firm performance, as Coles et al., (2008) and
7
Yermack (1996) used previously in their models. The variables board size, fraction of independent members in the board and the E-Index, independent variables, will be used as corporate governance tools trying to explain the performance of the company. Moreover, dummy variables to identify Nonforced leaves from previous CEOs will be added to the model to give more precision and to test if these have an impact on Tobins Q.
One of the challenges surrounding this subject is the endogeneity issues. Researches like Coles et al. (2008), Bhagat and Black (2002) had some problems with their methodology because of the relationship between Tobins Q, board structure and equity ownership. For example Coles et al. (2008), found that certain type of boards could actually lower the value of the company, because board structure and Tobins Q are jointly determined causing this endogeneity.
Other scholars like Murphy and Zimmerman (1992) found that some variables such as R&D, CAPEX and advertising are lower during the CEO turnover, resulting in endogeneity because these expenses could increase the value or improve the performance of the company, thus controlling for these variables could weaken the transition-year effect.
In this thesis, based on the previous work by Coles et al. (2008) and Yermack (1996), I am going to use an OLS regression using Tobins Q as the dependent variable and measure of the performance of the company. To test the impact of corporate governance, I will use the E-Index score, the fraction of independent members of the board and the size of the board. Based on the previous research and the endogeneity problems, I will use some of the control variables used by Yermack (1996) and Murphy and Zimmerman (1992). These variables are: Investment opportunities ( CAPEX/Sales), Firm size, CEO age, leverage and R&D expenses.
8
This thesis contributes to the existing literature in two ways. First, the corporate governance proxies used here, the E-Index, board size and fraction of independent members, have never been used in the same methodology. Previous researchers focused either on corporate governance indexes or in the impact of the board of directors, but combining these corporate governance proxies bring a different approach to this topic. The second and most important contribution is that there is no research regarding the importance of corporate governance practices when there is a new CEO in the office and how this could affect the performance of the company, furthermore this thesis also contemplates the impact in the performance when the previous CEO was forced to leave, the board of directors are the only ones that can take the decision of replacing the current CEO, and if a company has a solid board, it is expected that this decision has a positive impact in the performance of the company. A company with solid corporate governance practices should not be affected in a negative way in their results because of a management turnover.
The rest of the thesis is divided as follows: in the next section, papers related to this thesis are going to be discussed in order to see how they relate to my topic and to highlight the importance of their contribution for my thesis. After this first section, I am going to discuss the methodology followed in this thesis in order to answer the hypotheses made in the previous section. I will continue with the explanation of the data used and the summary statistics, after which the results obtained and their economic impact, as well as the contribution to the existing literature will be discussed in the result section. Following this section, there will be another section for robustness checks, and finally the conclusion, where I will discuss the impact of this thesis, its importance and some areas where can be improved for further research.
9 2. Literature review
In every successful company, efficient decision making involves delegation and diffusion of decision control and a separation between management and control (Fama and Jensen, 1983). All the weight of managerial decisions should not be held only by one person; this strategy also helps the company become more structured and with a higher level of corporate governance which separates ownership and control, otherwise the capability of the management can be scarce, due to the limited advice given by the board. Fama and Jensen (1983) mentioned that an effective separation between management and outside directors is beneficial for the company, because they do not or should not have incentives to collude with the management, and therefore be an unbiased force in the organization.
Bhagat and Bolton (2008) inspired by previous studies, sought to test the relationship between corporate governance and firm performance, using seven different measures of corporate governance, such as GIM and BCF Indices and CEO-Chair separation. The GIM index is a corporate governance index that uses 24 governance rules provisions as a proxy for shareholder rights. If a company has one of these provisions the index adds one point, meaning that if the company ends up with a high mark, the management has a lot of power and the shareholders are not very well protected. The firms in the lowest decile have stronger shareholder rights. The BCF index, also called the E-Index is very similar as the GIM index, the difference is that instead of using 24 provisions, Bebchuk et al., (2008) focused on the 6 provisions that showed a more direct link between corporate governance and firm performance. Bhagat and Bolton (2008) using these corporate governance proxies found that there is no relationship between corporate governance and stock market performance, meaning that the market does not take in consideration in any of the corporate governance practices and how can it affects the performance of the company. In contrast with the stock market performance, Bhagat
10
and Bolton (2008) found that good corporate governance practices, defined by GIM and BCF Indices have a positive impact on the operating performance of the firm. This point is quite important, because if some investors decide to also analyze some of the corporate governance practices from a company, like the board of directors, they could be ahead of the market and react quicker than the rest.
Corporate boards are one of the most important tools in corporate governance, and they have been the focus of many researchers. Lipton and Lorsch (1992) and Jensen (1993), studied board characteristics and firm performance, especially the premise that is there one structure of the board of directors that could fit all companies. In the study it was observed if the size and complexity of the firm should affect the structure of the board and if this structure is linked to the performance of the company. Coles et al. (2008) based on the previous research on this matter, conducted their study by separating complex and simple firms, to separate firms they took in consideration several measures: the number of operation lines, R&D expense and leverage. Using OLs regression with Tobin’s Q as a dependent variable, the fraction of independent members, the board size and the number of business segments (as a measure of the complexity of the firm) as independent variables, they found that bigger or more complex firms have lower fraction of insiders, and that companies with greater advising requirements, due to their complex structure, required more directors on the board, specifically more outsiders. Outside directors add value to complex firms in comparison to simple firms; in economic terms, adding one outside director increases firm value in $290 million in complex firms, but decreases by $21 million for simple firms.
