Corporate governance and firm value in the Netherlands,
a study on the effects of the actualization of the Tabaksblat code
L. Vermeij 6301991
Thesis supervisor: Ilko Naaborg
Bachelor Thesis Finance and Organization
Business and Economics, University of Amsterdam August 2013
Index
1. Introduction 3
2. Literature review 4
2.1 Corporate Governance 4
2.2 Previous research on corporate governance in relation to firm value 5
2.3 Code Tabaksblat 6
2.3.1 Structure of the code 7
2.3.2 Enforcement of the code 8
2.3.3 Actualization of the Code 9
2.4 Hypothesis 10
3. Methodology and data 11
3.1 Methodology 11
3.1.1 Corporate Governance measure 12
3.1.2 Firm valuation 13 3.1.3 Control variables 13 3.1.4 Sample 14 3.2 Data 14 3.2.1 Tobin’s Q 14 3.2.2 Compliance 15 3.2.3 Control variables 16 3.2.4 Correlations 17 4. Empirical results 18 4.1 Univariate analysis 18 4.2 Regression analysis 19
4.3 Robustness check and additional tests 21
5. Conclusion and discussion 22
Reference list 24
Appendix 1 - Variation description 26
Appendix 2 - Pearson’s correlation matrix 27
Appendix 3 – Results of additional tests 27
1. Introduction
January 2004, the Dutch code Tabaksblat was introduced. This code was intended to create more transparency in terms of financial reporting for shareholders, thereby limiting the possibility of fraudulent behavior and increasing the protection of shareholder rights
(Groenewald, 2005). After the onset of the financial crisis, a revision of the code was deemed necessary, as the implemented Tabaksblat code seemed to have become outdated. In 2009 the code was renewed to clarify rules that might have been unclear before. Besides clearing up existing best practices, new best practices were added to the existing code (Monitoring Committee, 2013). The clarifications and additions should have led to a better code, which is easier to comply with due to the clear rules. Among others, de Jong et al. (2005) state that effective corporate governance enhances firm value and is therefore good for the economy as a whole.
In the past, literature on the relationship between corporate governance and firm value has focused on a couple of large provisions1 that drive the relationship with firm value. Bebchuck et al. (2009) show that only some individual provisions of corporate governance codes would have a positive influence on firm value, while others show to be irrelevant or even to have negative effects. Relating this to the Netherlands, de Jong et al. (2005) provide a research that investigates some provisions of the Tabaksblat code. They too, find that only some provisions seem to drive firm value. These examples show that previous literature is inconclusive in its findings if a complete set of provisions as a whole might enhance firm value.
This study will focus on the relationship between firm value and compliance with the Tabaksblat code. It is interesting to see if this relationship has changed after the adaptation of the code in 2008. This way, companies and shareholders will be able to see the effect of the code on firm value so that investors can choose their investments more wisely, and companies can see the importance of good corporate governance. By highlighting the change in
effectiveness of compliance before and after the adaptation of the code, policy makers will be able to see if the changes that were put forward are proving to be effective. The question that arises is the following:
1 Provisions refer to corporate governance measures
To what extent did the actualization of the Dutch corporate governance code Tabaksblat in 2009 influence the relationship between compliance and firm value in Dutch listed
companies?
If a change in the relationship appears to have occurred, it is interesting to know if the renewal of the code has led to an increase in compliance with the code, as well as if firm value has changed after the adaptation. These questions will also be addressed in this thesis.
The organization of the remainder of the paper is as follows: Section two will focus on the literature on corporate governance and its relation to firm value. It will also describe the Tabaksblat code and its different chapters. A hypothesis will be developed from the presented literature. Section three provides both the research methodology and information about the data and companies. Section four shows the results of the univariate as well as the
multivariate regressions that have been done. Section five gives the conclusion, limitations of this research and suggestions for future research.
2. Literature review
The next section will give an overview of the relevant previous literature in the field of corporate governance and its relation to firm value. Moreover, a concise overview of the Tabaksblat code will be presented.
2.1 Corporate Governance
Corporate governance is a relatively new concept, which has only been used since the introduction of the agency theory by Jensen and Meckling in 1976, when they spurred the discussion on loss in efficiency due to the separation of ownership and control (Denis, 2001). The problem of separation between ownership and control is that shareholders are investing money into a company, but are at the same time the residual claimants of the benefits. A manager on the other hand can engage in behavior that might be harmful to the company as a whole, but might be beneficial to the manager himself (Denis, 2001). In her survey, Denis gathers the past 25 years of corporate governance research, and thereby comes up with an all-encompassing definition of corporate governance: “The set of institutional and market mechanisms that induce self-interested managers (the controllers) to maximize the value of the residual cash flows of the firm on behalf of its shareholders (the owners).”
Information asymmetry is one of the main reasons why it is possible for the manager to engage in behavior that is not in the best interest of the shareholder. Because the shareholder is unable to continuously observe the manager, he may engage in actions such as empire
building, shirking or excessive consumption (Denis, 2001). To control this unwanted behavior by the manager, shareholders try to monitor the managers’ actions, which makes them incur monitoring costs. To avoid losing their investors, managers undertake actions to show what they are doing with the investors’ money, and that they are investing it wisely. By doing so, they incur bonding costs, for example by providing annual reports and creating transparency within the company. Jensen and Meckling (1976) continue to explain that both of these actions result in a residual loss for the shareholders. The monitoring costs, bonding costs and the residual loss add up to total agency costs, which amounts the loss in efficiency.
Corporate governance codes are intended to lead to more transparency and increase company performance to the benefit of shareholders (Akkermans et al., 2007). Many researches have found positive correlations between corporate governance codes and firm value, hypothesizing that this is the result of transparency. The transparency would create a better environment for internal and external supervision. It would also lead to a decrease in agency costs caused by information asymmetry, since it is now easier for the shareholder to monitor the manager. According to Nordberg (2011) the managers are incentivized to make better decisions regarding efficient allocation of resources when they are monitored more closely.
2.2 Previous research on corporate governance in relation to firm value
Compliance with corporate governance codes in relation to firm value has been studied extensively. The main reason for the great body of literature on a wide range of countries is that it is hard to compare the results cross-border, since there isn’t one single measure of corporate governance performance (Gompers et al., 2003). Many studies have focused their research on the influence of several separate best practices in relation to firm value, or they have created their own set of corporate governance variables.
The results of previous researches have been mixed or have created unexpected outcomes with regard to the relation of corporate governance and firm value. However, the positive effects in most studies outweigh the negative (Ammann et al. 2011). Bebchuck et al. (2009) created a set of 24 entrenchment provisions, which are tested on having an influence on firm value. These entrenchment provisions are provisions that limit shareholder rights or increase the arrangements to prevent hostile takeovers. This set of provisions has been widely used and discussed in other studies, and is therefore relevant in the research on corporate governance and firm value. Only six of the provisions negatively influenced firm value, while the other eighteen had no correlation, or unexpectedly had a positive influence. The six
provisions that negatively influenced firm value were: 1) staggered boards; 2) limits to shareholder amendments of the by-laws; 3) supermajority requirements for mergers; 4) supermajority requirements for charter amendments; 5) poison pills and 6) golden parachute arrangements. Bebchuck et al. state that limiting these entrenchments by for example
corporate governance code should have a positive effect on firm value. Lemmon and Lins (2003), come to the same conclusion in their study on the East Asian financial crisis. They find that companies that did not have sufficient corporate governance control on average had 12 percent lower stock returns during the financial crisis.
