2 Abstract
In this thesis research is conducted with regard to benchmarking peer group and managerial ownership guidelines.
The first step in this thesis was to gain information, through existing literature, about
corporate governance, managerial ownership guidelines and benchmarking. In short, the literature states that benchmarks (i.e. peers) have a significant influence on a a company’s decision. This also applies for corporate governance (Granovetter, 1985).
The empirical part is divided in three research subtopics, namely in having, adopting and
changing a managerial ownership guideline. For the first subtopic there is a significant and positive relation between the fraction of peers that has a managerial ownership guideline and a S&P company having it. The data with regard to the second and third subtopic overlap each other. Due to this reason the estimators in the models are quite similar. For the second subtopic there is a non-significant negative relation between the fraction of peers making a change to the guideline and the S&P company doing it. For the third subtopic it is shown that the fraction of peers that adopts a guideline is significantly and negatively related to a S&P company making an adoption.
It is emphasized that the findings in this study can only be used as indications.
3
Table of Contents
Part 1: introduction ... 4
Part 2: literature review ... 8
2.1 Corporate governance and the agency problem ... 8
2.2 Managerial ownership guidelines ... 9
2.3 Benchmarking ... 14 2.4 Hypotheses ... 18 Part 3: methodology ... 19 3.1 Sample ... 19 3.2 Materials ... 19 3.3 Procedure ... 19
3.4 Information concerning sample ... 21
3.5 Variables and measures ... 21
3.6 Model ... 25 Part 4: results ... 28 4.1 General analysis ... 28 4.2 Regression analysis ... 34 4.3 Testing of hypotheses ... 37 Part 5: discussion ... 38 5.1 Analysis results ... 39 5.2 Omitted variables ... 41
5.3 Limitations of literature review ... 42
5.4 Limitations of empirical research ... 43
5.5 Suggestions for further research ... 43
Part 6: conclusion ... 43
Part 7: references & appendix ... 46
7.1 References ... 46
4
Part 1: introduction
Corporate governance relates to ways in which suppliers of finance to corporations assure themselves of getting a return on their investment (Shleifer, 1997). The concerning framework exists out of mechanisms that assure that the firm is run in the interest of shareholders. Corporate governance nowadays is indispensable because in most companies there is a separation between ownership and control. This separation creates a misalignment of interests, also known as the agency problem. Through incentive contracts the company tries to prevent agency problems from happening. These contracts can take the forms of for example share ownership, or threat of dismissal (Jensen, 1976). Managerial share ownership is the dominant form of corporate governance arrangement in continental Europe and other OECD countries (Holderness, 1999).
To improve managerial share ownership the SEC (Securities and Exchange Commission)
invented the managerial ownership guideline in 2005. This is a policy that is set up to encourage managerial ownership. The board of directors, who among focus on corporate governance, are interested in managerial ownership guidelines because they want to
effectively align the interests of different stakeholders. Research and information about these guidelines are however limited. Due to this lack of information there is an unclear view of what the factors are that determine the usage of managerial ownership guidelines.
Consulting firm Frederic W. Cook states that a managerial ownership guideline is a key
component in a company’s compensation program (2008). One of the most notable
determinants of compensation is benchmarking (Bizjak et al., 2008). Seen this information it is presumed that benchmarking also has an impact on using managerial ownership guidelines.
Managerial ownership guidelines has a broad scope, therefore it is important to define a
framework for this research. A few studies have been dedicated to determinants of
managerial ownership guidelines, but none of them focussed on the influence of benchmarks. In this thesis the determinant benchmarking peer group is studied. The research will focus on companies that were listed on the S&P index between 2006 and 2011. S&P companies are chosen because large firms tend to have a managerial ownership guideline. This research thus adds information to about the relation between benchmarks and managerial ownership
guidelines. This research topic is interesting to examine because it concerns a relatively new corporate governance mechanism which is quite unexplored.
5 The research question is as follows:
Does benchmarking peer group act a role in the usage of a managerial ownership guideline? This research question will be answered by studying existing literature and performing empirical research.
The literature shows that actions of managers are not only determined by economic
incentives and information asymmetries, but also by the social context of a firm (Granovetter, 1985). One of the ways that social context is used to gain information is by monitoring
competitors. This is done to improve the quality of decision making, anticipate on competitors and create a competitive advantage. Prašnikara et al. (2005) describes this as the concept of benchmarking. The literature also states that benchmarking should be used for all important activities so that the company can benefit from the advantages.
Benchmarking is usually performed on the basis of peers that are similar with the firm. For example, Bizjak et al. (2011) found that S&P 500 companies often choose peers that are similar for benchmarking. Smaller companies however often choose firms that are lager in terms of size and have more stock return.
Granovetter (1985) states that companies move with time towards homogenization because they get influenced by their peers. This mutual awareness across companies, one firm is doing what other firms are doing, is named an interorganizational network. This imitation across firms also applies for corporate governance (Davis, 1991). In addition, the use of
benchmarking has a significant impact on CEO compensation (Bizjak et al., 2008). A key component of CEO compensation is a managerial ownership guideline (Frederic W. Cook, 2011).
Seen this relation, it thus can be stated that benchmarks (i.e. peers) have an influence on the usage of a managerial ownership guideline. In other words, as the fraction of the peer group that has a managerial ownership guideline increases the probability of the firm using a
6 The relationship derived from the literature review will be tested in an empirical research. In this empirical study all the firms that were in the S&P 500 (large cap index) at any point between 2006 and 2011 are included in the sample. Peer group data of these firms is gathered until 2012. The total sample contains 4248 observations. These observations are based on 1294 unique companies. The research has been conducted through hand collection of data and using databases. Data with regard to peer groups and managerial ownership guidelines are almost unavailable. Therefore the required data had to be hand collected. For the control variables databases are used, namely Wharton Research Data Services (Risk Metrics & CRSP) and Thomson Reuters Datastream. In this study the main independent variables are the fraction of peers with a guideline, fraction of peers that made a change to the guideline and the fraction of peers that adopted a guideline. To control for other factors that are related to managerial ownership guidelines variables with regard to firm size, firm risk, firm performance, debt, free cash flow and corporate governance are included in the
regression models.
The goal of this research is to increase the information with regard to managerial
ownership guidelines and benchmarks. Currently the literature has focused almost entirely on managerial ownership, but focused less on managerial ownership guidelines and the influence of peers on it. The relevance is that firms gain knowledge about this guideline and are more conscious about it. Most U.S. firms compare pay and policy to a peer group of 10 to 20 rival firms. In addition, firms frequently tell shareholders that benchmarking is important
(Albuquerque, 2012). This shows that information produced through this research is of value. The research also increases knowledge about the alignment of interests of shareholders and executives on long-term.
Firstly the literature review is discussed. This is done by viewing the subjects corporate
7 the methodology of the empirical research is discussed. Subsequently, the results of the research are viewed. This is done by examining the descriptive and inferential statistics. In part four the discussion is treated. This part exists of analysis results, omitted variables, limitations of literature review, limitations of empirical research and suggestions for further research. In the last part the conclusion of this study is discussed.