Yermack (1996), similar to Coles et al. (2008), studied the effect of the board of directors in firm performance, but he focuses his research merely on board size. Also using Tobin’s Q as the dependent variable in an OLS regression, he found an inverse
11
relation between firm market value and the size of the board of directors. Yermack (1996) also suggests that smaller boards have a higher incidence of active monitors, which could add more value to the company. Some suggest that the board can be affected by the performance of the firm which will lead to reverse causality, but he failed to find evidence of this problem. In the same line, Boone et al. (2007) argued that the board structure is driven by the complexity of the firm, also implying that the board composition is the result of a negotiation between the outside members and the CEO. The board structure also depends on the stage of growth of the company; as a company keeps growing, it requires more specialized guidance and outside directors can give this type of advices Boone et al,. (2007). This conclusion goes in the same direction with the results found by Coles et al. (2008), that not all companies should have the same type of board of directors, because it could harm the company, and not the other way around. The board structure problem is something that has been questioned by many researchers, because it appears that none of them have come to any definite result. For example Baysinger and Butler (1985) argue that the lack of uniformity on board structures is due to the separation of ownership and control in large corporations. The market forces the management to adopt certain measures as control mechanisms, i.e. adding more outside directors as a measure to control agency problems. However, the importance of the board of directors is not the same for all companies; it depends on certain characteristics and unique circumstances, since market conditions tend to condition companies to conduct their process in a certain way. This is the reason why Baysinger and Butler (1985) and Coles et al. (2008) conclude that it is important that boards vary across companies. Because of the relationship between of board of directors and performance of the company, Hermalin and Weisbach (1991) used panel data as an attempt to eliminate the endogeneity issues because they found that poor performance
12
might lead to changes in the board, using cross-sectional regressions that include board composition and company performance might be biased, but using panel data they failed to find any relation between board composition and firm performance, these results converge with the one made by Yermack (1996). Continuing with the discussion regarding the importance of the structure of the board, Rosenstein and Wyatt (1997) argued that if a balanced board performs in the correct way, adding an insider instead of an outsider will increase the shareholder’s wealth. This is because inside directors might have some knowledge about the company that outside directors do not have. This is very important also when considering the type of company, i.e. a company investing considerable amount of money on R&D may require more inside directors.
Klein (1998), as a way to test the importance of corporate boards, changes the approach to not study only the board of directors, but also the finance and audit committee. She changes the approach because of the inconclusive results between the performance of the company and the members of the boards, analyzing only the board of directors; this was the reason why she was encouraged to study also the other boards in the company that have independent members, to see if she could have some more insights about the importance of the independent members. She found a positive relation between the performance and the percentage of inside directors in the investment and finance committees. This means that at first glimpse, there is no relationship between board and performance, but if you search deeper you could find some correlation. Analyzing the impact when a turnover takes place should lead me to different results, because it is at this point when the board of directors should show their value to the company. Klein (1998) concludes that inside directors could be a very valuable asset if used correctly. Bhagat and Black (2002) focused specifically on the role of the outside directors on firm performance. They found a strong inverse correlation between firm performance
13
and board independence. One of the reasons they gave for this inverse relation is that insiders can add value that in some occasions outsiders cannot.
So far the discussion has been focused in the impact of corporate governance in the performance of the company, seen in different approaches and using different proxies: the GIM, the E-Index, board size and board structure. But what about the impact when there is a turnover in the management, for example the CEO?
When the company is performing poorly, the board of directors come into action and decides whether replace the CEO or not (Kang and Shivdasani 1995). This is a serious decision because they have to decide if the poor performance of the company is because of the CEO or maybe some other conditions such as economics or political. In some cases hostile takeovers are a way to discipline the CEO for poor performance (Kang and Shivdasani 1995). It is important to mention that they also conclude that the turnover rate decreases when the management has high levels of ownership. This questions the autonomy of the board of directors, because even if the problem in the company is the management, they cannot do anything about it.
Kang and Shivdasani (1995) focused their study not only on firm performance resulting in CEO turnover, but also the different situations where is easier or more difficult for the company to replace their CEO, for example when the ownership of the top ten shareholders is considerably high, it is easier for them to replace the CEO because of their voting power. They also conclude that nonroutine turnover is related to ROA and negative pretax earnings, this goes in the same line as the conclusions of the researchers discussed above. Denis et al. (1997) as well as previous scholars mentioned above focuses their study in nonroutine turnover. They tried to test the impact of ownership structure, and the way to do it was by analyzing the rate of nonroutine top executive
14
turnover. Using an OLS regression, with the change of probability turnover as dependent variable and market stock return, ownership of officers and directors, outside blockholder dummy, the fraction of outside members and if there is any founding family member as a top executive as independent variables. They found that nonroutine turnover is higher in poor performing firms with low managerial ownership. These turnovers are also more frequent when the company has outside blockholders, because they may work as performance monitors.
Denis et al. (1997) argue that the ownership structure affects the turnover rate. Their results showed that twelve months before a top executive turnover there is an unusual increase in corporate control, and also the fraction of independent members in the board affects the turnover rate. Regarding the importance of fraction of independent members, they separate companies into two different groups depending if the fraction of outsiders was greater than 0.6 or lower than 0.6. They found a negative relationship between the performance of the company and the companies that have a fraction of outsiders higher than 0.6. This strong relationship between performance and turnover in firms with outsider-dominated board means that they work as performance monitors and based on the previous results discussed above, this also affects the turnover ratio.
