1
CEO Compensation in the Netherlands, the Influence of
Anglo-American Board Members and Anglo-Anglo-American Cross-Listings.
PAUL VAN BEIJEREN (s1529099)
University of Groningen Faculty of Economics and Business
2
CEO Compensation in the Netherlands, the Influence of
Anglo-American Board Members and Anglo-Anglo-American Cross-Listings.
ABSTRACT
This study examines CEO compensation in the Netherlands, I am particular interested to see if Anglo-American board membership and/or Anglo-American cross-listing variables can help explain the level of CEO compensation. Next to this I test if there is a positive pay-performance relationship. The results of my regression analysis suggest that there is no positive pay-performance relationship in the Netherlands. In contrast to my expectations the variables Anglo-American board membership and cross-listing are not significantly positive, but the model with those variables does have more explanatory power about the variance in CEO compensation, which leads to the conclusion that the Anglo-American variables have an influence on CEO compensation.
JEL classification: G30; G34; J33; M52
3 Contents
I. Introduction ...4
II. Theoretical background ...6
A. classical payment theories ...6
B. Empirical research on CEO compensation ...8
C. Anglo-American board membership. ... 11
D. Anglo-American cross-listings ... 12
E. CEO compensation in the Netherlands. ... 13
III. Hypotheses and Methodology. ... 14
A. Research question and Hypotheses. ... 14
B. Methodology. ... 15
IV. Data description ... 17
A. Data collection ... 17 B. Variables ... 18 B.1 Dependent variable ... 18 B.2 Explanatory variables ... 18 B.3 Control variables ... 19 C. Sample Characteristics. ... 20 D. Data restrictions ... 21 V. Results ... 23 A. Performance model ... 23
B. Anglo-American board member model ... 24
C. The Anglo-American cross-listing model ... 24
D. The full model ... 25
VI. Conclusion... 27
VII. References ... 30
4 I. Introduction
CEO compensation has been a topic of discussion for many years, not only in the academic world but most certainly also in the eyes of the public and politicians. For instance there has been a lot of discussion in the Netherlands about the compensation of the CEO of Ahold, Anders Moberg in 2004, followed by the more recent and still ongoing debate about CEO compensation and bonuses of financial institutions. This is also a topic where the government tries to intervene, since history has shown that this issue is not self-regulated by the people involved. In the United Kingdom for instance the government has been a forerunner by giving investors an annual non-binding shareholder vote about CEO compensation since 2002, which is named “say on pay”. This subject is empirically tested by Ferri and Maber (2009) and they find no substantial evidence that there was a reduced growth rate for CEO compensation; however, CEO compensation seems to be more sensitive to poor performance. The topic of CEO compensation in the academic world is one of the most discussed topics in the economic and financial literature; studies have been going on this topic for over 80 years. The reason why this topic has drawn so much attention for all these years can be related to a few issues. Maybe the most important reason for all this attention can be found in the fact that, the reasons behind certain levels of CEO compensation are still not clear. This leads to new topics of research, where new variables are empirically tested to see if they have an influence on CEO compensation. This paper will discuss two variables and their influence as market factors on CEO compensation.
5
market factors, where the supply of managers that can deal with the more demanding Anglo-American markets is lower and demand for those CEOs will be higher, because of the globalization aspects. These market factors combined with higher risk of losing their jobs for those CEOs and spillover effects from the Anglo-American market will lead to higher CEO compensation.
While earlier studies did focus mainly on only one of those for mentioned theories, this study will combine both the classical theory of optimal contracting and the theory that focus on markets forces, especially the globalization issue. To test both these theories I will do an empirical test of a dataset which consists of all the listed companies in the Netherland in the year 2010. By testing both theories for only one dataset, it will be possible to identify the theory that fits best in the Dutch market. In this way, if for instance the optimal contracting approach is not proven to be the right theory to explain the Dutch situation, it can be possible to show why this is the case and if the market theory is better suited to explain CEO compensation in the Netherlands. The open economy in the Netherlands can be useful to see if CEOs are indeed paid more when the firms are connected with “the most prestigious financial markets” (Oxelheim et al., 1998).
In my opinion this study can contribute to a wider knowledge on CEO compensation, using a dual theory approach and the Netherlands as research country. The Netherlands has some specific corporate governance regulations, for instance the two-tier board structure. With these specific regulations in the Netherlands this study can help to provide additional proof to the influence of Anglo-American variables on CEO compensation. To my knowledge, there has only been one other study on this specific topic, which has focused on Norway and Sweden, and these countries have other regulations on corporate governance than the Netherlands. In this sense this study can be helpful to strengthen the proof on the theory, stated by Oxelheim and Randøy (2005), that market factors play an important role on the topic of CEO compensation in non-Anglo-American countries. Next to this, the study can help to provide a wider knowledge on the topic of CEO compensation and this could lead to better regulations and/or better understanding of CEO compensation in general. To test if performance variables or Anglo-American market variables determine CEO compensation in the Netherlands, I will try to answer the following research question in this paper; Is CEO compensation for Dutch listed
companies explained by the pay-performance relationship or are Anglo-American market factors a better way to explain CEO compensation?
The results in this study do indicate that CEO compensation and firm performance are not related, since the performance variables both have a negative coefficient in all four models. There is some proof that Anglo-American market factors do explain CEO compensation, since the model with those factors has more explanatory power. The dummy variables Anglo-American board membership and Anglo-American cross-listing do have positive signs, but in this study they are not significant.
6
the analysis will be discussed, next the main results of the analysis are presented and finally there will be a conclusion and discussion section.
II. Theoretical background
The topic of CEO compensation has been discussed and empirically tested a lot in the past and there have been done a lot of studies on how CEO compensation can be explained by different theories. In this paper the focus will be on American board membership and cross-listings on Anglo-American stock exchanges as explaining variables of CEO compensation. In the following section I will first discuss earlier research on the various topics explaining CEO compensation, followed by an overview of prior literature discussing topics that are specific for this paper, namely Anglo-American board membership and Anglo-American cross listings.
A. classical payment theories
Starting with the optimal contracting approach (Jensen and Meckling, 1976) who view CEO compensation as a way to minimizing conflict of interest between shareholders and managers, so compensation and performance of the company are related. The objective is to set pay contracts in such a way that managers have an incentive to maximize firm value and to reduce conflict of interests and moral hazard problems. The main assumption is that CEO compensation and firm performance are positively correlated. So, when the manager has a higher incentive, this will have a positive effect on the firm performance. Although the idea of this theory sounds simple, there are many problems and flaws, in real world situations, that result in contracts that are not optimal and do not give the proper incentives to the managers. As a result of this, the optimal contracting approach and the link between performance and pay have been a topic of interest for academics for decades.