Bauer et al. (2004) also use stock returns as well as accounting measures such as Tobin’s Q to analyze the relationship between governance standards and firm performance. The analysis is done on companies in EMU countries as well as companies from the UK. The results differ between these two regions, as the Eurozone countries show a significant positive relationship between corporate governance variables and firm value, while the English firms do not. An explanation is could be that most Eurozone countries tend to have corporate governance standards that are lower than those in the UK. Results show that lower corporate governance standards correlate with a stronger relationship between corporate governance and firm value. This means that if corporate governance standards tend to be high already, adding extra corporate governance regulations will on average not necessarily further increase the value of the firm.
The great body of literature is not conclusive in its findings. Edwards and Clough (2005) find that structural arrangements, such as staggered boards, have a positive influence on firm value. Bebcuck et al. (2009) have opposing results in which, as named before, staggered boards is one of the provisions that negatively influence firm value. These conflicting results continue to show the importance of necessary further research, since the combination of several of these provisions together might be the key to a positive influence on firm value.
2.3 Code Tabaksblat
In January 2004 the Dutch code Tabaksblat was introduced, thereby replacing the old Peters Committee Code from 1997. The Peters Committee Code was thought to be hard to comply with due to its complexity and therefore not being effective in reaching its objectives to create a return as high as possible for the shareholders. Code Tabaksblat was intended to create more transparency in terms of financial reporting (Monitoring Committee, 2013). A sharp focus was put on the supervisory board and its role to supervise the executive board, thereby
making the controlling function stronger and creating more protection for the shareholders. All companies that are listed on the Dutch stock exchange and have their statutory seat in the Netherlands are obliged to apply the Tabaksblat code.
2.3.1 Structure of the code
The code consists of a preamble and the following five sections: I. compliance with, and enforcement of the code; II. the management board; III. the supervisory board; IV. the shareholders and the general meeting of shareholders; V. audit of the financial reporting and the position of the internal auditor function and of the external auditor. These five chapters contain 21 principles in total. The principles in their turn have been divided into 113 best practice provisions that contain the regulations of the code. The chapters of the code are explained in further detail below.
2.3.1.1 Compliance with, and enforcement of the code
This first chapter has two best practices. The first one creates the understanding that the supervisory board is responsible for reporting the corporate governance policy of the
company. The second best practice states that the corporate governance policy has to be in a separate chapter of the annual report.
2.3.1.2 The management board
This chapter contains the principle that has been complied with the least: the level and composition of board remuneration. The report of the Monitoring Committee states that best practice II.2.7 (II.2.8 in the new code)2, which is the best practice on maximum remuneration in the event of dismissal, has been least complied by AEX index companies. This is the result of board members that have been seated since before the introduction of the Tabaksblat code in 2004, for which old contracts have been respected (Monitoring Committee, 2013). Other principles that are incorporated in this chapter are on the subject of board appointment, responsibilities and conflicts of interest.
2.3.1.3 The supervisory board
Within the Tabaksblat code, there is a strong focus on the supervisory board. There are many directions on how the supervisory board should act within a company, usually incorporating a
2 “The remuneration in the event of dismissal may not exceed one year’s salary (the ‘fixed’ remuneration
component). If the maximum of one year’s salary would be manifestly unreasonable for a management board member who is dismissed during his first term of office, such board member shall be eligible for severance pay not exceeding twice the annual salary.” (Code Tabaksblat, 2009)
two-tier structure. A large portion of the chapter concentrates on the role of the supervisory board to monitor the board of management. Besides this, there are best practices on the independence of the members, what information about the supervisory board should be disclosed in the annual report and how it should be composed. Every supervisory board has to have a board composition that consists of members from different fields of knowledge, which they can apply in the different committees that have to be present.
2.3.1.4 The shareholders and the general meeting of shareholders
The chapter on shareholders is the one that had the most best practices added during the actualization of the code. In 2004 this chapter started with 27 best practices, while after the adaptation it consisted of 35. This is mainly due to the division of some of the best practices into more concise ones. A new best practice in this chapter is on response time; this means that there can be no more than 180 days between the notification of shareholders to introduce a new agenda point and the general meeting3. Other items in this chapter consist of reiterating
the power of shareholders by involving them in the general shareholders meeting, the prevention of anti-takeover measures by depositary receipts and the publication of
information for all shareholders. This means that information has to be shared without delay, even if it contains price sensitive information4.
2.3.1.5 Audit of the financial reporting and the position of the internal auditor function and of the external auditor
Rules on the responsibilities of the board, with respect to the financial reports, are covered in this final chapter. Principle V.1 states that the management board is responsible for the quality and completeness of the disclosed financials, while the supervisory board should ensure that the management board fulfills this responsibility. Further best practices are on the external auditor, which should be independent and appointed in the general meeting, as well as the internal auditor, for whom the management board is also responsible.
2.3.2 Enforcement of the code
The Dutch Tabaksblat code is a comply-or-explain based code. This means that companies are not required by law to comply with every best practice of the code, but if they do not, they are required to explain in the annual report why they have not complied with the provision. This creates a system of voluntary compliance and monitoring by the supervisory board, but
3 Code Tabaksblat (2009), best practice IV.4.4 4 Code Tabaksblat (2009), principle IV.3
without legal enforcement after deviations. Even though there are no legal consequences if a company does not comply, research has shown that companies will comply due to self-regulation (de Jong et al., 2005; Bauer et al., 2004; Shleifer and Vishny, 1997). Bauer et al. give two reasons for this self-regulatory effect. Firstly, they claim that well-governed
companies on average have better credit ratings, which results in a lower cost of capital from institutional investors. Secondly, the monitoring cost of the investor is lower, which is the reason why they require a lower rate of return, which again results in a lower cost of capital for the firm.
According to de Jong et al. (2005) the self-regulatory effect alone is not enough to have companies comply with a full extent. There still has to be some sort of legal enforcement that accompanies the voluntary code. The previous Dutch code drawn by the Peters
Committee (1997-2003) had no legal enforcement at all; de Jong et al. (2005) show that this code had little to no effect on company value. As company value by Tobin’s Q can be seen as a proxy to measure the quality of corporate governance measures in a company, the old Tabaksblat code was not deemed very successful in reaching a higher corporate governance standard (Denis, 2001). The Tabaksblat code was therefore supplemented with some legal obligations and constraints. Examples of the connections with the law are lined out in the Preamble of the Code Tabaksblat (2008). These include the open standards within the Dutch Civil code that embrace the reasonableness and fairness principles, which tie the corporate governance principles to corporate law by making it possible to suppress bad governance practices and thereby encouraging integrity.