Other scholars like Murphy and Zimmerman (1992) focus their study in the financial performance of the company and the CEO turnover, unlike Kang and Shivdasani (1995). In their study they analyze other aspects that could explain or be related with the CEO turnover, for example; R&D, advertising, CAPEX and accounting accruals. For example they argue that a decline in R&D expenses and CEO turnover maybe are related to managerial discretion, but this could also be driven by poor performance. Because of this possible correlation, it is more difficult to examine and interpret the
15
effects of poor performance, that could cause a decrease in some expenses like R&D or the CEO turnover, from the effects of managerial discretion.
They also found that turnover can occur if the relationship between the CEO and company deteriorates, and the possibility of turnover increases when there are some unexpected changes in the firm that the CEO cannot handle or does not have the necessary skills.
Denis and Denis (1995) aimed to test whether management turnover leads to improve firm performance. The reason to conduct this study was due to monitoring mechanisms which received attention because of highly publicized CEO dismissals at well-known firms. These CEO dismissals caused an increase in the importance of internal control mechanisms, but there is little evidence that these controls improve the performance of the company. They argue that if control mechanisms are effective, there should be two results that affect firm performance:
Greater incidence of top management changes in poorly performing firms. Improvement in firm performance following forced management changes.
This study presented solid evidence that firms where the CEO was forced to resign, the OROA decreased significantly prior the CEO turnover, and then there was a significant improvement following these changes. In contrast, in firms where the CEO wasn’t forced to resign, the OROA did not decline before the turnover, and after the turnover exhibited small performance improvements.
One of the biggest challenges of this topic is the endogeneity. Scholars like Hermalin and Weisbach (1991) claimed that poor performance leads to changes in the board, so there might be some biased in the analysis, because the changes in board composition might be the result of poor performances. Based on this, the board of directors is
16
modified by the performance of the company and not the other way around. There are some other scholars like Yermack (1996) that found no evidence that there exist some relationship between past performance and changes in the board.
Regarding the performance of the company and CEO turnover, scholars like Murphy and Zimmerman (1992) argue that one of the reasons is due because of the age of the CEO. Others like Kang and Shivdasani (1995) argue that the turnover of the CEO is also related to other variables like the decrease of R&D and CAPEX.
2.1 Hypotheses
H1: Companies with good corporate governance practices outperform companies with bad corporate governance practices
This hypothesis as Coles et al. (2008), Yermack (1996) and Bhagat and Bolton (2008) tries to see the impact of corporate governance in the performance of the company. The main difference with the previous researchers is that in this hypothesis I am going to use a different model with several corporate governance tools in order to control for some of the endogeneity issues that previous studies had, an also based on the work from Murphy and Zimmerman (1992), I am going to control for R&D, CAPEX and the age of the CEO, to mitigate some of the issues surrounding the turnover.
H2: Having good corporate governance practices in companies with an incoming CEO, given that the previous CEO did not leave the company voluntarily, improves their performance.
This hypothesis is inspired from the first hypothesis. Here I want to test the importance of having good corporate governance practices when there is a turnover, here we must see a bigger impact in the performance of the company, because some of these
17
practices, such as the board of directors could become more important under this circumstance.
3. Methodology
The econometric model that is going to be used in this thesis is an OLS regression inspired in the model used by Yermack (1996) and Coles et al. (2008), where they used an OLS regression and a 3SLS where they tried to test the impact of the board of directors in the company. What I used from this model was the measure for performance of the company, Tobins Q ratio. They defined Tobins Q ratio as follows:
Tobins Q = 𝑏𝑜𝑜𝑘 𝑎𝑠𝑠𝑒𝑡𝑠−𝑏𝑜𝑜𝑘 𝑒𝑞𝑢𝑖𝑡𝑦+𝑚𝑎𝑟𝑘𝑒𝑡 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝑒𝑞𝑢𝑖𝑡𝑦 𝑏𝑜𝑜𝑘 𝑎𝑠𝑠𝑒𝑡𝑠
And also based in the model used by Bhagat and Bolton (2008) where they used the GMI and BCF Indices as measures for corporate governance, but instead of using theses Indices, I decided to use the E-Index created by Bebchuk et al. (2008) as a measure of corporate governance practices because the provisions used in this Index had a stronger correlation with Tobins Q than the IRRC, that were the provisions used in the GMI Index.
The model for the first hypothesis is the following:
Tobins Q =β0 + β1E-Index i + β2 fraction of independent i + β3 Board size i + γ1CEOAGE i +
γ2Investop i + γ3 Firm Size i + γ4 Leverage i + γ5 R&D + α1 + ε
Where i=1998,…, 2006
18
For the second hypothesis the following model is going to be used:
Tobins Q = β0 + β1NewCEO*NotnotForced + β2E-Index*NewCEO + β3fraction of independent*NewCEO + β4Board size*NewCEO + γ1CEOAGE+ γ2Investop+ γ3Firm Size + γ4Leverage + γ5 R&D + α1+ ε.
Where i=1998,…, 2006
α1: is used to control for industry fixed effects
The dependent variable, Tobins Q is measure in the same way as in the models from is Yermack (1996) and Coles et al. (2008), explained above. The NewCEO * CEONotnotforced variable is the interaction between the dummy variable that is equal to 1 if the company has a turnover in the specific year and 0 otherwise and a dummy variable that equals 1 if the previous CEO did not resigned, retire or decease and 0 otherwise. This variable was supposed to be the interaction with the dummy variable for new CEO and a dummy variable if the previous CEO was fired, this in order to get stronger effect on Tobins Q. The information retrieved from Compustat only mention if the previous CEO resigned, retired or deceased, but there was no specification if he was fired, so I could not make the assumption that when there was not any specification about the reason he left the company was because he was fired. So this is the best approximation that I have of analyzing if the previous CEO was fired.