7
problem. The adverse selection problem arise when the principal has less information than the agent (Darrough and Stoughton, 1986). Arrow (1984) refers to this as hidden actions and hidden information problems. So to avoid the asymmetric information problem and align interests of the manager with those of the shareholder the CEOs compensation should provide the right incentive for the manager. According to Holmström (1979, 1982) the CEOs salary should be based on the actions the CEO has taken, are those actions creating the most shareholder wealth? But even if there would be perfect information symmetry between the manager and shareholders, it would be very difficult to monitor the action taken by the managers, hence the CEOs compensation would be the only “communication tool” (Ross, 1973), so the CEOs compensation should be in such a way that it “forces” the CEO to take the right action. It becomes clear that it is essential for an organization, that CEO compensation is set in such a way that the interests are aligned, so the principal-agent problem is avoided. Different studies have provided information and proof that the principal-agent problem can be highly damaging for a company (Jensen, 1986, Morck et al., Bertrand and Mullainathan, 2003).
8
In addition to these two theories about CEO compensation where the former focuses on the classical principal agent problem and the latter on corporate governance variables that influence compensation. There are more recent theories that focus on the market driven aspect of CEO compensation, which consist of the supply side and a demand side for talented CEOs that can be explained by the human capital theory. Murphy and Zábojnïk (2004) discuss the issue of rise in CEO compensation in the light of the markets of CEOs. They question the statements of Bebchuck et al. (2002) that rent extraction of CEOs is the full story on rise in CEO compensation and that this is all related to weak corporate governance structures. First of all board of directors have become more independent over the years (Holstrom and Kaplan, 2001), so this would not explain the raise in CEO compensation. Secondly, in an earlier study Murphy and Zábojnïk (2003) discuss the difference in CEO compensation for internally and externally hired CEOs they find that external CEOs are paid 15.3% more on average. This is in contrast with the theory that the relationship between the board of directors and the CEO would lead to an increase in CEO compensation. Finally they find that it has become less important over the years to have a relationship with the board of directors to be hired as a CEO, in the seventies 15% of all CEO replacements where externally hired, which had increased to 26% in the nineties (Murphy and Zábojnïk 2004). Under their approach the rise in CEO compensation can be related to the market pricing of transferable managerial skills of a CEO, since those skills have become more important in the modern organization and the organizations are willing to pay for those skills. This is one of the studies that focus on a CEO market, in this study the rise in CEO compensation is both explained by the supply and the demand side of CEOs. The supply side of the human capital theory focus on the aspects of globalization, increased uncertainty and the more demanding aspect of managing large organizations. On the demand side the focus is on the aspect that companies are willing to pay more for more talented CEOs (Oxelheim and Randøy, 2005, Conyon 2006, Kaplan and Minton 2006, Gabaix and Landier, 2008).
B. Empirical research on CEO compensation
There have been done a lot of empirical studies on the topic of CEO compensation, at first the pay-performance relationship and the pay-size relationship have been the topics that generated the most attention. The topic has been studied as early as 1925 where Taussig and Barker (1925), state that… “They must be content with intelligent, solid, prudent executives, who in the main follow those in the
foremost rank. Might they not follow faster, be more on their toes, if there were a clear, immediate relation between their earnings and the net earnings of their concerns? A strong sense of duty and a pride in performance are not inconsistent with a lively interest in one's own advancement. Is it clear that our practice of a long-postponed and uncertain adjustment of salaries to results is better than the more direct relation that prevails on the Continent?” (p.51). This quote shows that, already in 1925,
9
have been the main focus of studies on CEO compensation, later followed by studies that focus on other theories that explain CEO compensation. Topics that received a lot of attention are for instance the board characteristics and corporate governance variables and in more recent years the focus on market aspects that drive CEO compensation. The literature on CEO compensation is so extensive, that there are quite a few papers that give a general overview of the literature on CEO compensation (Gomez-Mejia and Wiseman 1997, Barkema and Gomez-Mejia 1998, Murphy 1999, Tosi et al. 2000, Core et al. 2003, Friedman and Jenter, 2010).
First the studies that focus on the pay-performance relationship, in early studies the goal was to find performance measures that could explain CEO compensation, Roberts (1956), and Lewellen and Huntsman (1970) find some evidence that profits and sales are significantly related to CEO compensation. These results would suggest that there is a pay-performance relationship and although this is the case in the years of those studies, one should keep in mind that the dramatic rise in CEO compensation only started from the mid-seventies on (Friedman and Jenter, 2010). The next generation of studies on the pay-performance relationship have been using stock price performance as the performance variables. These studies starting in the mid-eighties, show inconsistent results on the pay-performance relationship. Murphy (1985) finds highly significant results that indicate that an increase of 2.1% in CEO compensation would lead to a 10% increase in stock prices. A major drawback of this study is indicated by Benston (1985), since this study defines CEO compensation as base salary plus cash bonus, the total income of the CEO is severely underestimated. CEOs have on average significant share and option holdings of the company, which are part of their compensation package, those should be included in the CEO compensation variable. Jensen and Murphy (1990) are the first who do an extensive study, including stock and option compensation, their results indicate a significant pay-performance relationship, but they state that their results are too weak to generalize. The general consensus in papers is that the relationship between pay and performance is weak (Haubrich 1991, Garen 1994, Joskow and Rose 1994, Kaplan 1994). Hall and Liebman (1998) propose another view, they say that because of the increase in stock option compensation, pay and performance are more closely linked, since better firm performance will result in a big wealth increase for the directors, the so called value at stake.
10
in shareholder return is linked to a 0.95% increase in CEO total compensation. This is in contrast with the Greenbury report of 1995 which proposes that CEO compensation and firm performance should be more closely linked. The inconsistent results do not stop here. In Australia Izan et al. (1998) do not find a significant positive relationship in their sample from 1987 to 1992, whereas Merhebi et al. (2006) do find a significant relationship between the CEO compensation and the performance of the company. All over the world there have been done studies on the pay-performance relationship indicating conflicting results. Of course this could be one of the main reasons this subject have been drawing so much attention in the literature.