8
Part 2: literature review
2.1 Corporate governance and the agency problem
Corporate governance relates to ways in which suppliers of finance to corporations assure themselves of getting a return on their investment (Shleifer, 1997). The concerning framework exists out of mechanisms that assure that the firm is run in the interest of shareholders.
Corporate governance, between investors and shareholders, emerges from the
separation of ownership and control in modern corporations. This matter is also known as the agency problem. It refers to the issues financiers have in assuring that their investments are not expropriated or wasted on unattractive projects. This problem arises whenever investors want to exercise control differently from the managers. Ownership rights are the cash flow rights of shareholders and control rights are is the actual power of the management of a corporation (Holderness, 2003).There can be a difference between ownership and control rights because of for example diffused ownership.
Shleifer and Vishny (1997) state that product market competition may reduce the returns
on capital and thus cut the amount managers can possibly expropriate, but it does not prevent the managers from expropriating the competitive return after the capital is sunk. Thus, it can’t solve issues regarding corporate governance.
Berle and Means (1932) started the discussion of ownership and control and promoted
the view of the widely held corporation in the United States. They state that the separation of ownership and control has totally eliminated the checks and balances that owners once
exercised over management. When the ownership of a company is dispersed and managers do not own a considerable amount of equity a misalignment of interests is created. An example of this misalignment is that the company’s assets are used for the benefits of managers instead of shareholders.
Shleifer and Vishny (1997) note that incentive contracts are necessary to induce the
management to act in the investor’s interest. The optimal incentive contract is determined by the manager's risk aversion, the importance of his decisions, and his ability to pay for the cash flow ownership up front (Jensen, 1976). These contracts can take the forms of for example share ownership, or threat of dismissal. This most common approach is also known as the principal-agent view. It assumes away all coordination problems between shareholders and the board entirely. And it also presumes that the shareholders agree on an optimal incentive
9 contract with the CEO (Weisbach, 2007). On the other hand Bebchuk & Fried (2005) developed an alternative view concerning executive compensation. Their conclusion is that the evidence overwhelmingly supports the notion that CEO’s have great sway over their own pay, and that executive compensation is best understood through this lens, rather than through an optimal principal–agent contract in which the shareholders capture all the rents (Weisbach, 2007). This is also known as the managerial power approach.
Partial ownership and control in the hands of a few investors seems to be the preferred
mechanism to overcome the collective action problem in many countries. It is a dominant form of corporate governance arrangement in continental Europe and other OECD countries
(Holderness, 1999). Shleifer and Vishny (1997) also state that matching significant control rights to significant cash flow rights is a major approach to overcoming the agency problem.
2.2 Managerial Shareholder Guidelines
In the following section managerial shareholder guidelines and insider ownership are discussed extensively. These concepts are intertwined with each other and the different aspects of them are viewed. Managerial shareholder guidelines are the core of this study, but to fully understand the framework managerial ownership is also viewed.
2.2.1 Definition
A managerial stock ownership guideline is a policy with regard to executive ownership. These guidelines are commonly used to align director and shareholder interests (Cook, 2011). A concept that is connected to this guideline is managerial stock ownership. This concept implies the amount of equity an executive has in the company that he/she works for (Himmelberg, 1999).
2.2.2 Purpose of managerial ownership (guidelines)
Berle and Means (1932) noted that when shareholders are too diffuse to monitor managers, corporate assets can be used for the benefit of managers rather than for maximizing
shareholder wealth. A well-known solution to this problem is to give managers an equity stake in the firm. This helps to resolve the moral hazard problem by aligning managerial interests with shareholders’ interests.
10 inefficient. This is because the pay of executives lacks sensitivity with regard to firm wealth, this is consistent with the findings of Berle and Means (1932). Jensen and Murphy (1990) arrive at a striking number that executive pay rises (and falls) by about $3 per every $1000 change in the wealth of a firm's shareholders.
To increase the sensitivity and movement between executives’ pay and firm wealth
managerial ownership guidelines were developed. They stimulate the alignment of director and shareholder interests (Cook, 2011).
2.2.3 Determinants of managerial stock ownership
In this part the determinants of adopting a managerial ownership guideline are examined. Companies have different motivations to adopt a guideline. Demsetz and Lehn (1985) state that managerial ownership varies with firm characteristics.
The factors and determinants that lead to an adoption of a managerial ownership guideline are also various. Based on literature review it can be stated that the most notable determinants of adopting a guideline are: firm risk, firm size, firm performance, debt, free cash flow, corporate governance and benchmarks. The last determinant, benchmarks, is extensively discussed in paragraph 2.3 seen that it is the main independent variable in this study.
2.2.3.1 Firm risk
In 1985 Demsetz and Lehn conducted research with regard to the determinants of managerial ownership. Their research points out that there is a relation between the riskiness of the firm and managerial ownership. When firm risk increases, the control that shareholders exercise becomes less. This could lead to moral hazard among the management. This conclusion means that a relatively higher amount of managerial ownership is necessary in risky firms, because the scope for moral hazard is greater. Managers in risky firms thus need a higher ownership stake to align the interests with the shareholders.
More managerial ownership will however expose the management to more risk, because
their compensation is largely linked to the firm’s performance. The requirement to own a certain amount of equity will expose management to extra risks and thus increase contracting costs. In short, it means that the more volatile a firm is, the more risk is put on the management. Cao and Gu (2011) state that the adoption of a managerial ownership guideline is negatively related with firm risk.
11
The riskiness of the firm is measured by the volatility of the stock price (Himmelberg et
al., 1999). This method to measure firm risk is used in this study. 2.2.3.2 Firm size
The relation between firm size and moral hazard is ambiguous (Himmelberg et al., 1999). Monitoring and agency costs are higher in large firms and this increases desired managerial ownership. On the other hand, large firms will enjoy economies of scale with regard to monitoring.
A study of Watts and Zimmerman (1986) states that large firms are subject to more
criticism from the public. There corporate governance receives critical attention from the public. The study of Cao and Gu (2011) states that larger firms are under more pressure to adopt a managerial ownership guideline.
Firm size is measured by Himmelberg et al. (1999) by the logarithm of firm sales. This
method is also used in this study. 2.2.3.3 Firm performance
The existing literature does not provide consistent evidence about the relation between managerial ownership and firm performance (Core & Larcker, 2002). Mork et al. (1988) for example state that managerial ownership is too low at most firms and that there is positive relation between firm performance and managerial ownership. On the other hand, Himmelberg et al. (1999) state that there is little evidence of a relation between firm performance and managerial ownership. Core and Larcker (2002) clarified this issue and showed that firms that are below equilibrium levels of managerial equity ownership have a positive relation between firm performance and managerial ownership.
Cao and Gu (2011) elaborated on the research of Core and Larcker (2002) and showed
that poorly performing firms are more likely to adopt a managerial ownership guideline.
Firm performance is measured by Core and Larcker (2002) by the firm’s stock return.