For the interaction between NewCEO and E-Index first I had to make a distinction between companies with good corporate governance practices from companies with bad corporate governance practices, according to the E-Index. In this Index each company was given a score, from 0 to 6, the low score means good corporate governance practices and therefore a high score means bad corporate governance practices. The
19
companies in the 25th percentile were consider in the group of good corporate governance and the companies in the 75th percentile in the group of bad corporate governance.
The E-Index gives scores of the company every two years, meaning that in the analysis I will take in consideration only the year of 1998, 2002, 2004 and 2006. The reason why start with 1998 is that there is no information regarding the board of directors from ISS before 1996 and the E-Index has information in 1995 and 1998.
The variables fraction of outsiders and Boardsize are there to show the relationship between the size and degree of independence of the board in the performance of the company. These two variables are interacted with a dummy variable that equals 1 if the company has a new CEO and 0 otherwise. The fraction of insiders is divided by ranges depending on the percentage of independent members in the board this separation is inspired in the work by Denis et al. (1997) where they separate companies depending if their fraction of outsiders was higher than 0.6. In order to get a better insight about the best mix between insiders and outsiders, I am going to separate them into 4 different groups. The first group is for companies that have less than 25% of independence in the board, the second group is for companies between 25 and 50%, the third group is 50 and 75%, and the fourth group is for companies that give higher than 75% degree of independence.
As mentioned on the previous sections, one important issue in this topic is the endogeneity. Based on the work by Yermack (1996) the control variables that I am going to use are Investment opportunities, Firm size, leverage and R&D expenses. The addition of these control variables will help me mitigate the endogeneity issues. Based
20
on the work by Murphy and Zimmerman (1992) I will add the age of the CEO as a control variable as well.
The variable Investment opportunities was calculated in the same way as Yermack (1996), CAPEX/Sales. Firm size was also calculated as stated by Yermack (1996), as the total value of equity plus long term debt plus preferred stocks.
The data that is going to be used in this research is from companies in the S&P 500 from 1998 to 2006. The reason for choosing companies listed in the US stock market is because is easier to get information about corporate governance. This is vital because I need information regarding the number of member in the board and whether or not they are independent, also information regarding the CEO such as the age, when he became CEO, in order to see when there was a turnover. The rest of the information is very standard in term of availability, because it is retrieved from financial statements.
4. Data and descriptive statistics
4.1 Data
The sample that is going to be used in this thesis is from companies listed in S&P 500 from 1998 to 2006. The reason why I chose this sample for the research is because there is no information regarding board of directors before 1996 and regarding the E-Index there is no information after 2006 and there is only information every two years and there was no information in the years of 1996 and 1997. It is important to mention that using this sample, the effect of the most recent crisis is not been considered.
The information regarding the E-Index was collected using the file provided by the Professor Lucian Ayre Bebchuk in hard.edu website. This file shows the level of corporate governance from each company based on the number of the six provisions
21
that provide a relationship with firm performance. Each company was given a score, from 0 to 6, the low score means good corporate governance practices and therefore a high score means bad corporate governance practices. The E-Index is from 1990 until 2006, using the firms followed by the Investor Responsibility Research Center, which includes all the companies listed in the S&P500. For purposes of this thesis, data from 1996 until 2006 was used.
The data collected regarding the information of the CEO such as age, date he became CEO and reason why he left the company was gather from Compustat Executive Information on the long term incentive award section. The information collected from the board of directors was gathered from ISS (formerly Riskmetrics) on the section of directors, in this section I was able to get information regarding the number of members of the board and whether or not they were independent. The rest of the data that helped me construct Tobins Q ratio according to Coles et al. (2008), book assets, book equity and market value of equity were gathered from Compustat.
4.2 Descriptive statistics
In the following table the number of companies that had a turnover in each year is presented. The total number of companies that suffered a turnover in the 10 years analysis was 323, the year that had the highest number of turnovers was 2000, while the year with the lowest number of turnovers was 2003 .
22
In the following table I am presenting some descriptive statistics regarding Board characteristics, firm characteristics and CEO characteristics. In the research made by Coles et al. (2008) the mean of Tobins Q was 1.79, in my research the mean for Tobins Q was 1.95, this is a little higher than Coles et al. (2008), meaning that in the years of this research (1998-2006) the market is paying a higher price for the company than the book value.
Regarding board characteristics is interesting to see that the mean of board size is above 10 and most of the members in the board are independent members (mean of 7.75). This means that on average, the companies have a board with almost 80% of independent members. According to some researchers like Klein (1998), Rosenstein and Wyatt (1997) and Bhagat and Black (2002) having more independent members not necessarily means an improve in the performance of the company and in some times it could have negative impact. Also Yermack (1996) consider that smaller boards are better for the companies, because they act more as constant monitors and consequently are more beneficial for the company.
Th is ta b le s h o ws th e n u mb e r o f CEO tu rn o ve r fro m 19 9 6 u n til 2 0 0 6
Turnover year Number of companies
1996 23 1997 28 1998 31 1999 34 2000 40 2001 25 2002 27 2003 15 2004 35 2005 30 2006 35 Total 323
23
Regarding the E-Index, we can see that the mean is a score of 2.16, considering that the highest and worst score is 6, on average the companies have a good corporate governance level according to this Index.
Table 2. Summary statistics.
This table shows the statistics regarding board characteristics, E-Index level, firm characteristics and CEO characteristics. The sample is for companies of the S&P 500 from 1998 to 2006.
5. Results
The first part of this section is the analysis and discussion of the results of the first model that tries to answer the first hypothesis regarding the importance of having good corporate governance practices.