Since a general consensus on the pay-performance relationship could not be reached, this could be the reason that academics have start focusing on other variables that could influence CEO compensation. The human capital theory focuses on specific knowledge and abilities of managers and the managerial power approach brought the attention to the corporate governance mechanisms in the organization. First of all Agarwal (1981) studies the influence of the human capital on CEO compensation, in his opinion earlier studies have the limitation that they only focus on firm size and performance as an explanation of CEO compensation. In his sample of 168 insurance companies he finds that job complexity, the employer's ability to pay, and the executives' human capital characteristics account for almost 80% of the variance in executive compensation. The human capital variable consists of three components, education level, field of study and work experience. Of these three, only work experience was significantly related to CEO compensation. When CEO tenure is seen as work experience of a CEO, Hill and Phan (1991) have results that conflict with those of Agarwal. In their study on 104 companies over the period 1977 to 1988 they find that CEO tenure negatively affects the pay-performance relationship. The longer the tenure of a CEO, the more he has influence on his compensation package. So also human capital could yield different results and although managerial skills can be valuable, a lot of CEOs have comparable backgrounds. If a CEO can distinguish himself from others, with a rare and inimitable skill, he could create value for a firm and be rewarded for this. International assignment experience is one of those skills. Having a CEO with this experience will result in better performance and CEO compensation and performance would be more related (Carpenter et al. 2001). To test the managerial power approach, a range of corporate governance variables have been tested to see what their impact on the level of CEO compensation is. Results of Core et al. (1999), prove that weaker corporate governance structures result in more agency problems and higher paid CEOs, although firm performance will be weaker. They find that CEO compensation decreases with the existence of an external block holder (Ownership structure) and with non-internal CEO shareholdings. In a more recent paper Ozkan (2007) finds this same relationship in a sample of 414 large UK firms for the period 2003 to 2004. She finds evidence that corporate governance mechanisms can explain a significant amount of variation in CEO compensation.
11
the relationship of Swedish and Norwegian firms with the Anglo-American capital markets, by cross-listings in those markets. The market for corporate control is there second topic, that is when a firm has Anglo-American board members. In their study they use a sample of 132 listed Norwegian firms and 120 Swedish listed firms. They assume that cross-listing on Anglo-American stock exchanges will reduce the availability of capable CEOs which in turn will lead to an increase in CEO compensation. They also suggest that international spillover effects will play a role in this increase, Anglo-American board membership is a reason for this, because of the common practice of higher CEO compensation in Anglo-American markets. They find a positive relationship for CEO compensation in firms with cross-listings in Anglo-American and for firms with Anglo-American board members, these results are highly significant. They argue that higher CEO compensation is justified because of increased risk of dismissal because of competing in a more demanding environment. Next to this they argue that there are less CEOs that are able to handle the more complex situations the companies are in because of the Anglo-American involvement, so the number of capable CEOs reduces leading to a higher compensation for CEOs.
C. Anglo-American board membership.
The reasons why Anglo-American board membership can result in higher CEO compensation have to be reviewed in some more detail. Oxelheim and Randøy (2005) explain that when a company include an Anglo-American board member, this can be considered by investors as a step to compliance with the board member his home country`s corporate governance culture. The inclusion of Anglo-American board members would be considered as compliance with one of the most demanding corporate governance cultures of the world. This is the result of different law systems, on one side the common law practiced in Anglo-American countries and on the other side the civil law, which is more common in Continental Europe, also in the Netherlands.
The difference between common law and civil law is explained by La Porta et al. (1998). In common law countries there are more shareholder rights, which will result in better shareholder protection. This combined with a more demanding corporate governance structure could result in a reduction of the agency-problem. A result of this better shareholder protection can be found in the ownership structure of a company. In civil law countries there are mainly large controlling shareholders, families, states or other companies. In contrast with this only in countries with the highest shareholder protection, like the US, companies are widely held (La Porta et al., 1999). A conclusion could be that better shareholder protection, will reduce the agency problem and there will be a stronger pay-performance relationship.
12
payments in the United States. This could mean that even though the Anglo-American board members are independent, they are more likely to approve higher CEO compensation. So one could argue that higher compensation is a result of spillover effects, caused by Anglo-American board members, who more easily approve high remuneration schemes for the CEO.
Oxelheim and Randøy (2005) proof that this positive relationship between Anglo-American board membership and CEO compensation does exist and is significant for their sample of Norwegian and Swedish firms. In a study of the influence of independent foreign board members in the United States by Masulis, Wang and Xie (2012). They find that because of poor board meeting attendance records, companies with foreign board members are more likely to approve higher CEO compensation and that this compensation is less linked to performance of the CEO.
D. Anglo-American cross-listings
To see what the result of an Anglo-American cross-listing can be on CEO pay, it is important to know what reasons companies can have to cross-list. Pagano et al. (2002) find evidence that an increasing number of European firms do cross-list, mainly in the US. When companies have a more global market, it can be beneficial for them to get access to foreign capital markets. Another trend that can be seen is the difference between companies that cross-list in Europe and companies that cross-list in the US. Companies that cross-list in Europe do not see significant growth, companies that cross-list in the US are mainly young, export-oriented and experience high equity-based growth (Pagano et al., 2002). There are various reasons to cross-list, a nice overview on this reasons are also given by Pagano et al., they name and test a total of ten reasons, I will focus on the reasons that could be related to the pay-performance relationship. First companies that cross-list have more possibilities to raise capital for investment which in turn could result in higher company performance. Secondly a reason could be to strengthen the company’s output market; this could also result in a higher performance for the company. Thirdly a broader and new shareholders base, this could lead to more strict corporate governance since the company is more widely hold. The last reason is the most important reason if we consider our paper, companies will be forced to commit to disclosure and more strict corporate governance standards. In line with including Anglo-American board members in the company, this could lead to a reduction in the agency problem and to a better pay-performance relationship.
13
between Anglo-American cross-listing and CEO compensation does exist and is significant for their sample of Norwegian and Swedish firms.
E. CEO compensation in the Netherlands.
In the Dutch situation, because of the global aspect of the Dutch industries, it will be interesting to see if Anglo-American board membership and Anglo-American cross-listings will have an influence on CEO compensation. Like in other countries also CEO compensation in the Netherlands have been a topic that created lots of interest, again the pay-performance relationship yields conflicting results. In a recent study, on 107 Dutch listed companies, van Ees et al. (2007) find a positive significant pay-performance relationship, although this relationship was not the case for all remuneration components. A negative relationship between performance and pay is shown by Duffhues and Kabir (2008) in their study on top management team compensation and firm performance. So also in the Netherlands it can be said that the optimal contracting approach is not able to explain the CEO compensation.