This method is also used in this study. 2.2.3.4 Debt
One of the benefits of debt for shareholders is that it has a positive effect on corporate governance (Himmelberg et al., 1999). Debt reduces the agency costs of free cash flow by
12 reducing the cash flow available for spending by the management (Jensen, 1986). In addition, debtholders also monitor the actions of the management.
The effect of debt on managerial ownership is ambiguous. On the one hand debtholders
would prefer that the management holds equity. On the other hand due to the lower agency costs there becomes less necessity to own equity (Cao & Gu, 2011).
Cao and Gu (2011) also state that the relation between managerial ownership guidelines
and debt is ambiguous. Debt can have a motivating effect on mangers to work harder and improve firm performance. But on the other hand, debt can reduce agency problems and thus reduce the need to impose ownership guidelines on managers.
Debt is measured by Benson and Lian by the debt ratio. This is the book value of debt
divided by the total assets of the concerning firm. This method is also used in this study. 2.2.3.5 Free cash flow
Studies of Palia (1998) and Himmelberg et al. (1999) show that high free cash flows creates agency problems. These agency problems can be mitigated with higher levels of equity ownership by the management. In short, it can be stated that free cash flow has a positive association with managerial ownership (Palia, 1998). Through managerial ownership the incentives of shareholders and managers become more aligned. This reduces the risk that managers waste free cash flow or invest in bad projects (Jensen, 1986).
Free cash flow is measured by Palia (1998) by the operating income (proxy for free cash
flow) divided by the total assets. Operating income is divided by total assets to correct it on firm size. This method is also used in this study.
2.2.3.6 Corporate governance
Tirole (2006) describes corporate governance as the ways in which suppliers of finance assure themselves of getting a return on their investment. The aim of corporate governance is to let corporate insiders act in the best interests of the providers of the funds. If managers incentives are not aligned with those of the shareholders agency problems will exist (Shleifer, 1997). One of the tools to improve corporate governance is to let managers hold firm equity (Himmelberg et al., 1999). This improves the alignment of interests between the shareholders and managers.
Studies of Himmelberg et al. (1999) and Jensen (1976) state that firms with greater
13 ownership among managers. This is for the purpose of aligning the interests of shareholders and managers. This means that there is a negative relation between corporate governance and managerial ownership.
Following these studies Cao and Gu (2011) state that firms with agency problems are
more likely to use a managerial ownership guideline. The purpose of this guideline is to reach an efficient contract and maximize shareholder value. Thus, there is also a negative relation
between managerial ownership guidelines and corporate governance.
Gompers et al. (2003) and Bechuk et al. (2009) have each created an index to proxy for
corporate governance. The G-index has been developed by Gompers et al. and the E-index by Bebchuk et al. Both indexes are based on provisions that are related to corporate governance. The G-index and E-index are both used in this research to control for corporate governance. 2.2.4 Managerial shareholder guidelines among companies
Of the 100 largest companies in the S&P 500, 95% of them have policies in place (Cook, 2010). Another study of Cook (2011) based on a sample of 240 U.S. firms from the Large Cap, Mid Cap and Small Cap shows that fifty-five percent of the companies have some type of ownership guidelines in place. Of the companies that do not have ownership guidelines, 51% percent are Small Cap companies, followed by Mid Caps (32%), and a small minority of Large Caps (17%). From this can be inferred that Large Cap firms adopted a stock ownership guideline in higher degree.
These policies typically require executives to achieve pre-determined stock ownership levels within a specified time period, with the goal that executives build a significant equity stake in the company they manage to provide focus on long-term shareholder value creation. Requiring executives to hold a significant portion of their net worth in company equity is also believed to mitigate excessive risk-taking that propels short-term gains at the expense of long-term value creation (Cook, 2010).
There is a prevalence of various types of stock ownership policies among companies.
The most common approach to ownership guidelines is a multiple of salary approach, which requires executives to hold a defined dollar value of stock that grows in line with salary increases. The median multiple for the CEO is 5X.
Another manner to comply with the guideline is through the use of retention ratios in
14 executive should retain a certain percentage of ‘net profit shares’. The benefit of using a retention ratio in combination with a traditional multiple-of-salary or fixed-share guideline is that companies need not set a period of time by which the required ownership levels must be
achieved.
An alternative approach to a retention ratio is a holding period (Cook, 2010). The
difference between retention ratios and holding periods is based on the period of time executives are required to hold the net profit shares. Retention ratios are enforced for an indefinite period of time (e.g., the executive’s career or until multiple of salary guidelines are met). Holding periods are enforced for a specific period of time (e.g., one year after exercise of options).
2.3 Benchmarking
In the following part benchmarking is extensively discussed. Several subjects around this concept are viewed to create a complete view.
2.3.1 Definition
The following definition of benchmarking is used in this study. Benchmarking is the process of comparing and measuring your organization against others, anywhere in the world, to gain information on philosophies, practices, and measures that will help your organization take action to improve its performance (Coers et al, 2001).
Kyrö (2003) states that benchmarking is a general business concept and the core of it
are aspects of evaluation and improvement by learning from others. 2.3.2 The concept of benchmarking
When a company wants to be better than its competitors it should achieve above-average performance. For this reason companies develop their capabilities. In addition, the competitors are monitored so that the company can correctly anticipate on them. Information is gained and used to improve the quality of decision making and thus create a competitive advantage. Prašnikara et al. (2005) describe this as the concept of benchmarking.
With benchmarking the assumption is that an organization will improve its own
performance if it copies an organization that exhibits the best performance, product, or process (Coers et al, 2001). This assumption is proven in research of Carr (2006). It was shown that
15 there is a positive relation between benchmarking peer groups and firm performance.
Prašnikara et al. (2005) state that benchmarking can be divided in four types. These are
competitive advantages, strategies, process and performance.
The purpose of benchmarking competitive advantages is gain knowledge about factors which are the base of competitive advantages of other companies. It is used to improve the company’s long run advantages.
Benchmarking of strategies is used to gain knowledge about the strategies of other companies. The purpose is to use this knowledge to improve the company’s own strategy and realize strategic objectives.
The goal of process benchmarking is to learn about business processes and activities by which other companies successful implement their strategies. This type of benchmarking is used to improve the efficiency of the companies own activites, processes and strategies.
Performance benchmarking is used to gain knowledge about other companies performances. The own business performance can then be compared to those of other companies. The results of this are used to improve strategic objectives.
2.3.3 Model of benchmarking
Prašnikara et al. (2005) state that benchmarking is a useful tool if it is used integrally in the company. In other words, benchmarking should be used for all important activities so that the company can benefit from the advantages.
There exist different models for executing benchmarking. The following basic phases are
defined by Coers etl al. (2001) and Andersen et al. (1996); planning, collecting business information, comparison & analysis and application.
The planning phase is used to clearly define the objectives and purpose of benchmarking. It is important to select measures that are clear and accurate so that an effective gathering of information will be enabled.
In the collecting business information phase data about other companies are gained. The data that are gathered should be relevant and comparable to create quality knowledge.