The second part is the discussion of the results of the second model that tries to answer the second hypothesis regarding the importance of having god corporate governance practices when there is a new CEO in the office.
Mean 25th Percentile 75th Percentile
Board characteristics Board Size 10.99 9 15 Independent members 7.75 5 11 Insiders members 2.47 1 4 E-Index E-Index level 2.16 1 4 Firm Characteristics TobinsQ 1.95 1.17 2.73 Investment opportunities 0.06 .02 .08
Firm Size (millions) 14602 2474 16080
Leverage 0.34 0.18 0.46
CEO Characteristics
24
Table 3. Analysis of the performance of the company given by Tobins Q.
This table is separated into 2 different regressions, where regression (1) is given by Tobins Q = β0 + β1 E-Index + β2 fraction of independent + β3 Board size + γ1CEOAGE+ γ2Investop+ γ3 Firm Size + γ4 Leverage + γ5R&D + α1+ ε.. This table also reports the t-statistic for each variable, and (2) is using the same formula as in regression (1) but
analyzing β1 year by year separately. Constants were included in the regression but not reported, *, ** and *** indicate significance level at 10%, 5% and 1% respectively.
1 2 goodcorpgov .3625*** (8.15) badcorpgov -.2232*** (-4.83) goodcorpgov1998 .3918*** ( 2.33) goodcorpgov2000 .0398 (0.42) goodcorpgov2002 ,2697*** (2.61) goodcorpgov2004 .6122*** (5.15) goodcorpgov2006 1.2311*** (2.59) badcorpgov1998 .3479 (1.03) badcorpgov2000 -.6095*** (-4.57) badcorpgov2002 -.4301*** (-5.35) badcorpgov2004 .0316*** (0.26) badcorpgov2006 1.2609*** (-8.26) Boardsize .0679*** .0735*** (20.15) (10.75) Range 2 .2178*** .2605*** (2.31) (2.83) Range 3 .3445*** .3298*** (6.17) (5.92) Range 4 -.3350*** -.3250*** (-8.35) (-8.19) CEOage .0013 .0012 (0.45 ) (0.39 ) Investop 3.0291*** 3.1480*** (3.57) (3.88) Firmsize -.0016*** -.0032*** (-4.86) (-9.48) Leverage -.0345*** -.0335*** (-6.72) (-6.59) Fixed effects Yes yes
R2 0.4935 0.4941
25
In order to answer my first hypothesis H1: Companies with good corporate governance practices outperform companies with bad corporate governance practices. The following model of table 3 was used: Tobins Q = β0 + β1E-Index + α1 + γ1CEOAGE+ γ2Investop+ γ3 Firm Size + γ4 Leverage + ε
The first variable measures the impact of the E-Index on the performance of the company. Here we can see that having good corporate governance practices has a positive effect in the performance of the company, in the other hand having bad corporate governance practices has a negative impact in the performance of the company. This goes in the same line as Bhagat and Bolton (2008), they used the GIM index instead but the results are the same. When analyzing this variable on a year by year basis in regression 2, the results are the same, only the variables in 1998, the estimate is not significant, and in 2006 the effect is getting stronger. It would have been beneficial to have information for the following years in order to see the effect, because this could mean that the impact of this measure is higher in the recent years. In economic terms been in the 25th percentile means increasing your performance in $362 thousand USD. On the other hand been in t the 75th percentile could decrease your performance in $223 thousand USD.
For the other corporate governance measures; the board size and the fraction of independent members in the board. The board size has a positive effect on Tobins Q, this means that bigger boards benefits the company. This results are similar than the ones presented by Coles et al. (2008), where they found that Tobins Q is positively related to the size of the board in diversified and complex firms. On the other hand Yermack (1996) argue that smaller boards have more incentives to act as active monitors and therefore have a positive impact in the performance of the company. The fraction of outside members in the board was divided into ranges as explained in the
26
methodology section; it is observed that if the company has between 25 and 75% of independent members in the board there is a positive effect on the performance of the company. This goes in the same line as Coles et al. (2008) that found that for complex firms, adding independent members can improve the performance of the company. On the other hand if the percentage of independent members in the board is higher than 75% it has a negative impact in the company. This goes in the same line as Klein (1998) and Rosenstein and Wyatt (1997), they argue that there should be some balance in the board in order for the company to perform in a better way and adding inside members in the board might increase the performance of the company. Also Bhagat and Black (2002) found an inverse relationship between independent members in the board and firm performance; they argue that this might occur because of the knowledge of the day by day operation that insiders can add to the company. In economic terms the companies that have between 25 and 50% of independent members could benefit in $218 thousand USD and the companies between 50 and 75% have an increase of $345 thousand USD. On the other hand companies that have more than 75% of independent members in their board could have a negative impact of $335 thousand USD.
Based on the results obtained in table 3, it is possible to argue that companies that have good corporate governance practices, measured by E-Index level, board size and fraction of independent members in the board, have a better performance than companies that does not have good corporate governance practices. I can conclude this because with the results shown above, companies with a good E-Index score have a positive effect on the performance of the company as well as companies with big and balanced board of directors.
27
Table 4. Analysis of the performance of the company given by Tobins Q.
This table is separated into 2 different regressions, where regression (1) is given by Tobins Q = β0 + β1NewCEO*NotnotForced + β2 E-Index*NewCEO + β3 fraction of independent*NewCEO + β4 Board size*NewCEO + γ1CEOAGE+ γ2Investop+ γ3 Firm Size + γ4 Leverage + γ5R&D + α1+ ε.. This table also reports the t-statistic for each variable, and (2) is adding to the regression (1) the variable Notforced leaves as a complement for the regression (1), *, ** and *** indicate significance level at 10%, 5% and
1% respectively.