Since the Netherlands is a civil law country it has been argued that there is weak shareholder protection, in fact in the study of La Porta et al. (1998) the Netherlands is seen as one of the countries with the weakest shareholder protection. Next to this the Netherlands is not seen as a country with really high CEO remuneration, certainly not when compared with remuneration in Anglo-American countries, for instance the US. This can indicate that when Anglo-American board members are assigned to the board or when Dutch companies try to compete for capital on the most demanding capital markets, by cross-listing on Anglo-American stock exchanges, this could lead to higher CEO pay and a better relation between pay and performance.
14
two-tier (Jungmann, 2006). In Sweden companies are allowed to have one executive on the supervisory board which is in most cases the CEO of the company. In Norway the board of directors has both a supervisory function and a management function, so this would be a one-tier board. Since the Anglo-American countries have one-tier board structures, the addition of Anglo-American board members or the listing on Anglo-American markets could have an impact on the remuneration policy of the companies.
The literature indicates that becoming more global as a company will result in less available CEOs who are capable of managing the company, as a result of this CEO compensation will be higher. A study of Sanders and Carpenter (1998) has resulted in the evidence that there is a positive relationship between the internationalization of the product and service market and CEO compensation. Most of the Dutch economy is dependent on the export market, as shown in the last subsection, when exports are important a company could benefit from cross-listing. These aspects of the Dutch economy could yield results comparable by the results of the Oxelheim and Randøy study of 2005 where there was indeed an impact of Anglo-American board members and Anglo-American cross-listing on pay.
III. Hypotheses and Methodology.
In this section I will first explain the research question of the paper, followed by the hypotheses I will test to answer my research question. Next I will explain the methodology I will use to test the hypotheses.
A. Research question and Hypotheses.
The goal of this paper is to examine whether CEO compensation for Dutch listed companies can be explained by the pay-performance relationship or is more related to specific market variables. I will test four hypotheses to do so. These hypotheses are closely related to the study of Oxelheim and Randøy (2005). First of all the research question that I want to answer is:
Is CEO compensation for Dutch listed companies explained by the pay-performance relationship or are Anglo-American market factors a better way to explain CEO compensation?
15
protection, we can assume that there will not be a positive pay-performance relationship and the manager can have some power on his remuneration. As a result of this my first hypothesis will be:
C1: There is no positive and significant relationship between firm performance and CEO compensation in the Netherlands.
Secondly I will test the influence of having an Anglo-American board member on CEO compensation. The literature makes clear that there indeed can be an influence of having Anglo-American board members, either by being more focused on the performance of the company or by spillover effects. The only study that has included this variable so far, (Oxelheim and Randøy, 2005) found a significant relationship between Anglo-American board membership and CEO compensation. Therefore my second hypothesis will be the following:
C2: There is a positive and significant relationship between Anglo-American board membership and CEO compensation in the Netherlands.
My third hypothesis will be related to cross-listing on an Anglo-American stock market, in the previous section I have explained the various reasons a company can have to cross-list, especially on Anglo-American markets. These reasons indicate that cross-listing is indeed one of the factors that explain CEO compensation, this relationship will most likely be positive. This positive and significant relationship is also proven by Oxelheim and Randøy (2005). So my third hypothesis will be:
C3: There is a positive and significant relationship between Anglo-American cross-listing and CEO compensation in the Netherlands.
In my fourth and last hypothesis I will test whether the inclusion of the variables Anglo-American board membership and Anglo-Anglo-American cross-listing will lead to higher explanatory power of our regression model. The results of Oxelheim and Randøy (2005) study, indicates that for their sample, including the variable Anglo-American board membership lead to a 6.3% increase in explanatory power of the model and including the Anglo-American board cross-listing variable lead to a 5.8% increase in the explanatory power of the model. Considering the conflicting results about CEO pay it would be interesting to see if these variables are an important factor to explain CEO pay. Therefore my fourth hypothesis will be:
C4: The inclusion of Anglo-American board membership and Anglo-American cross-listing as explanatory variables in the model will increase the explanatory power of the regression model. B. Methodology.
The method I will use to test the hypotheses developed in the last subsection will be in line with earlier studies on the subject of CEO compensation, for example (Oxelheim and Randøy 2005, Ozkan 2007). I will use the Ordinary least-squares (OLS) regression, which will be run in the Eviews program.
16
(1) Where the dependent variable is the natural logarithm of total CEO compensation, α is the intercept and the β`s indicate de coefficients of the independent and control variables. The first two independent variables are performance measures, RI is the stock return over the year and ROE is the return on equity in the previous year, this is an accounting performance measure. The following five independent variables are control variables, the variable value is measured by Tobin’s Q, size is the natural logarithm of the company’s total sales. The next three variables are corporate governance measures, the variable board is the natural logarithm of the board size of the company. Nonexc. is the percentage non-executive board members with respect to the total board size, in this case because of the two-tier board structure in the Netherlands, the percentage non-executives are the members of the supervisory board as a percentage of the management board plus the supervisory board. Tenure is the natural logarithm of the total tenure of the CEO at the end of 2010. The last term, ε is the error term of the function. The goal of the first hypothesis is to test if the performance variables are not significantly positive related to total CEO compensation. In this way I can see if in the Dutch situation the contracts are related to optimal contracting theory. Following the theory, I expect that this will not be the case and that there will not be a positive pay-performance relationship. So C1 will be supported when β1 and β2 are not significantly positive.
The second hypothesis will be tested using basically the same equation as equation 1, including one extra independent variable, this will be the board membership model, tested by the following equation;
(2) As explained the equation and the variables are the same as equation 1, with the exception of the variable dummy B.M., this is a dummy variable with respect to Anglo-American board members, this variable will take a value of 1 if there is one or more Anglo-American board members in a company and a value of 0 if there is not an Anglo-American board member in the company. Following the literature I expect that there will be a positive relationship between Anglo-American board members and pay, so C2 is supported when β8 is significantly positive.
To test hypothesis 3 I will again refer to equation 1, as in equation 2 the only difference with that equation is the inclusion of one extra independent variable, this will be the cross-listing model, which will be tested by the following equation;
17
The difference in this equation with respect to equation 2 is that the dummy variable of American board membership is replaced with dummy CL., this is a dummy variable of Anglo-American cross-listing. This variable will take a value of 1 when a company does cross-list on an American stock exchange, and a value of 0 when a company does not cross-list on an Anglo-American stock exchange. I explained in the last subsection, that I expect that β8 will take a significant positive value, and if this is the case I will accept C3, which will mean that in the Netherlands there is a positive relationship between Anglo-American cross-listing and CEO pay.