The goal of the comparison and analysis phase is to analyze the data with regard to other companies and create new business knowledge about the subject of benchmarking. The methodology for the analysis depends on the available information and the purpose of benchmarking.
16 In the last phase the outcome of the analysis is used for the decision-making. This is called the application phase.
2.3.4 Information spread through benchmarking
Dimaggio and Powell (1983) conducted a study with regard to organizational theory. This theory presumes that companies that are related to each other, through their business practices, are active in the same organisational field. Companies in an organizational field are often
peers/benchmarks. Their research shows that companies in the same organisational field move with time towards homogenization. This homogenization is created by the social structure that is present in an organizational field (Granovetter, 1985).
Haunschild (1993) states that the imitation among companies in the same field is a
strategic response. Through this imitation the second-movers take advantage, because the first-movers absorbed the risk. This imitation corresponds to the theory of Granovetter (1985) and can lead to homogenization in an organizational field. This social mechanism of mutual
awareness across companies is named an interorganizational network. This social mechanism suggests that one firm is doing what other firms are doing (Haunschild, 1993).
The study of Davis (1991) shows that the interorganizational network influences the perception of interests and the manner in which they are pursued. Davis (1991) states that the imitation across companies, created by the social context, also applies to corporate governance.
In short, based on the organizational theory can be stated that actions of managers are
not only determined by economic incentives and information asymmetries. The social context has significant influence on managerial action (Granovetter, 1985).
Based on this theory is assumed that peers and benchmarks have significant influence on managerial action, in specific on corporate governance.
2.3.5 Benchmarking and remuneration
Consulting firm Frederic W. Cook states that a managerial ownership guideline is a key component in a company’s compensation program. It helps ensure an alignment of executive and director interests with those of shareholders. Seen that literature about benchmarks and managerial ownership guidelines is limited the relation between benchmarking and executive compensation is viewed.
17 the compensation with a peer group of similar companies. The use of benchmarking is a
common practice and it has a significant impact on CEO compensation. Bizjak et al. (2008) examined 100 random companies in the S&P 500 index. It was found that 96 out of the 100 companies use a peer group to determine CEO compensation.
Bizjak et al. (2008) states that the use of benchmarking, to determine CEO
compensation, has an advantage and disadvantage. A disadvantage of benchmarking CEO compensation is that it will become independent of CEO or firm performance. This means that CEO pay is not tied to CEO or firm performance and this can create a misalignment of interests with the shareholders.
An advantage of benchmarking pay is that it is an efficient mechanism to determine the reservation wage of the CEO. It thus is necessary information for the compensation process (Bizjak et al., 2008).
The Stock and Exchange Commission (SEC) requires companies that use benchmarks to
determine pay to discuss it in the proxy statement. On the basis of this information Bizjak et al. (2011) found that companies choose peer firms in a way that inflates CEO compensation. Murphy (1999) and Bizjak et al. (2011) state that companies tend to choose peers of similar size, same industry, similar credit ratings and accounting performance. S&P 500 companies tend to choose peers that are also part of the S&P 500 index.
Bizjak et al. (2011) found that S&P 500 companies often choose peers that are similar.
Smaller companies however more often choose peers that are larger in terms of firms size and have more stock return. The result of this is that they benchmark themselves against larger and more highly paid peers. This will then increase the chances on higher CEO compensation.
Seen that benchmarking has a significant impact on executive’s compensation it is
assumed that it will also have a significant impact on a company’s decision to use a managerial ownership guideline.
18 2.4 Hypotheses regarding managerial ownership guideline and benchmarking
Several studies in the theoretical framework state that benchmarks have a significant impact on the compensation program and corporate governance of a company (Bizjak et al., 2008; Davis, 1991). Since benchmarks have an impact on the company’s corporate governance decisions it can be derived that they also have an impact on the use of managerial ownership guidelines. The social mechanism of mutual awareness across companies thus leads to imitation
(Granovetter, 1985). In this study is examined if benchmarks also have an impact on decisions concerning managerial ownership guidelines.
The following hypotheses are tested;
Hypotheses 1: a firm is more likely to use a managerial ownership guideline if firms in its peer
group also use it.
Hypotheses 2: a firm is more likely to make a change to a managerial ownership guideline if
firms in its peer group also make it.
Hypotheses 3: a firm is more likely to adopt a managerial ownership guideline if firms in its
19
Part 3: methodology
In this part the methodology of the empirical study is discussed. First the sample used to conduct a survey is discussed. This is followed by materials and procedure regarding the research tool. Finally the variables and the model of the empirical research will be presented. 3.1 Sample
In this empirical study all the firms that were in the S&P 500 (large cap index) at any point between 2006 and 2011 are included in the sample. Peer group data of these firms is gathered until 2012.
The reason that firms at any point between 2006 and 2011 are included is to avoid a survivorship bias. Namely if the sample only consists of firms that were in the S&P 500 until 2012 the sample will be biased to including only firms that survived until then.
The total sample contains 4248 observations. These observations are based on 1294 unique companies.
3.2 Materials
The research has been conducted through hand collection of data and using databases. Data with regard to peer groups and managerial ownership guidelines are almost unavailable. There is only one source worldwide, Equilar database, that provides data concerning managerial ownership guidelines. Unfortunately the University of Amsterdam has no access to this
database. As a consequence the required data had to be hand collected.
Information regarding peer groups and managerial shareholder guidelines were extracted from proxy statements of the companies. These proxy statements were provided by dr. Tomislav Ladika and downloaded from the website of the U.S. Securities and Exchange Commission. The second research tool that is used are databases, namely Wharton Research Data Services (Risk Metrics & CRSP) and Thomson Reuters Datastream. These tools are chosen because it is simple in use and very efficient with regard of time.
3.3 Procedure
The first step was to explore which companies were included in S&P 500 between 2006 and 2012. This information was extracted from Wharton Research Data Services (WRDS).
20 In the period April – May 2013 data about peer groups was hand collected by using proxy statements of the S&P 500 companies. Thereafter was checked whether the S&P companies and peers have adopted a managerial shareholder guideline. This took place in the period May – June and was partially done through manually checking annual reports and financial statements for this information. In addition dr. Tomislav Ladika provided data concerning managerial shareholder guidelines from 2006 until 2008.
After data was gathered regarding the dependent and the independent variable in interest information was collected about the control variables in September 2013. This was done using Thomson Reuters Datastream. After this stage the dataset was prepared for statistical research. In addition, data concerning some control variables were gathered in February and March 2014.
21 3.4 Information concerning sample
In appendix 1 general information is given about the sample. This provides some
background information of the observations. Appendix 1 shows that a majority of the S&P
companies in the sample have no managerial shareholder guideline (63.46%). It also shows that a majority of the S&P companies have an ex-CEO on the board (54.93%).
In table 1 is shown that on average 25% of the peers have a managerial shareholder guideline. This is substantially lower in comparison to the proportions within S&P companies.
Another notable observation can be extracted from the percentiles. It shows that all the companies in the sample substantially differ across the percentiles.