2.3.3 Actualization of the Code
At the request of CNV, Eumedion, FNV, NCD, NYSE Euronext, VEB, VEUO and VNO-NCW, the Tabaksblat code has been updated on December 10th, 2008 (Code Tabaksblat, 2009). The request intended to achieve a mentality change within companies from ‘box ticking’ behavior, in which there is just formal compliance, to actual compliance with the spirit of the code. This mentality change could be achieved by making the best practices clearer, more principle based, and therefore easier for companies to implement on a
conceptual basis. This is also why there is a larger emphasis on performing duties correctly beforehand, instead of accounting for them afterwards (Press release, 2008). In June 2008 a proposal for the revised code was published. Different interest groups gave their feedback on the proposal, after which the revised code went into action January 1st, 2009.
Changes consisted mainly of clarifications and slight changes within the existing best practices. A clearer focus was put on principles in the areas of risk management, executive pay, shareholder responsibility, diversity in the composition of the supervisory board and corporate social responsibility. Sixteen provisions were added to the code, including new best practice provisions on executive remuneration, which should reflect long-term objectives (Code Tabaksblat, 2009). The new provisions in the code with respect to directors’ pay (II.2.10 and II.2.11) also have the intention to increase the power of the supervisory board in the level of directors’ remuneration. With the sixteen new provisions, the new code now consists of 129 best practices.
2.4 Hypothesis
The actualization of the Tabaksblat code has led to a code that is clearer and more principle based than the code that was in place since 2004. This means that it is made easier for companies to comply with the new code on a principle basis, instead of checking boxes with the detailed old code. Principles were clarified and new important issues on several topics were added to the new code to make it up-to-date. The new code has led to more
transparency within the applicable companies, and created a better environment for supervision of the board (Code Tabaksblat, 2009). Akkermans et al. (2007) and Nordberg (2011) argue that this increase in transparency and supervision would lead to a decrease in agency problems caused by the information asymmetry, leading to better decision-making regarding efficient resource allocation. Among other researches, Goncharov (2006) shows that better governance is accompanied by better company performance. Taking the results of Bauer et al. (2004) in consideration, the Dutch corporate governance code was already quite good before the actualization of the code. It could therefore be the case that the effect of the new code on company performance is somewhat tempered. Keeping the above-mentioned literature in consideration, the main hypothesis is as follows:
H1: The actualization of the Tabaksblat code has had a significantly positive effect on the relation between compliance with the code and firm valuation
This main hypothesis can be subdivided into two parts: the compliance with the code and firm valuation. The next two hypotheses have been created to address the two parts of the main hypothesis. The previous literature has led to believe that the new code is easier to comply with due to the best practices that are clearer in what is expected of the company. Following this new code will create a more transparent company (Code Tabaksblat, 2009). The
transparent company will be valued higher, since literature assumes that due to the diminishing agency problems, the managers will make better investments (Jensen and Meckling, 1976). These theories lead to the following hypotheses:
H2: The actualization of the Tabaksblat code has had a significantly positive effect on firm valuation
H3: The actualization of the Tabaksblat code has had a significantly positive effect on compliance with the code.
3. Methodology and data
The methodology used in this research will be similar to those of other researches done before by for example Goncharov, Werner and Zimmerman (2006), Gompers et al. (2003) and de Jong et al. (2005). Section 3.1 will give a thorough explanation of the choice of variables and the model that has been created with them. After that, section 3.2 gives the descriptive statistics of the variables that are included in the models.
3.1 Methodology
The different research methods of studies mentioned above have been combined into a new model. Following Goncharov’s research method a dummy variable for time will be added to the model, making it possible to run a regular OLS regression. In this case this dummy will be the variable ‘TIME’, which will distinguish the period before the adaptation and the period after. The variable of interest in this study is the interaction variable ‘TIME*COMPL’, since this variable indicates the difference in the effect of compliance before and after the
actualization of the code. The control variables are added to limit the effect of other factors on Tobin’s Q. 𝛽! is the intercept of the model, which indicates the value ‘TBQ’ when all
independent variables are equal to zero.
TBQ = 𝛽! + 𝛽! TIME + 𝛽! COMPL! + 𝛽! TIME*COMPL!+ 𝛽!LOG_AGE + 𝛽!LOG_ASSETS
+ 𝛽!ROA + ε
The indicator ‘i’ on the compliance variables can take the values {E, 2, 2.2, 3, 4}. This approach avoids the different chapters to be in one model, which has been done to prevent multicollinearity (de Jong et al., 2005). Five different models will be created, one for
every relevant chapter5. The indicator i for ′COMPL!′ and ‘TIME*COMPL!’ indicates that there are separate models for: the entire code (E), which includes chapters one through five; chapter 2 on the management board (2); principle 2 of chapter 2 on management remuneration (2.2); chapter 3 on the supervisory board (3) and chapter 4 on the shareholders (4).
The dependent variable in every model is ‘TBQ’, which is the Tobin’s Q ratio according to the definition by Chung and Pruitt (1994), further explained in section 3.3. The dummy variable ‘TIME’ will take the value of 1 when addressing values in the period after the adaption of the Tabaksblat code (2009-2010), and will be 0 during the period before the adaption (2007-2008). ‘COMPL!’ is the compliance variable in the regression, measuring the compliance of the indicated chapter(s) by taking the percentage of compliance with respect to all best practices of the specific chapter(s). ‘TIME ∗ COMPL!’ is the actual variable of interest, and is an interaction variable that indicates the relationship between compliance and Tobin’s Q. In line with previous research, age and assets have been taken into the model as a natural logarithm. ‘LOG_AGE’ is the natural logarithm of the company in years, measured from the year of incorporation to the year of the relevant data. ‘LOG_ASSETS’ is the natural logarithm of the book value of assets measured in euros. ROA is the return on assets in percentages.
Besides assessing the data with the regression analysis, t-tests are also used on all variables to see if there are any relevant changes between the two periods. The hypothesis of increased compliance with the code after the adaptation can be tested for every separate chapter. Besides that, a change in Tobin’s Q before and after the adaptation may also become apparent.
3.1.1 Corporate Governance measure
As a measure of corporate governance compliance with the best practices of the entire Tabaksblat code will be used. For 2007 and 2008 this means 113 best practices, and in 2009 and 2010 there are 129 best practices. Compliance for every best practice is hand collected from the annual report of the companies, as they are required to include this information in their yearly report. The data for 2008 and 2009 have been obtained from Vriens (2011). Previous research has shown that some corporate governance measures might have a negative effect on firm value, which means that they may cancel out the positive effect of other measures, leaving a net zero effect on firm value (Bebchuck et al., 2009). Keeping this
5 Chapters 2, 3 and 4 can be seen as relevant since these are the chapters that include non-compliance. Chapters 1
and 5 have been fully complied with and are therefore not included in a separate regression. Chapter 2 principle 2 has been included since this is the principle least complied with. It is interesting to see if there have been improvements within the compliance in this principle.
in mind, I will divide the code into the different chapters and include a variable for every chapter that has not been fully complied with in the period 2007-2010. Since chapter one and five have been fully complied with by all companies in the sample period, these will not be used in the regression models. Chapter 2 principle 2 (II.2) will also be included, since this principle has the lowest compliance of the code. It is the principle on director remuneration, and deserves special attention to see if the new code has improved compliance with this principle. The overall compliance will be measured as a fraction of complied best practices divided by the total amount of best practices. To see the overall effect of the revision of the code, another regression will be run with the compliance of the code as a whole.