In addition to equation 2 and 3, I will use an equation that will include both the dummy variables, dummy B.M. and dummy CL.. Equation 4 will be the full model, this model will be tested by the following equation;
(4) This equation includes all the independent, control and dummy variables to explain CEO pay. I will use this equation to test the first three hypotheses again, to see if the results change when all variables are included. Next to this I will test C4 to see if equation 4 will have more explanatory power than equation 1 with regard to CEO pay. To test this I will compare the adjusted R-squared of equation 1 and 4, if the adjusted R-squared is higher in equation 4 I will accept C4. The R-squared will indicate how well the equation explains the real data, I will use the adjusted R-squared since this statistic, adjust the value of R-squared for the number of variables used. This is necessary since the inclusion will lead to an increase in the normal R-square anyhow, the adjusted R-squared will indicate if the inclusion of the new variables increases the explanatory power of the model, so is the model improved by the new variable.
IV. Data description
The upcoming section will give an overview of the data used in the analysis, I will first explain the data collection method, followed by an overview of the variables. Next I will give a description of the data and finally the restrictions of the data.
A. Data collection
18
The data was partially collected using the DataStream Worldscope software, to find the variables related to the financial data of the company. DataStream provided the data for the performance measures, return on stocks and return on equity, and the data for the control variables Tobin`s Q and company size. The collection of the data regarding to the CEO compensation and the corporate governance variables was more challenging. Companies are forced to comply with the corporate governance code, because of this the CEO compensation could be found in the corporate annual reports of 2010. The corporate governance variables, board size and percentage non-executives and CEO tenure as a control for individual CEO characteristics, where found using a combination of the annual reports and the corporate websites. The data for the explanatory dummy variables, Anglo-American board membership and Anglo-Anglo-American cross-listing, was found using corporate websites and the websites of the Anglo-American stock exchanges.
B. Variables
This subsection will list and explain the variables used in my regression models. It will explain how the variables are measured and the relevance of these variables for the research. The variables used will be separated in the categories, dependent variable, explanatory variables and the control variables.
B.1 Dependent variable
The dependent variable in this study will be the total CEO compensation, which will be referred to as CEO pay. Total CEO compensation is build up from four components; base cash salary, a cash bonus component, a long term incentive plan and other benefits. The first component needs no explanation, the cash bonus is related to firm performance and generally measured by a selection of accounting measures. The long term incentive plan is an important tool to align management and shareholders’ interests, it consists of company stock and stock options. The latter will not be part of the CEO compensation in this study, since there was not enough data available to value these options using the Black-and-Scholes formula. Excluding stock options from the study should not have a considerably effect on my results since Fernandes et al. (2011) show in their sample option compensation in the Netherlands is low compared to other countries, only 5% of total compensation in 2006. Other benefits for a CEO can for example be pension cost covered by the company, loans, living expenses or golden parachutes. In the Netherlands pension benefits is the most common form of other benefits that is included in CEO compensation. In my models total CEO compensation is measured by the natural logarithm of the total CEO compensation, this is done since CEO compensation is heavily skewed and this will lead to non-normality of the sample.
B.2 Explanatory variables
19
To make this data useful the RI of 2010 is divided by the RI of 2009, next one is deducted to create the increase or decrease in the stock price over the year 2010, this will give the stock returns including dividends. This measure of performance is useful since CEO compensation should be in line with the results of the owners. The second measure of performance is the accounting measure return on equity (ROE), the data of this measure used in this study was from the previous year (2009) since in earlier literature (for example Jensen and Murphy, 1990) it is suggested that CEO compensation is related to one year lagged accounting measures. Including two different measures of performance will create the opportunity to see if there is a pay-performance relationship at all.
The two other explanatory variables are the dummy variables, Anglo-American board membership and Anglo-American cross-listing. First Anglo-American board membership, variable is measured as follows, I have assigned this variable a value of 1 when there are Anglo-American board members and a value of 0 when the company did not have Anglo-American board members. I have looked at both executives and non-executives and their nationality, a board member is considered Anglo-American when he or she has one of the following nationalities, American, Australian, British, Canadian or Irish. In my sample of 92 companies, a total of 34 companies had at least one Anglo-American board member or almost 37%. This indicates that in the Netherlands it is quite common to have an international board, which could be related to the highly internationalized economy in the Netherlands. The second dummy variable is the Anglo-American cross-listing variable, again this variable was given a value of 1 when a company did cross-list in either the United Kingdom or the United States and a value of 0 if a company did not cross-list in one of these countries. From the total sample of 92 companies, a total of 8 companies did cross-list, or 8.7 %, which is considerably smaller than the other dummy variable. This could be caused by all the rules a company has to face when it cross-lists.
B.3 Control variables
20
CEO compensation (Baumol 1967, Oxelheim and Randøy 2005). Since company sales are usually heavily skewed I will use the natural logarithm of the total sales to deal with normality issues.
I will control for two corporate governance characteristics. First I will control for total board size, measured by the amount of executives (management board) and non-executives (supervisory board) that have a seat on the company’s board. This variable is also commonly used in earlier studies, for example Core et al. (1999) and Oxelheim and Randøy (2005). As for firm size, the natural logarithm is taken from the total companies board members, again to deal with normality problems. Secondly I will use the percentage non-executives on the board of the company as a control variable, this is the percentage of board members that are on the supervisory board. This is a proxy for the independency of board members, prior studies have shown that firms with more independent boards are more likely to pay the CEO less (Core et al., 1999).
The last control variable I will use is related to individual CEO characteristics. To control for these characteristics I will use CEO tenure as a proxy. CEO tenure will be measured as the amount of years a CEO operates in his current position for the company. CEO tenure will have an influence on CEO compensation and the pay-performance relationship as studied earlier by Hill and Phan (1991). The natural logarithm of CEO tenure will be taken to deal with non-normality issues.
C. Sample Characteristics.
In this part I will present the descriptive statistics of the sample I used in the study, the descriptive statistics can be found in table 1. From the table we can see that for the variables were I used the natural logarithm, the spread was considerably reduced (compared with appendix A where the un adjusted values of these variables are presented), because of this the variables are less skewed than they were without taken the natural logarithm. For total compensation we see that there is still considerable variance in the sample, this is because of the pay that vary widely in the various industries and companies. In appendix A it can be seen that in real numbers the average total CEO pay is 997,734.20 in Euros, which is considerably higher than the median of 664.000 Euros, this is caused by outliers both on the upside and downside, where the former plays the biggest role with a maximum CEO compensation of 5.363.000 Euros.