3.5 Variables and measures
Through statistical analysis one seeks to analyze how one variable (the dependent variable) is related to other variables (the independent and control variables). In this part the
variables used for the empirical study are discussed. 3.5.1 Independent variables
In this study there are 32 types of independent variables, of which 21 are control variables. The composed control variables concern firm size, firm risk, firm performance and corporate
governance.
Tabel 1 General information
Research item mean std. dev. percentiles 10% 25% 50% 75% 90%
Fraction of peers with guideline 0.25 0.15 0.05 0.14 0.24 0.33 0.45
Fraction of peers that made a change 0.013 0.03 0 0 0 0 0.056
with regard to the guideline
Fraction of peers that adapted a guideline 0.007 0.022 0 0 0 0 0.03
Firm size (sales) 1.0e+07 2.0e+07 419515 1.0e+06 3.2e+06 9.4e+06 2.4e+07 Firm size (assets) 2.3e+07 6.7e+07 535988 1.4e+06 4.3e+06 1.5e+07 4.4e+07
Board meetings 10.86 2.28 8 9 11 12 14
Debt ratio 0.24 0.18 0.003 0.094 0.220 0.351 0.481
Stock price volatility 4.666 5.283 1.002 1.777 3.099 5.501 9.313
Return on equity 10.881 26.081 - 10.14 5.685 12.860 20.830 30.790
Operating income/Total assets 0.089 0.083 0.006 0.039 0.084 0.134 0.192
Corporate governance score 78.743 12.051 62.470 71.890 80.860 87.720 92.930
22 3.5.1.1 Main independent variables
The independent variable of most interest is the ‘fraction of peers that have adopted a managerial shareholder guideline’. This variable is determined for each S&P company in the sample by dividing the amount of peers with a managerial shareholder guideline by the total amount of peers.
Also an independent variable is included which measures the ‘fraction of peers in a group that made a change to their managerial shareholder guideline’. In other words, it is checked whether the peers in a group adopted or removed the guideline. This variable is determined for each S&P company in the sample by dividing the amount of peers that made a change with regard to the managerial shareholder guideline by the total amount of peers in the group. In addition a
variable is created which states what ‘fraction of the peer group adopted a managerial guideline’ during the sample period.
3.5.1.1 Lagged value
The control variable ‘lagged value’ refers to the past period values of the dependent variable. This variable is whether the S&P company in the sample adopted a managerial shareholder guideline. The lagged value of the dependent variable is included because it is expected that the current level of the dependent variable is heavily determined by its past level. This variable is not used in the model because of multicollinearity, see 4.1 results.
3.5.1.2 Firm size
The control variable ‘firm size’ is based on research of Himmelberg et al. (1999) and is measured through the variable sales.
3.5.1.3 Firm risk
The variable ‘firm risk’ is based on research of Demsetz & Lehn (1985) and Himmelberg et al. (1999) and is inferred through stock price volatility. The monthly stock prices of all the S&P companies in the sample were collected and with this data the yearly variances were calculated. 3.5.1.4 Firm performance
The variable ‘firm performance’ refers to research of Core & Larcker (2002) and is inferred by return of equity of the S&P company in the sample. Their research proposes that governance
23 problems can be revealed through firm performance. As a result, poorly performing firms are more likely to adopt ownership guidelines to provide managers stronger incentives to work hard. In addition the firm performance of the S&P company in the sample is stated as the stock
return. This variable is calculated by dividing the appreciation in the price and any dividends paid by the original price of the stock.
3.5.1.5 Corporate governance
A managerial ownership guideline is a tool align the interests of managers and investors. The adoption of such a guideline depends thus on the corporate governance of the company (Shleifer and Vishny, 1997). Through four variables the quality of the corporate governance of the S&P company is inferred, namely through board meetings, ex-CEO in the board and two indexes.
In addition a proxy for the G-index (Gompers et al., 2003) and the E-index (Bebchuk et al., 2009) is calculated by using data from Risk Metrics. Because of unavailability of all the provisions a proxy is made for the G-index. The proxy for the G-index contains the following provisions: blank check preferred, limit ability to amend bylaws, limit ability to amend charter, classified board, cumulative voting, fair-price, golden parachutes, poison pills, limit ability to call special meeting, supermajority and limit ability to act by written consent.
The E-index I matches with the research of Bebchuk et al. (2009) and is based on the following provisions: limit ability to amend bylaws, limit ability to amend charter, classified board,
supermajority, golden parachutes and poison pills. 3.5.1.6 Cash flow
Based on research of Palia (1998) and Himmelberg et al. (1999) the ratio operating income divided by total assets is included as an independent variable to infer the free cash flow of the S&P company in the sample.
3.5.1.7 Debt
Debt is measured by Benson and Lian by the debt ratio. This is the book value of debt divided by the total assets of the concerning firm. This method is also used in this study.
24 3.5.2 Dependent variables
The dependent variable is described as what is measured in an experiment and what is affected during the experiment (Keller, 2009). It other words, the dependent variable responds to the independent variables.
In this study the first dependent variable is whether the S&P firm has adopted a managerial shareholder guideline. To infer this variable data was hand collected from proxy statements for all the S&P companies in the sample.
The second independent variable is whether the S&P company made a change with regard to the managerial shareholder guideline during the sample period. This data was calculated through the first independent variable. In addition a third variable was created that states whether a S&P company adopted a managerial guideline during the sample period.
25 3.6 Model
The above information concerning independent and dependent variables is translated into a multiple regression model (table 2).
In this model the independent variable and the different categories of control variables are included. Appendix 2 shows in detail which variables are included in this empirical study. Based on the gathered data a multiple regression analysis is performed. The statistics are used to assess the model’s fit and the validity of the model.
To construct a view of the variables, included in the regression model, the scale of
measure of each one is discussed. The control variables are quantitative variables, this means that they have numerical values.
The first independent variable ‘fraction of peers that adopted a managerial shareholder
guideline’ is measured through a ratio that will be between 0 and 1. The degree of adoption by peers is positively related with the amount between 0 and 1. This means that when all peers adopted a guideline the independent variable would end up being 1. It can be stated that there is a positive relation between the fraction of peers with a guideline and whether a S&P company in the sample will adopt the guideline. Hence, it is expected that the coefficient of fraction of peers with guideline is a positive number.
The independent variables ‘fraction of peers that made a change to the guideline’ and ‘fraction of peers that adopted a guideline’ are also measured as a ratio between 0 and 1. This means that as more peers in a group make a change with regard to the guideline, the ratio will become higher. It can be stated that these variables have a positive relation with whether a S&P
Table 2 Model
Dependent variable = S&P_comp
Independent variable = peer_group
Firm size = f_size
Debt = debt
Firm risk = f_risk
Firm performance = f_perfor
Free cash flow = cash_flow
Corporate governance = cov_gov
26 company has a guideline. Hence, it is expected that the coefficients of these variables are positive.