3.1.2 Firm valuation
As in several previous studies, company value will be measured by Tobin’s Q. In many corporate governance studies, such as by Goncharov et al. (2006), stock returns are used as a performance measure. In my study, I will intentionally use Tobin’s Q instead of stock returns, because several studies show that there is a clearer relationship between corporate governance and Tobin’s Q than there is with stock returns (Edwards and Clough, 2005; Bhagat and Bolton, 2008).
In this research the definition of Tobin’s Q will be the one that Chung and Pruitt (1994) present in their paper. They show an easy way of calculating Tobin’s Q, where data consists of market value of equity, market value of liabilities, book value of equity and book value of liabilities. These data have been obtained from the Datastream database.
𝑇𝑜𝑏𝑖𝑛!𝑠 𝑄 = !"#$%& !"#$% !" !"#$%&!!"#$%& !"#$% !" !"#$"!"%"&' !""# !"#$% !" !"#$%&!!""# !"#$% !" !"#$"!!!"#$
3.1.3 Control variables
Since Tobin’s Q can be influenced by several factors, control variables have to be used. The main control variable that is used in other research that has been done, is company age (Bebchuck et al., 2009; Goncharov et al., 2006; Gompers et al., 2003). Other control variables proven to be of significance are ROA (Bebchuck et al., 2009; Gompers et al., 2003) and book value of assets (Bebchuck et al., 2009; Jong et al., 2005). These three control variables will be used in my regression model. Based on prior research, the
coefficient for book value of assets and company age is expected to be negative, while ROA is to have a positive effect. The data on these control variables have been collected using the
Datastream database. Since the data from Datastream had several missing values, the dataset has been supplemented with data from the Amadeus database.
3.1.4 Sample
The Tabaksblat code applies to all listed companies that have their statutory seat in the Netherlands. Because the data on compliance with the code has to be hand collected from annual reports, only a relatively small sample is used. Since larger companies usually have a larger degree of compliance (Gompers et al., 2003) and have their information readily available, this research will initially focus on the companies listed at the Amsterdam Exchange Index (AEX), thereby including the 25 largest Dutch companies. Of these 25 companies, three have their statutory seat outside the Netherlands; hence they do not have to comply with the Dutch corporate governance code and will be taken out of the sample group. Another requirement to be included in the sample is that the companies have to be listed throughout the entire period on which data is gathered, which is two years before the
introduction of the adaption, to two years afterwards (2007-2010). A group of 20 companies remains after the conditions have been taken into account. These companies can be found in table 2.
3.2 Data
In this section, an elaboration is given on the descriptive statistics of the sample, the compliance statistics per company and the variation of the parameters. These tables can be found in table 1, 2 and appendix 1 respectively, and will be referred to in the text below.
3.2.1 Tobin’s Q
Table 1 shows the overall descriptive statistics for Tobin’s Q. It can be noted that the minimum and maximum values are fairly spread ranging from 0.80 to 3.518, while the standard deviation is rather small with a value of 0.601. This means that during the years 2007-2010, most firms had a Tobin’s Q that was similar to the other firms with a value of around 1.478, the mean of the sample. Appendix 1 shows that a large part of the variation is due to the variation between the different companies, which can also be seen in table 1. However, there was also some variation within the companies itself, meaning it varied over time within the same company. It should be noted that Reed Elsevier has been taken out of this data as an outlier. Tobins’ Q of this firm is ranged from 5.08 to 10.02. This means that its Tobin’s Q ranges anywhere from 2.6 to 6.1 standard deviations from the mean. It is therefore
far outside the 99 percent confidence interval for all individual years and is therefore excluded6. For Tobin’s Q, this leaves a sample of 19 firms with a total of 76 observations.
Table 1. Descriptive statistics
Variable Mean Median Minimum Maximum Std. dev. 5% 95% TBQ 1.478 1.30 .80 3.518 0.601 1.341 1.615 COMPL! .964 0.97 .832 1 .033 .957 .971 COMPL! .917 .93 .64 1 .078 .900 .935 COMPL!.! .878 .93 .5 1 .122 .850 .905 COMPL! .973 .98 .816 1 .042 .964 .982 COMPL! .976 1 .852 1 .032 .969 .983 AGE 79.375 70.5 2 267 65.244 64.856 93.894 LOG_AGE 3.951 4.25 .693 5.587 1.051 3.718 4.186 ASSETS (in millions) 14600 7400 1010 67200 14500 1140 17900 LOG_ASSETS 22.940 22.73 20.733 24.931 1 22.718 23.163 ROA .057 0.05 -.045 .239 .056 .045 .070 This table indicates the descriptive statistics for every variable included in the regressions, as well as assets and age of the companies from the sample. The presented data include all companies from the sample; the only exception is the data on TBQ, for which Reed Elsevier has been taken out as an outlier. COMPL! is the
compliance with the Tabaksblat code, with an indicator for the separate chapters. 2 stands for the chapter on the management board, 2.2: management remuneration, 3; supervisory board, 4; shareholders and E stands for the entire code. The data is obtained from the Amadeus database as well as from Datastream.
3.2.2 Compliance
With an overall compliance average of 96.4%, table 1 shows that the overall compliance with the code is relatively high. Appendix 1 indicates that there is also very little variation in the total compliance. Table 2 shows the compliance statistics per company, in which it can be seen that ING noticeably deviates from the sample in 2007 and 2008 by only complying with 83.2 percent on average. After the actualization of the code, ING increased its compliance to 96.5 percent and therefore no longer shows to have a large deviation from the sample.
Noteworthy is the large variation of the compliance in chapter 2 principle 2 on director remuneration. While ASML only complied with 53.6% of this principle, there were also 5 companies that fully complied with this principle during the entire sample period. This standard deviation of .122 mainly points to variation between the different firms from the sample. The variation over time for each separate company was relatively small, which can also be seen in table 2. This means that companies that did not comply with the principle in 2007 will most likely deviate in the end of the sample period (2010) as well.