1 2
NewCEO*prev CEO not not Fired .2577***
(3.77)
goodcorpgov * NewCEO .0751***
(3.71)
badcorpgov* NewCEO -.3039***
(-3.58)
NewCEO1998 * Prev CEO not not Fired 2.0950***
(5.35)
NewCEO2000 * Prev CEO not not Fired -.6728***
(-4.37)
NewCEO2002 * Prev CEO not not Fired .4989***
(2.34)
NewCEO2004 * Prev CEO not not Fired 2.7022***
(11.47)
NewCEO2006 * Prev CEO not not Fired .9482***
(3.37) goodcorpgov1998 2.0249*** (5.07) goodcorpgov2000 1.8992*** (6.00) goodcorpgov2002 -.0407 (-0.17) goodcorpgov2004 -3.2094*** (-13.24) goodcorpgov2006 .5812* (1.85) badcorpgov1998 1.1465* (1.22) badcorpgov2000 -.0442 (-0.31) badcorpgov2002 -1.2380*** (-6.67) badcorpgov2004 -2.2702*** (-11.72) badcorpgov2006 -2.2702*** (-11.72) Boardsize * NewCEO .1691*** .1847*** (19.47) (21.07) Range 2 * NewCEO .1977 .1649 (1.56) (1.39) Range 3 * NewCEO .4546*** .4291*** (7.62) (7.37) Range 4 * NewCEO -.4792*** -.4789*** (-10.12) (-10.29) CEOage .0079*** .0048*** (5.14 ) (3.16 ) Investop 4.3058*** 4.2323*** (13.87) (13.53) Firmsize -.0013 -.0087 (-1.00) (-0.86) R&D .0039*** .0014*** (7.12) (7.45) Leverage -.0455*** -.0441*** (-8.04) (-8.06)
Industry Fixed effects Yes Yes
R2 0.7322 0.7436
28
For the second hypothesis: Having good corporate governance practices in companies with an incoming CEO, given that the previous CEO did not leave the company voluntarily, improves their performance. The following model was used: Tobins Q = β0 + β1NewCEO*NotForced + β2 E-Index*NewCEO + β3 fraction of independent*NewCEO + β4 Board size*NewCEO + γ1CEOAGE+ γ2Investop+ γ3 Firm Size + γ4 Leverage + γ5R&D+ α1+ ε.
The results of this regression are shown in table 4.
The first variable (β1) that shows if the previous CEO did not resigned, retired or deceased, has a positive effect on Tobins Q. The interpretation of this variable could be delicate because I actually tried to test the impact when the previous CEO was fired and therefore test the effect of the board of directors, as the only ones that could replace the CEO, but this information was not available and could only collect if the previous CEO resigned, retired or deceased. So I cannot assume that if the previous CEO did not left the company for one of the reasons stated above is because he was fired, but is the best approximation that I have. The positive effect of this variable means that the replace of the CEO benefits the company. This goes in the same line as Denis and Denis (1995) who also found that after the CEO turnover, the company suffered an important improvement in his performance. These results also complement the research made by Denis et al. (1997) and Kang and Shivdasani (1995), they argue that top executive turnover occurs after negative results. In here we can see that replacing the CEO was a good decision for the company. In economic terms this replacement means an increase of $258 thousand USD.
For companies with a CEO turnover that have good corporate governance practices, given by the E-Index level, the effect on firm performance is positive, although it is not
29
as strong as in the results shown in table 3. On the other hand, in the same line as the results shown in the previous table, companies with new CEO and bad corporate governance practices, have a negative impact on their performance. When analyzing this variable on year by year basis, companies with new CEO and good corporate governance practices have mix results, some years the effect is positive and in some others the effect is negative. For companies with new CEO and bad corporate governance show more constant results an easier to give a general meaning about the impact of having a bad corporate governance level. In economic terms, having a new CEO with bad corporate governance practices reduces their performance in $304 thousand USD.
The board size in companies with new CEO also have a positive relationship with Tobins Q, meaning that big boards benefit improves the performance of the company. This goes in the same line with the regression shown in the previous table. In economic terms an increase of members in the board size, increases the performance of the company by $169 thousand USD.
Having between 50 and 75% of independent members in companies with a new CEO in the office showed a positive effect on Tobins Q. Companies with new CEO and more than 75% of independent members in the board, showed a negative effect on Tobins Q. This goes in the same line with the analysis in the previous model but on the other hand Denis et al. (1997) found a negative relationship between outside dominated boards and firm performance. They divided the companies into two different groups: companies with less than 60% of outsiders and companies with more than 60% of outsiders. Based on my results I can conclude that having outside members is beneficial for the company, this is also claimed by Coles et al. (2008), but there should be certain amount of balance and the outside members cannot be over than 75%. In economic terms to have between
30
50 and 75% of independent members in the board could benefit the company by $454 thousand USD, but having more than 75% could harm the company by $479 thousand USD.
Kang and Shivdasani (1995) maintain that during times of uncertainty the board of directors becomes more relevant because they have to decide whether fire the CEO or not. Based on the previous results, I can conclude that companies with a new CEO benefit with big boards and balanced boards, with more outside members but no more than 75%. This conclusion goes in the same line with Rosenstein and Wyatt (1997), Baysinger and Butler (1985), and Coles et al. (2008), that companies benefit with balanced and well-structured boards.
6. Robustness checks
In order to test the robustness of the model, the way good and bad corporate governance practices was divided has been modified, in the main model, companies were divided into quantiles depending the level of E-Index, the companies that where in the 25th were considered of having good corporate governance practices, while the companies in the 75th percentile, for purpose of the analysis instead of using the 25th and 75th percentile I used 50th percentile, this in order to see if the effect of having good and bad corporate governance practices remain the same.