21
where appointed in 2010 and CEOs that have been in the company in the excess of 20 years. The two dummy variables can only take the value of 0 or 1, as described in the last subsection a total of 34 companies have one or more Anglo-American board members and 8 companies do cross-list in an Anglo-American country.
Table I Descriptive Statistics
This table presents the descriptive statistics of the used sample. The mean, median, maximum, minimum, standard deviation and the size of the sample are listed for all the variables used in the study. Total CEO pay are all remuneration components excluding stock options. Stock performance is measured by the return index (RI) from DataStream (RIt / RIt-1) – 1. ROE is the return on equity in the previous year. Tobin`s Q is calculated by is calculated as the sum of the book value of assets, plus the market value of common stock, minus book value of common stock and this value is divided by the total assets. Firm size is
measured by total sales. Board size is the total number of board members, and percentage non-executives is number of non-executives divided by the board size. CEO tenure is the total number of years a CEO is
appointed. Anglo-American board members and cross-listing are dummies.
Variables Mean Median Maximum Minimum
Std.
Deviation N
Total CEO pay (ln) 13.484 13.406 15.495 11.661 0.792 92
Performance, stocks 3.966 0.442 323.674 -0.664 33.702 92
ROE (2009) -0.144 0.095 0.599 -10.798 1.222 92
Tobin`s Q 1.798 1.612 7.006 0.730 0.805 92
Firm size (sales LN) 19.940 20.187 26.354 13.259 2.484 92
Board size (LN) 1.977 1.946 2.773 1.099 0.376 92
% non-executives 0.658 0.667 0.889 0.400 0.093 92
Ceo-tenure (LN) 1.461 1.658 3.207 -1.609 1.008 92
Anglo-American Board members
(dummy) 0.370 0 1 0 0.485 92
Anglo-American cross listing (dummy) 0.087 0 1 0 0.283 92
D. Data restrictions
To justify the use of the OLS regression, the data used has to comply with some rules. One of these rules is already discussed, namely the sample has to be normally distributed, to do so from a few variables the natural logarithm is taken as can be seen in table I. This procedure resulted in normal distribution of the variables CEO pay, company size and board size. CEO tenure was still not normally distributed, which is a result of the relatively small sample.
22 Table II
Pearson correlation test.
In this table the results of the Pearson correlation test are presented, *indicates significance on the 5% level and ** indicates significance on the 1% level. The table find relatively high and significant correlations between the independent variables, board size and company size of 0.64 and between the variable board size and the
dummy variable Anglo-American board membership of 0.561.
Variables 1 2 3 4 5 6 7 8 9
(1) Total compensation (in 1000)
(2) Performance, stocks -0.088
(3) ROE (2009) 0.158 0.024
(4) Tobin`s Q 0.153 -0.044 -0.057
(5) Firm size (sales LN) 0.809** -0.044 0.289** -0.179
(6) Board size (LN) 0.723** -0.100 0.022 0.148 0.640**
(7) % non-executives 0.075 -0.066 0.066 0.041 0.083 -0.039
(8) Ceo-tenure (LN) 0.112 -0.004 0.092 -0.126 0.138 -0.013 -0.024
(9) Anglo-American Board members
(dummy) 0.410** -0.080 -0.135 0.134 0.242* 0.561** 0.083 0.059
23
also relatively highly correlated with Anglo-American board membership, the correlation is 0.561 this result is also significant. It makes sense that larger boards have a higher chance of having an Anglo-American board member(s). Although this correlation is high it should not be a problem, since usually there are higher cutoff points that indicate multicollinearity.
A third problem that we could face in the sample is the presence of heteroscedasticity, I deal with this issue by using White`s Standard Errors in the OLS regression analysis. These are robust standard errors that will make sure that there is corrected for heteroscedasticity. Finally there the Durbin-Watson statistic does not indicate any sign of autocorrelation of the standard errors.
V. Results
In this section I will discuss the results of the OLS estimation of the models discussed in section III. The results will be presented in table III. This section will discuss separately the results of each of the four models constructed, based on the four hypotheses.
A. Performance model
24
The adjusted R-squared of this model indicates that the model can explain 76.3% of the variation in CEO pay, which is high.
B. Anglo-American board member model
This second model is basically the same as the previous model, with one exception. The dummy variable Anglo-American board membership is included in the model as an explanatory variable. This OLS regression model is used to test the second hypothesis C2. The hypothesis is supported when there is a significant and positive relationship between Anglo-American board membership and CEO pay. The result for the variable can be found at the bottom of column two of the variables in table III, we see that, although the sign of the beta is positive the result is not significant. It falls short of the 10% significance level. The p-value is 0.144. This result means that the relationship between Anglo-American board membership and CEO pay cannot be generalized and therefore hypothesis C2 is not supported.
The results of the other variables show a lot of resemblance with the performance model. The betas for the performance measures are still negative and only the stock performance measure is significant. The other control variables indicate that there is a change in the significance level for the variable board size, this variable is still positive but in this model only on the 10% level, again the intercept is highly significant on the 1% level. R-squared indicates that there is an increase of explanatory power of the model of 0.6% when I include the Anglo-American board membership variable, when adjusted for the increase in independent variables by using the adjusted R-squared we see that the real increase is 0.4%.
C. The Anglo-American cross-listing model
The third model is also basically the same as the performance model, but instead of including Anglo-American board membership as a dummy variable, the dummy variable Anglo-Anglo-American cross-listing is included in the model. The results can be found in table III, in the third column. This regression model is used to test my third hypothesis C3. Is there a positive and significant relationship between Anglo-American cross-listing and CEO compensation. The results of the regression model do indicate that the beta of the Anglo-American cross-listing variable is positive, but as in the previous model this result is not significant, with a p-value of 0.114. This result suggest that hypothesis C3 is not supported, and indicates that being cross-listed on an Anglo-American stock exchange, does not result in higher CEO compensation in the Netherlands.