6 Most likely this is a result of wrong or missing data in Datastream, which resulted in a wrong calculation within
Table 2. Compliance per company, before and after actualization
Compliance in %
Entire code Chapter 2 Chapter 3 Chapter 4 Principle II.2 time before after before after before after before after before after Aegon 96.9 97.7 92.0 93.3 99.0 100 96.3 97.1 92.9 93.3 Ahold 100 100 100 100 100 100 100 100 100 100 Akzo Nobel 96.9 98.4 94.0 96.7 96.9 100 98.1 97.1 89.3 100 ASML 93.4 95.3 74.0 83.3 98.0 98.0 100 100 53.6 66.7 BAM 97.3 98.4 92.0 93.3 99.0 100 98.1 100 85.7 86.7 Corio 100 100 100 100 100 100 100 100 100 100 DSM 99.1 97.7 100 93.3 98.0 98.0 100 100 100 100 Fugro 92.9 93.8 88.0 93.3 95.9 96.0 87.0 88.6 85.7 93.3 Heineken 92.0 93.8 94.0 93.3 85.7 88.0 98.1 100 96.4 96.7 ING 83.2 96.5 66.0 95.0 81.6 96.0 94.4 97.1 67.9 90.0 KPN 96.0 98.4 86.0 93.3 98.0 100 100 100 75.0 86.7 Philips 95.6 100 96.0 100 93.9 100 96.3 100 92.9 100 Randstad 96.5 96.9 86.0 90.0 99.0 98.0 100 100 82.1 86.7 Reed -Elsevier 92.9 93.8 88.0 93.3 95.9 96.0 87.0 88.6 85.7 93.3 SBM -Offshore 98.2 98.4 100 100 98.0 98.0 96.3 97.1 100 100 TNT 97.8 97.7 92.0 90.0 99.0 100 100 100 85.7 80.0 TomTom 92.9 98.1 88.0 95.0 91.8 99.0 96.3 98.6 78.6 90.0 Unilever 95.6 96.1 88.0 90.0 100 100 92.6 94.3 78.6 80.0 Wereldhave 99.1 99.2 100 100 100 100 96.3 97.1 100 100 Wolters-Kluwer 94.7 97.3 80.0 88.3 98.0 100 100 100 71.4 83.3 Average 95.6 97.4 90.2 94.1 96.4 98.4 96.8 97.8 86.1 91.3
This table reports the compliance variables separated by company. The stated numbers in the column before are averages in compliance of the years before the adaptation of the Tabaksblat code (2007-2008). The numbers in the column after are averages of the years 2009-2010.
3.2.3 Control variables
In the regression models, firm age, assets and return on assets have been used as control variables. The large standard deviations of all control variables show that the sample consists of a widely varied set of firms with respect to these variables. The standard deviation for firm age is a result of variation between firms, as the only variation within the firm can be
explained as the firm aging a year every consecutive year of the sample period. With an overall standard deviation of 65.244, the youngest firm, TomTom, had an incorporation date only 2 years before the start of the sample period in 2007. The oldest firm is ING with an incorporation year dating 267 years back counting from 2010.
For book value of assets the large standard deviation also mainly comes from the variation between the different companies, which is logical since the change in book value of assets over the years is not as large as the difference in book value between the companies.
Table 1 shows that the lowest assets value, which belongs to Fugro, is 1.01 billion euros. ING is not only the oldest company, but also has the highest book value of assets, since in 2010 they had a book value of 67.2 billion. The sample is somewhat split in half when it comes to larger and smaller companies according to book value of assets.
The return on assets varies between -4.5% to 23.9% (see table 1). Appendix 1 shows that this large spread is primarily due to variation within the sample, meaning that some firms have had a specifically high or low ROA during one year of the sample period, while other years they have had returns around the mean of 5.7% return. There was a slight insignificant decline in return on assets after the actualization. This decline is due to an overall decline in ROA and cannot be attributed to one or several separate companies.
3.2.4 Correlations
Appendix 2 shows the correlation matrix for the variables that have been used in the
regressions. The correlations between the compliance of different chapters can be disregarded since these will not be included in the same regression model. However, it is interesting to see that the degree of compliance of a single chapter does not necessarily correlate with the degree of compliance of another chapter, but that compliance in almost every single case significantly correlates with the dummy variable ‘TIME’. This implicates that compliance increased after the adaptation of the code. High correlation between independent variables may lead to multicollinearity within the regression model, which leads to a model that is less precise. In this case, the highest correlation is .507, which is the correlation between
‘LOG_AGE’ and ‘LOG_ASSETS’. Although this correlation is significant, it is the only one exceeding 50%; ‘LOG_ASSETS’ will therefore not be excluded from the regression analyses. All compliance variables present a negative correlation with ‘TBQ’. The negative correlation between TBQ and COMPL! contradicts prior research, since the overall compliance should lead to a positive effect on Tobin’s Q. On the other hand, the correlation of the compliance of different chapters with Tobin’s Q can be explained by Bebchuck et al. (2009) and Gompers et al. (2003), as they state that not every corporate governance variable has a positive influence on Tobin’s Q; some even have a negative one. In line with prior research, ‘LOG_AGE’ and ‘LOG_ASSETS’ both show a negative correlation with ‘TBQ’, although the latter one is not significant.
4. Empirical results
The next section focuses on the results of the univariate tests, as well as the regression results. Furthermore, there have been some tests on robustness and omitted variable bias, which will
4.1 Univariate analysis
In table 3, the descriptive statistics are separated into the periods before and after the actualization. The two means are compared using a two-tailed t-test. For Tobin’s Q, a small decline from 1.5 to 1.46 is present. This decline is not significant, so this finding might be a result of random variations within the sample. However, this decline is not attributed to a decline from a specific company, but is the result of small variations in the value of all companies.
Table 3. Univariate analysis of the included variables
Before adaptation After adaptation Difference in means Mean Std. dev Mean Std. dev Difference t-value
TBQ 1.50 0.67 1.46 0.53 -0.04 -0.324 COMPL! 0.95 0.04 0.97 0.02 0.02*** 2.639 COMPL! 0.90 0.09 0.94 0.05 0.04** 2.168 COMPL!.! 0.85 0.13 0.90 0.10 0.05* 1.919 COMPL! 0.96 0.05 0.98 0.03 0.02** 2.227 COMPL! 0.97 0.04 0.98 0.03 0.01 1.305 AGE 78.30 65.72 80.45 65.59 2.15 0.146 ROA 0.06 0.12 0.05 0.05 -0.02* -1.750 ASSETS (in millions) 14500 13600 15000 15600 1440 0.478 LOG_ASSETS 22.90 0.99 22.98 1.02 0.08 0.546 LOG_AGE 3.91 1.11 3.99 1.00 0.08 0.344
This table states the varying descriptive statistics before (2007-2008) and after (2009-2010) the adaptation of the Tabaksblat code. In column 3, the difference between ‘before’ and ‘after’ is given. The sample consists of 76 observations; 38 before and 38 after the adaptation. COMPL! is the compliance with the Tabaksblat code, with an indicator for the separate chapters. 2 stands for the chapter on the management board, 2.2: management
remuneration, 3; supervisory board, 4; shareholders and E stands for the entire code. The t-value is obtained from running a two-tailed t-test. *, ** and *** indicate a statistical significance on the 10%, 5% and 1% level respectively.
Table 3 shows that compliance with the code in general, as well as chapter 2 and 3 on the management and supervisory board, and principle 2 of chapter 2 on directors
remuneration have significantly increased after the actualization with 2, 5, 2 and 4 percent additional compliance respectively. Chapter 4 compliance which incorporates shareholder rights has also increased from 97 to 98 percent, but this has not shown to be significant (t = 1.305). Besides that, all standard deviations have dropped, meaning that compliance has not
only increased but there are also not as many variations within the sample with regard to compliance.
Return on assets has a significant decrease between the periods before (2007-2008) and after (2009-2010) the adaptation of the code. From this research it cannot be concluded if this is the result of actualization of the Tabaksblat code or if there are other reasons for this significant decline. None of the other control variables show significant changes in the period before and after the adaptation.