The lagged value of the adoption of a guideline by the S&P firm is included as a dummy variable. This independent variable can be 0 or 1 and takes on the value 1 if a firm had a managerial shareholder guideline in the previous year. It is expected that this variable will be a positive number. However, this variable is not included in the model because of multicollinearity, see 4.1 results.
The control variable sales, that infer firm size, is a continuous. A logarithm is added to this variable due to interpretation reasons. With the study of Himmelberg et al. (1999) it can be stated that sales are positively related to the likelihood to adopt a managerial shareholder guideline. Hence, it is expected that this variable will be a positive number.
The independent variable stock price volatility is a continuous variable. Demsetz and Lehn (1985) points out that there is a relation between the riskiness of the firm (stock price volatility) and managerial ownership. A higher amount of managerial ownership stakes is necessary in risky firms, because of the scope for moral hazard is greater. Based on this information it can be stated that firm risk is positively related to the likelihood to adopt a managerial shareholder guideline. Hence, it is expected that this variable will be a positive number.
The independent variable return on equity is a ratio. A study of Cao & Gu (2011) state that poorly performing firms are more likely to adopt ownership guidelines to provide managers stronger incentives to work hard. This means that return on equity is negatively related to the likelihood of a company to adopt a managerial shareholder. It is thus expected that this variable will be a negative number. The performance of the firm is also inferred by the stock return of the S&P company. Seen the research of Cao & Gu (2011), that poorly performing firms are more likely to adopt ownership guidelines to provide managers stronger incentives to work hard. Also is expected that stock return is negatively related to the likelihood of a company to adopt a managerial shareholder.
The control variables board meetings and ex-CEO in the board are used to infer the quality of corporate governance. A lower quality of corporate governance suggests that there is less likelihood that a managerial shareholder guideline has been adopted (Cao & Gu, 2011). It is thus expected that there is a positive relation between the quality of corporate governance and whether the company adopted a guideline.
27 In addition the variables proxy for G-index and E-index are used to control for corporate
governance. As the indexes increases the rights of shareholders decreases. This means that a high amount of the indexes refer to a bad quality of corporate governance. This means that the indexes are negatively related to whether the S&P company adopted a guideline.
The effect of debt ratio is however ambiguous according to Cao & Gu (2011) and can go each direction.
The independent variable operating income divided by total assets is a ratio. Palia (1998) and Himmelberg et al. (1999) state that there is a positive association between operating income (as a proxy for free cash flow) and managerial ownership. Thus it is expected that there will be a positive relation between the concerning variable and the likelihood to adopt a guideline. The first dependent variable whether the S&P firm in the sample adopted a guideline is a dummy variable. Whether a S&P firm in the sample has a guideline is measured on the basis of a 0 or 1. The degree of the concerning variable is positively related to the likelihood of adoption. This means that a firm with a managerial shareholder guideline has the value 1.
The second variable’ if the S&P firm made a change with regard to the guideline’ is also a
dummy variable. This means that a firm that made a change to her guideline has the value 1 for that year in the sample period. The same interpretation counts for the third variable that
28
Part 4: results
This part contains the results of the statistical analyses with Stata. First a general analysis is made to show how the results are processed.
Then the correlations between the variables used in this study are viewed and a regression analysis is made. In the discussion (part 5) of this thesis the results are also discussed. 4.1 General analysis
Through a statistics programs named Stata the data of this research is analyzed.
The data, that is collected by hand or exported from a database, is then further processed in Stata to create the variables in need. For example, the fraction of peers with a managerial guideline is calculated by dividing the peers with a concerning guideline by the total amount of peers. Also variables are created to check whether the S&P company or peer made an
adoption/change with regard to the managerial guideline across time.
After the database is complemented with the dependent, independent and control variables the data is checked for outliers. These are unusually large or small observations and its validity is therefore suspect (Keller, 2009).
In the following part descriptive statistics are used to describe the features of the main
variables. An analysis is performed at different levels to examine the S&P companies with regard to managerial shareholder guidelines and how they behave. Table 3, 4 and 5 focus on industry, corporate governance and control variables.
In table 3 the S&P companies that are included in the sample of the study are categorized by ten industries. In the table a distinction is made between S&P companies with or without a managerial shareholder guideline. The information of the previous chapter is in line with this table, namely that the majority of the S&P companies in the sample don’t have the concerning guideline. The proportions however depend on the industry as can be seen in table 3. Industries like Electric Utilities and Aerospace & Defense show for example a majority of companies with a managerial shareholder guideline.
29 In the following table is shown whether a certain clustering or movement can be found at different qualities of corporate governance. This is done based on a proxy for the G-index and the E-index. A low level refers to a good level of corporate governance and a high level refers to the opposite.
At the proxy for the G-index no movement can be found with regard to S&P companies with or without a managerial shareholder guideline. The ratios are proportional across the different levels of corporate governance.
However, at the E-index a certain movement is observed. Namely, as the quality of corporate governance reduces the proportion of S&P companies with a managerial shareholder guideline becomes less. In other words, S&P companies in this sample with a high level of corporate governance have proportionally more a managerial shareholder guideline.
In Appendix 3 and 4 a general analysis is made for the variables concerning changes and adoptions of guidelines on the basis of corporate governance.
In appendix 3 a positive movement is found between the proxy G-index and the variable that concerns S&P companies that made a change to the guideline. It namely shows that the group
Tabel 3 Descriptive statistics
Industry S&P firms with guideline S&P firms with no guideline Percentage with guideline Percentage of total sample
Regional Banks 58 78 42,65% 3,35%
Semiconductors 51 81 38,64% 3,25%
Communications Equipment 21 76 21,65% 2,39%
Health Care Equipment 50 56 47,17% 2,61%
Oil & Gas Exploration 33 57 36,67% 2,22%
Industrial Machinery 42 43 49,41% 2,09%
Apparel Retail 30 52 36,59% 2,02%
Pharmaceuticals 31 50 38,27% 2,00%
Aerospace & Defense 43 35 55,13% 1,92%
Electric Utilities 45 28 61,64% 1,80%
Remaining industries 931 2167 30,05% 76,34%
Tabel 4 Descriptive statistics
Corporate governance S&P firms with guideline S&P firms with no guideline Percentage with guideline Percentage of total sample Proxy G-index Low (1 - 4) 309 517 37,41% 24,00% Intermediate (5 - 8) 923 1529 37,64% 71,24% High (9-11) 62 102 37,80% 4,76% E-index Low (0 - 1) 119 127 48,37% 9,22% Intermediate (2-4) 700 1142 38,00% 69,01% High (5 - 6) 212 369 36,49% 21,77%
30 S&P companies with a less quality of corporate governance relatively make more changes to their guideline.
However, for the E-index the opposite direction is observed. Namely, as the quality of corporate governance reduces the proportion of S&P companies that made a change to their managerial shareholder guideline becomes less.
This analysis also corresponds with appendix 4. The reason for these similarities is that the data of the concerning tables each other largely overlaps.
In table 5 a categorization of S&P companies with or without a managerial shareholder guideline is made based on a number of control variables in this research.