25
D. The full model
The full model and final model of this study is the model, where I include both dummy variables in the regression estimation of the performance model. Again the results can be found in table III, the results of this model are used to see if the first three hypothesis are supported or not, when all the variables are used in the model. Next to this hypothesis C4 will be tested to see if the inclusion of the dummy variables lead to a model that has more explanatory power of CEO pay. First we look at the performance variables, the stock performance variable still has a small negative beta of -0.0005 and this variable is still significant on the 1% level. The results of the second performance variable ROE has a beta of -0.023, so also the second performance variable shows a negative sign, but this result is not significant. The results of the performance variables indicate as in the performance model that in the Netherlands there is no pay-performance relationship.
The dummy variable Anglo-American board membership still has a positive beta, this time of 0.137, but again this variable is not significant. The p-value even reduced compared to the Anglo-American board membership model to a value of 0.239. So the results of the full model are not in line with my expectations, it still seems that in the Netherlands there is no significant proof for a positive relationship between, the CEO pay and companies that have Anglo-American board members on their supervisory boards. The second dummy variable, Anglo-American cross-listing yields comparable results to the Anglo-American board membership dummy. Again there is a positive beta, in this case of 0.191, but again this result is not significant, the p-value is 0.261, so the results cannot be generalized. The results of both my dummy variables, conflict with the results of the other study on these variables and their effect on CEO pay. Oxelheim and Randøy (2005) did find positive and significant results for these variables in their study on CEO pay in Norway and Sweden. This could mean that such a relationship in the Netherlands does not exist.
When we compare the beta values of the control values in the performance model and the full model, what stands out is that for most of the variables there is not a big difference in the beta values. Only the positive beta for board size is smaller in the full model, from 0.495 to 0.357 and is only significant on the 10% level instead of the 5% level. CEO tenure is another control variable that goes through a change, in the performance model the sign of the beta was positive, in the full model the sign has changed to a negative sign, although in both models these results are not positive.
26 Table III
Results of the regression models
In the table the results of the four OLS regression models are given. The equation’s of these models can be found in the methodology section. The dependent variable is the natural logarithm of CEO compensation. As can be seen in the table the dummy variables are not included in model 1, they are both included in model 4. Anglo-American board members are included in model 2 and – cross-listings in model 3. Significance levels are indicated by *<10% **<5% and ***<1%.
Beta values are reported with t-statistics in parentheses.
Independent Variables Dependent variable: CEO compensation(LN)
Performance model (1) Board members model (2) Cross-listing model (3) Full model (4) Performance, stocks -0.0005*** (-2.841) -0.0004*** (-2.818) -0.0005*** (-2.973) -0.0005*** (-2.892) ROE (2009) -0.028 (-1.547) -0.021 (-1.037) -0.030 (-1.596) -0.023 (-1.184) Tobin`s Q 0.243*** (3.967) 0.243*** (4.152) 0.210*** (3.254) 0.216*** (3.490)
Firm size (Total sales LN) 0.225***
(6.670) 0.231*** (7.020) 0.218*** (6.308) 0.225*** (6.564) Board size (LN) 0.495** (2.414) 0.349* (1.742) 0.476** (2.350) 0.357* (1.798) % non-excecutives 0.154 (0.360) 0.038 (0.088) 0.082 (0.188) -0.001 (-0.003) Ceo-tenure (LN) 0.042 (1.181) 0.033 (0.887) 0.039 (1.126) 0.033 (0.888)
Anglo-American Board members (dummy)
0.162 (1.473)
0.137 (1.185)
Anglo-American crosslisting (dummy)
0.237 (1.599) 0.191 (1.132) Intercept 7.417*** (14.372) 7.609*** (13.997) 7.681*** (12.982) 7.792*** (12.955) Number of firms 92 92 92 92 R-squared 0.781 0.787 0.787 0.791 Change in R-squared - 0.006 0.006 0.010 Adjusted R-squared 0.763 0.767 0.766 0.768
27
VI. Conclusion
In this section I will discuss how the results, presented in the last section, helps to answer the research question and how we have to interpret these results. To do so I will first discuss the four hypotheses, to see if the results are in line with my predictions. Finally I will discuss the limitations of this study and give some ideas about future research on the topic.
To answer the first part of the research question hypothesis C1 was constructed; C1: There is
no positive and significant relationship between firm performance and CEO compensation in the Netherlands. This hypothesis helps to explain if the CEO compensation in the Netherlands is
performance based, as predicted there does not exist a positive pay-performance relationship. Two performance measures are used, first stock performance, the results are conclusive in all models and indicate a negative relationship between stock performance and CEO compensation. The next performance measure I used was return on equity (ROE), this performance measure also had a negative beta in all models, but was not significant. These results suggest that optimal contracting theory is not able to explain CEO compensation in the Netherlands. These results are in line with earlier findings, not only in the Netherlands, but all over the world, this finding adds to the general consensus that the pay-performance relationship is hard to proof. The hypothesis leads to the first conclusion of this thesis, that CEO compensation in the Netherlands cannot be explained by the pay-performance relationship.
The next three hypotheses are related to the second part of the research question, are the Anglo-American market factors a better way to explain the CEO compensation in the Netherlands. The second hypothesis addresses the inclusion of one or more Anglo-American board members to the board of the firm. C2: There is a positive and significant relationship between Anglo-American board
membership and CEO compensation in the Netherlands. This hypothesis is both tested in the model
that only includes the Anglo-American board membership dummy variable, as well in the full model that also includes the dummy variable for the Anglo-American cross-listing. Although in both models the beta of the dummy is positive, the results are not significant, so they are not in line with the predictions and the earlier research of Oxelheim and Randøy (2005), so the second hypothesis C2 is not supported.. Before I comment on this, let’s first look at the model where the Anglo-American cross-listing dummy is included. The third hypothesis is tested to see of this variable is one of the Anglo-American market factors that influences CEO pay. C3: There is a positive and significant
relationship between Anglo-American cross-listing and CEO compensation in the Netherlands. Again
28
explain why hypotheses C2 and C3 are not supported we have to look at earlier studies on CEO compensation, as explained in section II there has been done a lot of research on the subject of CEO compensation and all kinds of variables have been tested. All this prior research yields to one main important conclusion and that is that the topic of CEO compensation has seen a lot of mixed results. The topic discussed in this paper has only been studied by Oxelheim and Randøy and they did find the relationship between the dummies and CEO compensation in their sample of companies listed in Norway and Sweden. A reason that I do not find the same results can lay in the fact that, although Norway, Sweden and the Netherlands are all three civil law countries, there is a difference in the civil law in the Netherlands and the civil law in Scandinavian countries (La Porta et al., 1998). This could indicate that Dutch corporate governance is less influenced by the inclusion of Anglo-American board members or listing on an Anglo-American stock exchange. A second reason for these results could be the relatively small sample size that is used, which brings some problems with non-normality of the sample, this could influence the OLS regression estimation, although it was assumed that this should not be a problem. Another problem with the sample could be that the dummy variables are small percentages of the total sample, especially the dummy for Anglo-American cross-listing. To solve this and still use the Netherlands as a research object it could be possible to include other countries that also have the same origin of civil law code, in the Dutch situation this would be the French origin civil law, Belgium and France are examples of countries from Europe that also use this law system (La Porta et al. 1998). A final explanation of these results can be, that the relatively high correlation between board size and firm size and between board size and Anglo-American board membership. These correlations are a frequently observed, since there is a relationship between those independent variables, usually this should not be a problem, but it has the potential to lead to biased results.