4.2 Regression analysis
The empirical model described in section 3.1 will give an overview of the different variables effecting the dependent variable, Tobin’s Q. The results of the OLS regression can be found in tables 4 and 5 presented below. The variable of interest for this study is the interaction variable ‘TIME*COMPL’ since this variable shows if the effect of compliance on ‘TBQ’ has changed after the adaption of the code. Before running the regressions, a Shapiro-Wilk test for normality of the residuals was performed. Since the residuals of model 1, 2 and 3 were
normally distributed at the one percent level, they were regressed using an OLS regression. Model 4 and 5 did not have normally distributed errors and have therefore been regressed using a robust OLS regression. Furthermore, a Ramsey RESET test and vif test have been performed; these results are discussed in section 4.3.
In table 4 the regression results of the three regular OLS regressions are presented. All models show a similar pattern in which ‘LOG_ASSETS’ is the only independent variable that is not significant. With 49-60.8%, the R² of the regressions are in a similar range of those presented in the study by de Jong et al. (2005). This value means that the presented models can explain 49-60.8% of the variations in ‘TBQ’. As the model contains an interaction variable between ‘TIME’ and ‘COMPL’, these variables cannot separately be interpreted in the regressed models. The interaction term, however, shows that the effect of compliance in all cases has significantly decreased after the adaptation of the code. This contradicts the hypothesis, but may be partially explained by the high standard of corporate governance present before the adaption of the code, as explained by Bauer et al. (2004). ‘ROA’
demonstrates to have a positive effect on ‘TBQ’, which means firms with a higher return on assets tend to have a higher Tobin’s Q, keeping the other variables constant. ‘LOG_AGE’ has a negative impact on firm value measured in Tobin’s Q. Both of these findings are in line with previous research by Bebchuck et al. (2009). Previous literature together with the Tobin’s Q formula creates an expectation of a positive effect of book value of assets on ‘TBQ’. In my
study, however, my findings have resulted in a small, negative effect of ‘LOG_ASSETS’ on ‘TBQ’, but this effect is not significant in any regression.
Table 4. Regression results of the OLS regressions
TBQ (1) Entire Code (2) Chapter II (3) Chapter II.2 Coefficient t-value Coefficient t-value Coefficient t-value Constant 6.720*** 2.70 3.961** 2.61 3.899*** 2.97 TIME 8.943** 2.09 4.821*** 3.32 1.870** 2.53 COMPL! -4.497** -2.34 TIME*COMPL! -9.067** -2.06 COMPL! -1.887** -2.55 TIME*COMPL! -5.035*** -3.24 COMPL!.! -1.804*** -3.60 TIME*COMPL!.! -1.940** -2.37 ROA 3.099*** 5.46 3.187*** 6.09 3.110*** 6.27 LOG_AGE -.163*** -2.89 -.136** -2.58 -.098* -1.92 LOG_ASSETS -.020 -0.31 -.018 -0.30 -.028 -0.50 F-statistic 11.03*** 14.83*** 17.85*** R² 0.490 0.563 0.608
Every column presents a different regression model. Column 1 presents the results of the regression including the entire Tabaksblat code, while column 2 and 3 represent the models on chapters II (the board) and II.2 (remuneration of the board) respectively. COMPL! is the compliance with the Tabaksblat code, with an indicator
for the separate chapters. 2 stands for the chapter on the management board, 2.2: management remuneration, 3; supervisory board, 4; shareholders and E stands for the entire code. The t-value is obtained using a two-tailed test. *, ** and *** indicate significance at the 10, 5 and 1 percent level respectively.
Table 5 presents the results of the robust OLS regressions, were done since the residuals of these models did not fit the normal distribution. What strikes the attention is that both models 4 and 5 have less significant variables than the other models that were run with the regular regression. The R² is also lower in the regressions on Chapter III and IV. Both of these effects may be caused by the robustness of the regressions, which makes the regressions less precise. However, ‘ROA’ is still significantly positive in both regressions (t = 9.76 resp. t = 10.33), which is in line with prior research. The results cannot be given statistical
significance however, because the chapter III model has an F-value of .72, which means that with a corresponding p-value of .609, the model does not significantly explain ‘TBQ’. The same is the case for the chapter IV model, which has an F value of .83 with a corresponding p-value of .536. This value leads to the conclusion that model number 4 and 5 are not of statistical significance.
Table 5. Regression results of the robust regressions
TBQ (4) Chapter III (5) Chapter IV Coefficient t-value Coefficient t-value Constant 1.223 0.73 -4.360** -2.44 TIME -0.407 -0.19 5.817** 2.52 COMPL! -0.324 -0.28 TIME*COMPL! 0.478 0.22 COMPL! 4.470* 2.90 TIME*COMPL! -5.850** -2.48 ROA 6.667*** 9.76 6.720*** 10.33 LOG_AGE -0.021 -0.43 -0.020 -0.46 LOG_ASSETS 0.008 -0.16 0.045 1.02 F-statistic 7.37*** 6.66*** R² 0.382 0.414
This table presents model 4 and 5, which have been regressed using a robust regression. Model 4 includes chapter III on the supervisory board, while model 5 includes chapter 4 on the shareholders. COMPL! is the
compliance with the Tabaksblat code, with an indicator for the separate chapters. 2 stands for the chapter on the management board, 2.2: management remuneration, 3; supervisory board, 4; shareholders and E stands for the entire code. The t-value is obtained using a two-tailed test. *, ** and *** indicate significance at the 10, 5 and 1 percent level respectively.
4.3 Robustness check and additional tests
In order for to test for multicollinearity, a vif (variance inflation factor) test has been
performed. Before this test could be performed, the compliance term in every model had been centered to avoid collinearity between the interaction term and its components. This resulted in a centered interaction term in models 1 through 5. The resulting regressions have not been presented, since the parameters are merely centered to perform the multicollinearity test. The interpretations of the models are performed on the regular regressions, since these give more meaningful results in terms of interpretations of the parameters. A mean vif output value above 10 would indicate substantial multicollinearity. The mean vif value in the regression models described above range anywhere from 1.30 (model 2) to 1.42 (model 1). All results can be found in appendix 3. The results of the vif test indicate that there is no substantial multicollinearity, implying that the parameters can be precisely estimated. This result is in line with the low t-values of the individual parameters.
Furthermore, a Ramsey RESET was done to test for omitted variables. Omitted variables are variables that are not included in the model but are still affecting the dependent variable TBQ. Omitted variables can lead to inconsistent estimations of the parameters in the regression models. In the RESET test, the null hypothesis states that the model has no omitted variables. The resulting p-values range from 0.000 (models 2 and 3) to 0.072 (model 4); meaning in all models there is a significant case of omitted variable bias. The additional
results are presented in appendix 3. Due to the scope of this thesis, no additional parameters are added to the presented models.