In addition to this, in model 2, instead of using the dummy variable for previous CEO that didn’t left the company because they resigned, retired or deceased; I am going to use the dummy variable for CEO that were not fired. In order to test the impact of the variable used before. The adjusted model is as follows:
31
Tobins Q = β0 + β1NewCEO*NotForced + β2 E-Index*NewCEO α1 + γ1CEOAGE+ γ2Investop+ γ3 Firm Size + γ4 Leverage + ε.
Where i=1998,…, 2006
α1: is used to control for industry fixed effects.
Table 5. Analysis of the performance of the company given by Tobins Q.
This table is separated into 2 different regressions, where regression (1) is given by Tobins Q = β0 + β1 E-Index + β2 fraction of independent + β3 Board size + γ1CEOAGE+ γ2Investop+ γ3 Firm Size + γ4 Leverage + γ5R&D + α1+ ε.. This table also reports the t-statistic for each variable, and (2) is using the same formula as in regression (1) but
analyzing β1 year by year separately. Constants were included in the regression but not reported, *, ** and *** indicate significance level at 10%, 5% and 1% respectively.
1 2 goodcorpgov .2103*** (6.10) badcorpgov -.4120*** (-5.13) goodcorpgov1998 .2741*** ( 2.65) goodcorpgov2000 .02470 (0.25) goodcorpgov2002 ,1985*** (2.98) goodcorpgov2004 .6475*** (4.85) goodcorpgov2006 1.1518*** (1.49) badcorpgov1998 .3514 (1.00) badcorpgov2000 -.5097*** (-4.74) badcorpgov2002 -.4201*** (-5.14) badcorpgov2004 .0214*** (0.16) badcorpgov2006 -1.1450*** (-8.56) Boardsize .0654*** .0475*** (21.02) (10.90) Range 2 .1804*** .2314*** (2.57) (3.03) Range 3 .3741*** .3148*** (6.58) (6.14) Range 4 -.2550*** -.3140*** (-8.50) (-8.40) CEOage .0024 .0017 (0.45 ) (0.24 ) Investop 2.1450*** 2.0514*** (2.87) (2.95) Firmsize -.0014*** -.0012*** (-4.98) (-4.48) Leverage -.02478*** -.0145*** (-6.10) (-6.69) R&D .0015*** .0074*** (7.12) (7.45) Fixed effects Yes yes
R2 0.4350 0.4341
32
Table 6. Analysis of the performance of the company given by Tobins Q.
This table is separated into 2 different regressions, where regression (1) is given by Tobins Q = β0 + β1NewCEO*NotForced + β2 E-Index*NewCEO α1 + γ1CEOAGE+ γ2Investop+ γ3 Firm Size + γ4 Leverage + ε. This table also reports the t-statistic for each variable, and (2) is adding to the regression (1) the variable Notforced leaves as a
complement for the regression (1), *, ** and *** indicate significance level at 10%, 5% and 1% respectively.
1 2
NewCEO*prev CEO not Fired .1314*** (2.55) goodcorpgov * NewCEO .0457***
(3.85) badcorpgov* NewCEO -.2014***
(-3.40)
NewCEO1998 * Prev CEO not Fired 1.010*** (5.42) NewCEO2000 * Prev CEO not Fired -.4157***
(-4.78) NewCEO2002 * Prev CEO not Fired .4547***
(2.58) NewCEO2004 * Prev CEO not Fired 2.4051***
(11.25) NewCEO2006 * Prev CEO not Fired .8578***
(3.37) goodcorpgov1998 .5332*** (5.23) goodcorpgov2000 1.0825*** (6.05) goodcorpgov2002 -0,8345*** (-13.50) goodcorpgov2004 -.1254 (-1.45) goodcorpgov2006 -.3506*** (-5.33) badcorpgov1998 -1.4305*** (-8.37) badcorpgov2000 .0725 (0.19) badcorpgov2002 -.7051*** (-2.81) badcorpgov2004 .5534*** (1.97) badcorpgov2006 .-54952 (-3.67) Boardsize * NewCEO .1457*** .1574*** (3.54) (11.05) Range 2 * NewCEO .1189*** .1649 (3.34) (1.39) Range 3 * NewCEO .0534 .4291*** (0.53) (7.37) Range 4 * NewCEO -.2628*** -.4789*** (-3.54) (-10.29) CEOage .0042*** .0048*** (5.78 ) (3.16 ) Investop 3.2057*** 4.2323*** (13.53) (13.53) Firmsize -.0087 -.0087 (-0.86) (-0.86) R&D .0014*** .0014*** (7.45) (7.45) Leverage -.0441*** -.0441*** (-8.06) (-8.06)
Industry Fixed effects Yes Yes
R2 0.5102 0.5047
33
On table 5 we can see that with the modification of how to consider good corporate governance companies, the impact of been in the group of good corporate governance practices is weaker than in the previous model. This is because in here instead of separating the companies in the 25th and 75th percentile, I decided to use the 50th percentile. So companies that are now in the group of good corporate governance practices, have a higher score and don’t have such as good practices as in the previous model.
On table 6 we can see that when we use the variable not fired instead of the variable not not fired, the impact on Tobins Q is lower, meaning that when use this variable, we have a good approach to test the case if the previous CEO was fired.