The control variables stock price volatility and operating income/total assets don’t show a certain movement. However, stock return and debt ratio show a positive trend. As the
concerning variables increase the proportion of S&P companies with a managerial shareholder also increase.
Another striking observation is the movement between firm size and managerial shareholder guidelines. Table 5 clearly shows that as the firm size increases the proportion of S&P companies with a managerial shareholder guideline also increases.
31 In appendix 5 a similar categorization is made based on S&P companies that did or didn’t made a change to the managerial shareholder guideline.
The control variable stock price volatility doesn’t show a certain movement. However, stock return, firm size and operating income/total assets show a negative trend. There are relatively less S&P companies that made a change to their managerial shareholder as the concerning variables increase. On the other hand debt ratio shows a positive movement with the concerning dependent variable.
This analysis also corresponds with appendix 6. The reason for these similarities is that the data of the concerning tables each other largely overlaps.
In the following part a correlation analysis is made of the variables that are included in this research. This analysis gives a view of whether and how strongly pairs of variables are related (Keller, 2009).
First the dependent variable S&P company with guideline (<1>) is examined. Appendix 7 shows
Tabel 5 Descriptive statistics
Control variables S&P firms with guideline S&P firms with no guideline Percentage with guideline Percentage of total sample Stock price volatility
Low (< 1,77) 362 635 36,31% 24,76% Intermediate ( 1,77 - 5,501) 728 1298 35,93% 50,32% High (>5,501) 386 617 38,48% 24,91% Stock return Low (<-0,266) 339 630 34,98% 24,29% Intermediate (-0,266 - -0,167) 111 187 37,25% 7,47% High (>-0,167) 1020 1702 37,47% 68,24% Debt ratio Low (<-0,094) 347 668 34,19% 25,21% Intermediate (0,094 - 0,351) 747 1270 37,04% 50,10% High (> 0,351) 383 611 38,53% 24,69%
Operating income/total assets
Low (<-0,039) 390 626 38,39% 25,24%
Intermediate (0,039 - 0,134) 713 1287 35,65% 49,68%
High (> 0,134) 374 636 37,03% 25,09%
Log sales (firm size)
Low (<-13,84) 328 684 32,41% 25,14%
Intermediate (13,84 - 16,06) 723 1273 36,22% 49,58%
High (> 16,06) 426 592 41,85% 25,29%
Log total assets (firm size)
Low (<-14,14) 328 680 32,54% 25,04%
Intermediate (14,14 - 16,51) 682 1324 34,00% 49,83%
32 that there is a positive and significant relation with the fraction of peers with a guideline (<4>). Other variables that have a significant and positive relation with this dependent variable are firm size, stock price volatility, return on equity, debt ratio and board meetings. Another notable observation is the negative relation between the concerning dependent variable (<1>) and the E-index. This means that there is a positive relation between higher quality of corporate
governance and S&P companies with a managerial shareholder guideline.
The second dependent variable, S&P companies that made a change with regard to the
guideline (<2>), has less significant relations. A striking observation is that the direction of the relationship with most control variables has changed in comparison to the first dependent
variable (<1>). There is a significant negative relation with lagged value guideline S&P company (<7>), firm size sales (<8>), return on equity (<11>) and operating income/total assets
(<15>). The relationships that this dependent variables has correspond with those of the dependent variable S&P companies that made an adoption with regard to a managerial shareholder guideline (<3>).
Appendix 7 is also used to check for multicollinearity. This means that two or more of the variables are highly correlated. A consequence of multicollinearity is that the coefficients in a regression of at least one individual regressor will be imprecisely estimated (Keller, 2009). Multicollinearity is present when the correlation values approach 1. From table 6 is derived that there is a high correlation between S&P company with a guideline (<1>) and lagged value guideline S&P company (<7>). For this reason the variable lagged value will not be used in the regression model. There is also a high correlation between variables that are relate to a change or adoption. Also sales (<8>) and total assets (<9>) show a high correlation. To avoid
multicollinearity the variables that have a high correlation should not be included in the same regression.
33 4.2 Regression analysis
In this part a regression analysis is performed. This type of analysis is used when the focus is on the relationship between a dependent variable and one or more independent variables. It helps to understand in which way the value of the dependent variable changes when one of the independent variables is altered, while the remaining independent variables are held fixed (Keller, 2009).
The first regression describes the relation between S&P companies with or without a
managerial shareholder guideline (dependent variable) and the fraction of their peers that have the concerning guideline (independent variables). This regression is complemented with control variables regarding firm size, debt, firm risk, return, firm growth and corporate governance. There will be looked at what effect the independent and control variables have on the dependent variable.
Table 6 shows that the fraction of the peer group with a managerial shareholder guideline is positively and significantly (independent variable) related to S&P companies with a guideline.
Tabel 6 Regression analysis
Variable S&P companies with guideline Std. Err.
β
Fraction of peer group with guideline 0.1948** 0.0926
Log sales (firm size) 0.0112 0.0117
Debt ratio 0.089 0.0731
Stock price volatility 0.0063*** 0.0023
Return on equity 0.0007 0.0006
Ex-CEO in Board (0.0031) 0.0239
Operating income/total assets (0.1836) 0.1953
Board meetings 0.0055 0.0059 Stock return 0.0045 0.0264 E-index (0.0129) 0.0087 LR chi2(10) = 28.35 Prob > chi2 = 0.0016 Pseudo-R2 = 0.0119 * ρ ≤ 0.10 ** ρ ≤ 0.05 *** ρ ≤ 0.01
34 This observation is in line with the literature.
Another notable observation is the positive and significant relation between stock price volatility and S&P companies with a guideline. The model thus states that more volatile stock prices (i.e. more firm risk) have a positive effect on S&P companies that have a managerial shareholder guideline.
Table 6 also shows that most of the control variables are not significant.
As expected the regressors Ex-CEO in board & E-index show a negative relation with the dependent variable.
The Likelihood Ratio Chi-Square test (LR chi2) is used to test whether at least one of the
coefficients is not equal to zero. The probability of the LR chi2 is used to test the null
hypothesis; it states that all the coefficients in the model are equal to zero (Keller, 2009). The small p-value (0.0016) concludes that at least one of the coefficients in the model is not equal to zero.
The Pseudo-R² is used because it concerns a probit regression. This ratio could range
between 0 to 1. It measures the fit of the model by using the likelihood function. It measures the quality of fit of a probit model by comparing values of the maximized likelihood function with all the regressors to the value of the likelihood with none (Stock & Watson, 2012). In this model the Pseudo-R² is 0.0119.
In addition, appendix 8 has been added that corresponds to table 6 but has a proxy for
the G-index included to control for corporate governance (instead of the E-index). The
movement of all the coefficients is similar to the model in table 6. The Pseudo-R² is however a bit lower, indicating a relatively lesser quality fit of the probit model.
In the following part the relation is examined between S&P companies that made a change with regard to their guideline and the fraction of their peer group that made a change to the
concerning guideline. This model (table 7) contains the same control variables as the previous model.