5. Conclusion and discussion
The main hypothesis has been tested using a multivariate analysis. The results of the regression contradicted the hypothesis of a positive effect on the relation between compliance with the Tabaksblat code and firm valuation. This means that the new code has less effect on firm value than the old code did, using Tobin’s Q as a measure of firm value. This result may be explained by the theory presented in the paper by Bauer et al. (2004). The theory states that in case there is a better base of corporate governance present, compliance with a code will not necessarily increase firm value. The decreasing effect of compliance on firm valuation may therefore be a result of a good base of corporate governance that has already been present with the old code. The influence of the new code is therefore smaller than the influence of the old one, since major changes in corporate governance practices had already been made during the introduction of the old code. Caution has to be used when concluding this, because the sample period before the adaption of the code were in 2007 and 2008, so the economic
recession might have influenced the results. The results may therefore indicate that at the start of the economic recession there was a sharper focus on compliance by investors, which strengthened the relationship between compliance and firm value compared to the later years of the recession.
The two supporting hypotheses have been tested using univariate tests in the form of two-sample t-tests. The t-tests on compliance present a significant increase in compliance with the code in its entirety, which is in line with the results found by Vriens (2010). The separate chapters have also been tested on improvement of compliance. Chapters I and V on enforcement and audit had full compliance before and after the adaption of the code. Chapters II and III on the management and supervisory board show significant improvements in
compliance, while chapter IV on shareholder rights improves, but not significantly (t = 1.305). The increase in compliance may be attributed to the clarity of the best practices, which means that companies now know more of what is expected of them and can better comply with the best practices accordingly. Tobin’s Q did not increase significantly when tested with a t-test (t = -0.324), indicating that it remained stable during the sample period. The regression that includes the entire model, however, results in a positive coefficient for the time dummy variable TBQ, meaning that there was a significant increase in Tobin’s Q when all other variables included in the model are kept constant.
The results of this study need to be interpreted with care, since there are several limitations to this research. The sample size is fairly small with only 20 companies that were observed over a period of four years, creating an initial sample of 80 firm years. Besides that, there was no control for endogeneity, which might be addressed in a future study using instrumental variables. The Ramsey RESET test has proven that there are omitted variables in the regressions, meaning that there are some other variables influencing Tobin’s Q. Further research may be focused on extending the models with other variables that have proven to be significant in previous literature, such as insider block holdings of stocks, leverage of a firm and growth of the firm’s assets (Jong et al., 2005). It would then also be beneficial to run the regressions using a larger sample; this way the model will be more precise in estimating the effect on Tobin’s Q.
Reference list
Akkermans, D., Van Ees, H., Hermes, N. et al. (2007) Corporate Governance in the Netherlands: an overview of the application of the Tabaksblat Code in 2004. Corporate Governance: An International Review, 15(6), 1106–1118
Ammann, M., Oesch, D., & Schmid, M. M. (2011). Corporate governance and firm value: International evidence. Journal of Empirical Finance, 18(1), 36-55.
Bauer, R., Guenster, N. & Otten, R. (2004). Empirical Evidence on corporate governance in Europe: The effect on stock returns, firm value and performance. Journal of Asset Management, 5(2), 91 –104
Bebchuk, L.A., Cohen, A. & Ferrell, A. (2009) What Matters in Corporate Governance? Review of Financial Studies, 22(2), 783-827
Bhagat, S., & Bolton, B. (2008). Corporate governance and firm performance. Journal of Corporate Finance, 14(3), 257-273
Brown, L. D., & Caylor, M. L. (2006). Corporate governance and firm valuation. Journal of Accounting and Public Policy, 25(4), 409-434.
Chung, K.H. & Pruitt, S.W. (1994) A Simple Approximation of Tobin's Q. Financial Management, 23(3), 70-74
Code Tabaksblat (2008) Dutch Corporate Governance Code.
http://www.commissiecorporategovernance.nl/corporate-governance-code Denis, D. K. (2001). Twenty-five years of corporate governance research … and
counting. Review of Financial Economics, 10(3), 191-212
Edwards, M., & Clough, R. (2005). Corporate governance and performance: An exploration of the connection in a public sector context. University of Canberra, Canberra. Gompers P., Ishii J., Metrick A. (2003) Corporate Governance and Equity Prices. Quarterly
Journal of Economics, 118(1), 107-156
Goncharov, I., Werner, J.R. et al. (2006) Does Compliance with the German Corporate Governance Code Have an Impact on Stock Valuation? An empirical analysis. Corporate Governance: An International Review, 14(5), 432-445
Groenewald, E. (2005). Corporate Governance in the Netherlands: From the Verdam Report of 1964 to the Tabaksblat Code of 2003. European Business Organization Law Review, Vol. 6, p. 291 – 311.
Jensen, M. C., & Meckling, W. H. (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305-360 Jong, A. de, DeJong, D. V., Mertens, G., & Wasley, C. E. (2005). The role of self-regulation
in corporate governance: Evidence and implications from the Netherlands. Journal of Corporate Finance, 11(3), 473-503
Lemmon, M. L., & Lins, K. V. (2003). Ownership structure, corporate governance, and firm value: Evidence from the east asian financial crisis. The Journal of Finance, 58(4), 1445-1468.
Monitoring Committee Corporate Governance Code
http://www.commissiecorporategovernance.nl/ Consulted 18 Mei 2013
Nordberg, D. (2011) Corporate Governance Principles and Issues, SAGE Publications Press Release (2008). Monitoring Committee presents updated Corporate Governance Code.
Retrieved on 4th June 2013 from: http://commissiecorporategovernance.nl/nieuws/
554/Monitoring-Commissie-presenteert-geactualiseerde-corporate-governance-code Shleifer, A. & Vishny, R. (1997) A Survey of Corporate Governance. The Journal of
Finance, 52(2), 737-783
Vriens, Z. (2011) Heeft de actualisatie van de Code Tabaksblat in 2008 geleid tot een grotere toepassing van de code onder de Nederlandse AEX ondernemingen? Bachelor Thesis Universiteit van Tilburg
Appendix 1 - Variation description
Variable Std. dev. Variable Std. dev.
TBQ overall .601 COMPL! overall .032
between .503 between .031 within .343 within .010
COMPL! overall .033 AGE overall 65.244
between .027 between 66.505 within .019 within 1.354
COMPL! overall .078 ASSETS
(in millions)
overall 14600 between .065 between 14700 within .044 within 2270
COMPL!.! overall .122 ROA overall .090
between .112 between .041 within .053 within .080
COMPL! overall .042 LOG_AGE overall 1.051
between .036 between 1.068
within .022 within .090
This table gives an explanation of the variances of the variables. Overall variation is the standard deviation. Between can be explained as variation in the sample between the different companies. Within variation is variation in the same company but over the different years of the sample. COMPL! is the compliance with the
Tabaksblat code, with an indicator for the separate chapters, 2 stands for the chapter on the management board, 2.2: management remuneration, 3; supervisory board, 4; shareholders and E stands for the entire code.