7. Conclusion
This thesis, like previous research, was focused on the importance of having good corporate governance practices and how this affects the company. The main contribution of this thesis is the study made on the importance of corporate governance practices when a new CEO enters the company. Previous research focused on the importance of different corporate governance proxies, like Coles et al. (2008), Yermack (1996), Bhagat and Bolton (2008) among others, or on the impact of CEO turnover in the performance of the company, like Denis and Denis (1995); but there has been no research about the importance of corporate governance when there is a new CEO in the office. When the previous CEO did not leave the company voluntarily, the impact on the performance of the company was positive. I could not make the assumption that the reason that the previous CEO did not leave the company voluntary was because he was fired, but it is the best approximation based on the data available. In this thesis, I found that companies whose previous CEO did not leave voluntarily improve their
34
performance. Also, companies with a new CEO suffered if they had a bad corporate governance level, given by the E-Index. Likewise, they benefit from large boards and from a fraction of independent members between 50 and 75%.
The other contribution is the combination of corporate governance measures used to test their importance on the performance of the company. Researchers like Coles et al. (2008), Yermack (1996), Bhagat and Black (2002) focused on the importance of the size and the structure of the board, and others like Bhagat and Bolton (2008) used corporate governance indexes to test their importance. In this thesis, I decided to combine all these proxies and see if they can relate to the performance of the company.
One of the limitations of this thesis had to do with the data. The data available did not specify if the previous CEO was fired; it only specified if he retired, resigned or deceased. Therefore, I could not assume that the reason the data base did not specify any of these situations was because he was fired. The other data limitation had to do with the E-Index; there was no available information after 2006, and after the crisis of 2008, corporate governance practices were adapted for more companies as precaution measures. Thus, the impact on the performance of the company could be different from the one presented in this thesis.
Another important limitation of my research concerns endogeneity issues, where the performance of the company is linked to economic issues that my model is not
considering, and therefore the interpretation of the results might be biased. Also, some scholars such as Hermalin and Weisbach (1991) claimed that poor performance leads to changes in the board; thus there might be some bias in the analysis, because the changes in board composition might be the result of poor performances. Hence, the
35
be biased, even though some other scholars like Yermack (1996) found no evidence that there is a relationship between past performance and changes in the board. Regarding the CEO turnover, Kang and Shidasani (1995) argue that these turnovers might occur because of market conditions that are difficult to control. Murphy and Zimmerman also argue that the turnover might occur because of the deterioration between the
36 References.
Baysinger, B. & Butler, H. (1985). Corporate Governance and the Board of Directors: Performance Effects of Changes in Board Composition. Journal of Law, Economics, and Organization, Vol. 1 (1), 101–124.
Bebchuk, L. A., Cohen, A. & Ferrell, A. (2008). What Matters in Corporate
Governance? The Review of Financial Studies, Vol. 22 (2), February 2009, 783-827. Bebchuk, L.A. & Weisbach, M.S. (2009). The State of Corporate Governance Research. The Review of Financial Studies, Vol. 23 (3), March 2010, 939-961.
Bhagat, S. & Black, B. (2002). The Non-Correlation between Board Independence and Long Term Firm Performance. Journal of Corporation Law, Vol. 27 (2), 231-274 Bhagat, S. & Bolton, B. (2008). Corporate Governance and Firm Performance. Journal of Corporate Finance, Vol. 14 (3), 257–273.
Boone, A.L., Field, L.C., Karpoff, J.M. & Raheja, C.G. (2007). The Determinants of Corporate Board Size and Composition: An Empirical Analysis. Journal of Financial Economics, Vol. 85 (1), 66–101.
Coles, J. L., Daniel, N.D. & Naveen, L. (2008). Boards: Does One Size Fit All? Journal of Financial Economics, Vol. 87 (2), 329-356.
Denis, D. J. & Denis, D. K. (1995). Performance Changes Following Top Management Dismissals. The Journal of Finance, Vol. 50 (4), 1029-1057.
Denis, D. J., Denis, D.K. & Sarin, A. (1997). Ownership Structure and Top Executive Turnover. Journal of Financial Economics, Vol. 45 (2), 193-221.
Fama, E. & Jensen, M. (1983). Separation of Ownership and Control. Journal of Law and Economics, Vol. 26 (2), 301–325.
Hermalin, B. & Weisbach, M. (1991). The Effect of Board Composition and Direct Incentives on Firm Performance. Financial Management, Vol. 20 (4), 101–112. Jensen, M. C. (1993). The Modern Industrial Revolution, Exit and the Failure of Internal Control Systems. Journal of Finance, Vol. 48 (3), 831-880.
Kang, J-K. & Shivdasani, A. (1995). Firm Performance, Corporate Governance and Top Executive Turnover in Japan. Journal of Financial Economics, Vol. 38 (1), 29-58. Klein, A. (1998). Firm Performance and Board Committee Structure. Journal of Law and Economics, Vol. 41 (1), 275-304.
Lipton, M. & Lorsch, J. W. (1992). A Modest Proposal for Improved Corporate Governance. Business Lawyer, Vol. 48 (1), 59-77.
37
Murphy, K. J. & Zimmerman, J. L. (1993). Financial Performance surrounding CEO Turnover. Journal of Accounting and Economics, Vol. 16 (1-3), 273- 315.
Rosenstein, S. & Wyatt, J. (1997). Inside Directors, Board Effectiveness and Shareholder Wealth. Journal of Financial Economics, Vol. 44 (2), 229–250.
Schleifer, A. & Vishny, R. W. (1997). A Survey of Corporate Governance. The Journal of Finance, Vol. 52 (2), 737-783.
Yermack, D. (1996). Higher Market Valuation of Companies with a Small Board of Directors. Journal of Financial Economics, Vol. 40 (2), 185-211.