A striking observation is that the main independent variable has a negative, but not
significant, relation with the dependent variable. It thus suggests that S&P companies will make less changes to their guideline if the fraction of peers that make changes increases. This
movement is further examined in the discussion.
35 suggests that the control variables have a different relation with the two dependent variables. Only the regressors firm size, debt ratio and E-index have a positive relation with both
dependent variables.
The high p-value of the chi2 (0.2496) suggests that the unexplained variation in this
model is large. This leads us to conclude that at least one of the coefficients is equal to zero.
In addition, appendix 9 has been added that corresponds to table 7 but has a proxy for
the G-index included to control for corporate governance (instead of the E-index). The
movement of all the coefficients is similar to the model in table 7. The Pseudo-R² is however a bit lower, indicating a relatively lesser quality fit of the probit model.
In table 8 the relation is examined between S&P companies that adopted a managerial shareholder guideline and the fraction of the peer group that also adopted the concerning guideline. The model is complemented with the same control variables that are used in the previous regressions.
Tabel 7 Regression analysis
Variable S&P companies that made a change Std. Err.
with regard to the guideline β
Fraction of peer group that made a change (0.0532) 0.0973
with regard to the guideline
Log sales (firm size) 0.0009 0.0029
Debt ratio 0.0214 0.0204
Stock price volatility 0.0005 0.0005
Return on equity (0.0003)* 0.0001
Ex-CEO in Board 0.0149** 0.0067
Operating income/total assets 0.0284 0.0533
Board meetings (0.0013) 0.0017 Stock return (0.0011) 0.0074 E-index (0.0031) 0.0024 LR chi2(10) = 12.56 Prob > chi2 = 0.2496 Pseudo-R2 = 0.0362 * ρ ≤ 0.10 ** ρ ≤ 0.05 *** ρ ≤ 0.01
36
As expected the independent variables in table 8 show similar movements as the
variables in table 7. This is because the main independent variables of table 7 and 8 have a overlap with regard to data. The fraction that made a change to the guideline namely consists partly of the fraction that made an adoption. This creates the similarity in coefficients.
In table 8 however the main independent variable, fraction that made an adoption to the
guideline, shows a significant and negative relation with S&P companies that made an adoption. This coefficient implies that as the fraction of peers that adopt a guideline increases the S&P companies will make less adoptions. Thus more adoptions by the peers have a negative relation on the adoptions of S&P companies.
The p-value of the chi2 is 0.0565. At an alpha of 0.10 it can be concluded that at least
one of the coefficients in the model is not equal to zero.
In addition, appendix 10 has been added that corresponds to table 8 but has a proxy for
the G-index included to control for corporate governance (instead of the E-index). The
movement of all the coefficients is similar to the model in table 8. The Pseudo-R² is however a bit lower, indicating a relatively lesser quality fit of the probit model.
Tabel 8 Regression analysis
Variable S&P companies that made an adaption Std. Err.
with regard to the guideline β
Fraction of peer group that made an adaption (0.3162)* 0.1676
with regard to the guideline
Log sales (firm size) 0.0028 0.0024
Debt ratio 0.0156 0.0173
Stock price volatility 0.0007 0.0004
Return on equity (0.0003)** 0.0001
Ex-CEO in Board 0.0105* 0.0056
Operating income/total assets 0.0468 0.0432
Board meetings (0.0004) 0.0014 Stock return (0.001) 0.0061 E-index (0.002) 0.002 LR chi2(10) = 17.91 Prob > chi2 = 0.0565 Pseudo-R2 = 0.0597 * ρ ≤ 0.10 ** ρ ≤ 0.05 *** ρ ≤ 0.01
37 4.3 Testing of hypotheses
In the following section the formulated hypotheses are tested. The meaning of the conclusion will briefly be discussed.
Hypothesis 1
This hypothesis is as follows; a firm is more likely to use a managerial ownership guideline if
firms in its peer group also use it. To test this statement a null hypothesis is
formulated that suggests there is no relation between a firm using a managerial ownership guideline and her peer group using it also. The regression in table 6 shows that the null hypothesis is rejected because the p-value is less than 0.05 and the β is 0.1948.
Hypothesis 2
This hypothesis is as follows; a firm is more likely to make a change to a managerial ownership
guideline if firms in its peer group also make it. To test this statement a null hypothesis is formulated that suggests there is no relation between a firm making a change to a managerial ownership guideline and its peer group also making it. The regression in table 7 shows that the null hypothesis is not rejected because the p-value is higher than 0.05. The β fraction of peer group making a change is -0.0532. This means that there is a non-significant negative relation between the company making a change concerning a managerial ownership guideline and the peer group making changes.
Hypothesis 3
This hypothesis is as follows; a firm is more likely to adopt a managerial ownership guideline if
firms in its peer group also make it. To test this statement a null hypothesis is formulated that suggests there is no relation between a firm adopting a managerial ownership guideline and its peer group also making an adoption. The regression in table 8 shows that the null hypothesis is rejected at an alpha of 0.10. The main coefficient has namely a p-value less than 0.10.
The β fraction of peer group that made an adoption is -0.3162. This means that there is a significant negative relation between a company adopting a managerial ownership guideline and the peer group making adoptions.
38
Part 5: discussion
In this part the results are further analyzed and explained. Subsequently omitted variables, limitations of literature review and limitations of empirical research are discussed. Finally the suggestions for further research are treated.
5.1 Analysis results
Based on the regressions in part 4 it can be stated that the fit of the models is not that well. The Pseudo-R² is used to measure the fit of the models, because it concerns probit models. By comparing the Pseudo-R² of the probit models it can be stated that they range between 0.0119 and 0.0597. Seen that the models address different main variables the Pseudo-R² values cannot be compared to make a statement about the relative quality of the models.
The first regression to be examined is the model in table 6. The concerning table shows
that there is a significant and positive relation between the fraction of peers that has a
managerial ownership guideline and a S&P company having it. The β of the main independent variable is 0.1948. Accordingly, an increase in the fraction of peer group with guideline of 0.1 is estimated to increase the probability of a S&P company using a guideline by 1.95 percentage points. The movement of this independent variable is in line with the existing literature.
Granovetter (1985) namely states that peers and benchmarks have a significant influence on a company’s decision, in specific on corporate governance.
The regressors of Log sales, stock price volatility, ex-CEO in board, operating income/total assets and E-index also correspond with the existing literature. Literature namely expects that log sales, stock price volatility and operating income/total assets have a positive relation with the probability of a S&P company using a managerial ownership guideline. The coefficient of stock price volatility is even significant and has a magnitude of 0.0063. This means that a one unit change in stock price volatility generates a 0.63 percentage points change in the probability that a S&P company has a managerial ownership guideline. As corresponds with literature, the E-index has a negative relation with the dependent variable. As corporate governance becomes better the probability that a S&P company has a guideline becomes less.
The literature review showed that the movement of debt ratio, return on equity and stock return is ambiguous. The results of the model are not significant but show a positive relation between return on equity & stock return and the dependent variable.