Appendix 2 - Pearson’s correlation matrix 1 2 3 4 5 6 7 8 9 10 (1) TBQ 1.000 (2) 𝐂𝐎𝐌𝐏𝐋𝐄 -0.298*** 1.000 (3) 𝐂𝐎𝐌𝐏𝐋𝟐 -0.399*** (0.813***) 1.000 (4) 𝐂𝐎𝐌𝐏𝐋𝟐.𝟐 -0.543*** (0.635***) (0.909***) 1.000 (5) 𝐂𝐎𝐌𝐏𝐋𝟑 -0.081 (0.830***) (0.435***) (0.218**) 1.000 (6) 𝐂𝐎𝐌𝐏𝐋𝟒 -0.170 (0.480***) (0.181) (0.113) (0.257**) 1.000 (7) ROA 0.516*** -0.049 -0.063 -0.097 0.041 -0.171 1.000 (8) LOG_AGE -0.347*** 0.030 0.124 0.256** -0.129 0.133 -0.086 1.000 (9) LOG_ASSETS -0.186 -0.129 -0.062 0.022 -0.204* 0.102 -0.085 0.507*** 1.000 (10) TIME -0.038 0.286*** 0.238** 0.212* 0.245** 0.146 -0.070 0.039 0.063 1.000
This table states the Pearson’s correlations between the variables that are used in the empirical regression model. COMPL! is the compliance with
the code, with an indicator for the separate chapters. 2 stands for the chapter on the management board, 2.2: management remuneration, 3; supervisory board, 4; shareholders and E stands for the entire code.. The values in parentheses have been presented this waysince they will not be included in the same model, so the correlations between these variables are not of any significance to the results. *,** and *** indicate significance at the 10%, 5% and 1% level respectively.
Appendix 3 – Results of additional tests
Regression model vif test Ramsey RESET test vif - value F-value p-value 1) entire code 1.42 3.54 0.019 2) chapter 2 1.30 10.62 0.000 3) chapter 2.2 1.38 9.56 0.000 4) chapter 3 1.31 2.45 0.072 5) chapter 4 1.38 2.98 0.038
This table presents the results of the vif test and the Ramsey RESET test. *,** and *** indicate significance at the 10%, 5% and 1% level respectively.
Appendix 4 – Compliance with the Tabaksblat code for the years 2007-2010
Appendix 4 gives an overview of the compliance with the Tabaksblat code for the separate firms from 2007 through 2010. The best practices are listed in order of the different chapters. An ‘x’ marks non-compliance with that specific best practice. The best practices that had been fully complied with have not been included in these tables to obtain a table that is as concise as possible.
Non-compliance with Tabaksblat code in 2010
Chapters/ Best practices Ae go n Ah old Ak zo No be l AS M L Ho ld in g BA M g ro ep Co rio DS M Fu gr o He in ek en IN G g ro ep KP N Ph ili ps Ra nd sta d Re ed E lse vie r SB M O ffs ho re TN T To m To m Un ile ve r We re ld ha ve Wo lte rs -Kl uwe r II 1.1 x x x x 1.8 x 2.3 x 2.4 x x x 2.5 x x x x x 2.8 x x x x x x x x x x x x 2.9 x 2.13 x x x 2.14 x 3.3 x 3.4 x III 2.1 x x 2.2 x x 2.3 x 3.4 x 3.5 x x 4.1 x 5.11 x x x x x 6.3 x 6.6 x 7.3 x IV 1.1 x x x x x x 1.2 x 2.1 x 2.2 x 2.5 x 2.8 x 3.1 x x 3.10 x Compliance (out of 129) 126 129 128 125 127 129 126 121 121 125 127 129 125 120 127 126 127 124 128 126
Non-compliance with Tabaksblat code in 2009 Chapters/ Best practices Ae go n Ah old Ak zo No be l AS M L Ho ld in g BA M g ro ep Co rio DS M Fu gr o He in ek en IN G g ro ep KP N Ph ili ps Ra nd sta d Re ed E lse vie r SB M Of fs ho re TN T To m To m Un ile ve r We re ld ha ve Wo lte rs -Kl uwe r II 1.1 x x x x x 1.2 x 1.8 x x x 1.9 x 2.1 x 2.2 x 2.3 x x 2.4 x x 2.5 x x x x x x 2.8 x x x x x x x x x x x x 2.9 x 2.10 x 2.11 x 2.13 x x x x 2.14 x 3.3 x 3.4 x III 2.1 x 2.2 x x 3.1 x 3.4 x 3.5 x x x 5.6 x 5.11 x x x x x 6.3 x 6.6 x 7.3 x IV 1.1 x x x x x 1.2 x 2.1 x 2.2 x 2.5 x 2.8 x 3.1 x x 3.10 x Compliance (out of 129) 126 129 126 121 127 129 126 121 121 124 127 129 125 119 127 126 126 124 128 125 Source: Vriens (2011)
Non-compliance with Tabaksblat code in 2008 Chapters/ Best practices Ae go n Ah old Ak zo No be l AS M L Ho ld in g BA M g ro ep Co rio DS M Fu gr o He in ek en IN G gr oe p KP N Ph ili ps Ra nd sta d Re ed E lse vie r SB M O ffs ho re TN T To m To m Un ile ve r We re ld ha ve Wo lte rs -Kl uwe r II 1.1 x x x x x 2.1 x x x 2.2 x x x 2.3 x x 2.4 x x 2.5 x x x x 2.6 x x x x x 2.7 x x x x x 2.8 x x x x x x x x 2.9 x 2.10 x x x 2.11 x x x 2.13 x 3.1 x 3.2 x 3.3 x x III 2.1 x 2.2 x x x 3.4 x x 3.5 x x x 4.2 x 5.1 x 5.2 x 5.3 x 5.4 x 5.6 x 5.11 x x x x x x x 6.1 x 6.3 x 6.4 x 6.6 x 7.3 x x x x x 7.4 x IV 1.1 x x x x x x 1.2 x 2.1 x 2.2 x 2.5 x 2.8 x 3.1 x x x 3.8 x Compliance (out of 113) 110 113 111 105 110 113 112 105 107 95 109 108 109 104 111 110 105 108 112 107 Source: Vriens (2011)
Non-compliance with Tabaksblat code in 2007 Chapters/ Best practices Ae go n Ah old Ak zo No be l AS M L Ho ld in g BA M g ro ep Co rio DS M Fu gr o He in ek en IN G g ro ep KP N Ph ili ps Ra nd sta d Re ed E lse vie r SB M O ffs ho re TN T To m To m Un ile ve r We re ld ha ve Wo lte rs -Kl uwe r II 1.1 x x x x x 2.1 x x x 2.2 x x x x 2.3 x x x x x x 2.4 2.5 x 2.6 x x x x x x x 2.7 x x x x x x x x x x x x x 2.8 x x x x 2.9 2.10 x x x 2.11 x x 2.13 3.1 x 3.2 x 3.3 x x 3.4 x III 2.1 x 2.2 x x x 2.3 x 3.4 x x x x 3.5 x x x x x 4.2 x 5.1 x x 5.3 x 5.4 x 5.11 x x x x x x x x 6.1 x 6.3 x x 6.4 x 6.6 x 7.3 x x x x x x x 7.4 x 8.1 x IV 1.1 x x x x x x 1.2 x x 2.1 x 2.2 x 2.8 x 3.1 x x x 3.8 x x Compliance (out of 113) 109 113 108 106 110 113 112 105 101 93 108 108 109 102 111 111 105 108 112 107