The fourth hypothesis, C4: The inclusion of American board membership and
Anglo-American cross-listing as explanatory variables in the model will increase the explanatory power of the regression model, is tested by comparing the adjusted R-squared of the four models. The adjusted
29
variables have a higher explanatory power. The increase in explanatory power is higher in the study of Oxelheim and Randøy, but they do not use the adjusted R-squared, but the normal R-squared instead which has the potential of inflated results. Although the fourth hypothesis is supported, there are some limitations on this kind of analysis, since the first models adjusted R-squared is already considerably high, and the increase in adjusted R-square is small this could mean that these results are not complete. With the results I should be able to answer the research question, is CEO compensation for
Dutch listed companies explained by the pay-performance relationship or are Anglo-American market factors a better way to explain CEO compensation? The first part of this question is easy to answer, I
do not find evidence that there is a positive pay-performance relationship, since the performance variables do not indicated that there exist such a relationship for Dutch listed companies. The second part is harder to answer since the results are conflicting, on one side both Anglo-American dummy variables are not significant, which would indicate that there is also no proven relationship between CEO supply and demand markets or spillover and risk effects. If we look at the explanatory power of the model, we see that although the explanatory power of the performance model is already high, this number even improves when the dummy variables are included in the equation. This result together with the positive sign of the betas of Anglo-American board membership and the possible explanations for the insignificant results, lead me to the conclusion that Anglo-American market factors are a better way to explain CEO compensation. This conclusion would of course need future research, possibly by increasing the sample as suggested before.
Looking at the limitations of this study, first the earlier discussed, small sample. Having such a small sample can give problems with the OLS estimation, since the Netherlands is a small country, with a limited number of companies, a solution could be to include comparable countries as discussed before. Another solution for this problem could be, increasing the research period, since a number of variables are not available in digital form, this would need a long and intensive data collection procedure, but it could be an interesting topic for future research.
30
other countries, it could be interesting to see if there is an effect of culture and the opinion of a nation’s society on CEO compensation, this would require a measure for these effects and this could be related to the spillover effects. If we for instance look at the cultural difference between the Netherlands and the US, we see that in the US there is a culture called the “American Dream”, the society does not see people that get paid a lot as something strange, but as someone who has made it and lives the “American Dream”, although this has changed a bit because of the crisis. In the Netherlands it is seen as something strange when people earn a lot, the public opinion is more based on, they do not deserve it. These are just a few examples for future research, but I think it is clear that there will still be a focus on CEO compensation for many years.
VII. References
Agarwal, Naresh C., 1981, Determinants of Executive Compensation, Industrial Relations 21, 36-46. Arrow, Kenneth J., 1984, The Economics of Agency. Stanford University IMSSS Technical Report No.
451
Barkema, Harry G., and Luis R. Gomez-Mejia, 1998, Managerial compensation and firm performance: a general research framework, Academy of Management Journal 41, 135-145.
Baumol, William J., 1967, Business Behavior, Value and Growth, Harcourt, Brace and World, (New York: Macmillan)
Bebchuk, Lucian A., Jesse M. Fried, and David I. Walker, 2002, Managerial power and rent extraction in the design of executive compensation, The University of Chicago Law Review 69, 751-846.
Bebchuk, Lucian A., and Jesse M. Fried, 2003, Executive compensation as an agency problem,
Journal of Economic Perspectives 17, 71-92.
Benston George J., 1985, The self‐serving management hypothesis: some evidence, Journal of
Accounting and Economics 7, 67‐83.
Berle, Adolf A., and Gardiner C. Means, 1932, The Modern Corporation and Private Property, (Macmillan, New York).
Bertrand, Marianne, and Sendhil Mullainathan, 2001, Enjoying the Quiet Life? Corporate Governance and Managerial Preferences, Journal of Political Economy 111, 1043-1075.
31
Conyon Martin J. 2006. “Executive Compensation and Incentives”, Academy of Management
Perspectives 20, 25-44.
Core, John E., Robert W. Holthausen, and David F. Larcker, 1999, Corporate governance, chief executive officer compensation, and firm performance, Journal of Financial Economics 51, 371-406. Core, John E., Wayne R. Guay, and David F. Larcker, 2003, Executive equity compensation and incentives: a survey, Economic Policy Review 9, 27-50.
Darrough, Masako N., and Neal M. Stoughton, 1986, Moral Hazard and Adverse Selection: The Question of Financial Structure, The Journal of Finance 41, 501-513
Duffhues, Pieter J.W., and Rezaul Kabir, 2008, Is the pay-performance relationship always positive? Evidence for the Netherlands, Journal of Multinational Financial Management 18, 45-60.
Fernandes, Nuno, Miguel A. Ferreira, Pedro Matos, and Kevin J. Murphy, 2011, The Pay Divide: (Why) Are U.S. Top Executives Paid More?, Working Paper Series.
Ferri, Fabrizio, and David Maber, 2009. Say On Pay Votes and CEO Compensation: Evidence from the UK, Working Paper, Harvard Business School.
Friedman, Carola, and Dirk Jenter, 2010, CEO compensation, Annual Journal of Financial Economics 2, 75-102.
Gabaix, Xavier and Augustin Landier, 2008, Why Has CEO Pay Increased So Much?, The Quarterly
Journal of Economics 123, 49-100.
Garen, John. 1994. Executive Compensation and Principal-Agent Theory, Journal of Political
Economy 102, 1175–99.
Gomez-Mejia, Luis R., and Robert M. Wiseman, 1997, Reframing executive compensation: an assessment and an outlook, Journal of Management 23, 291-374.
Gomez-Mejia, Luis R., and Robert M. Wiseman, 2007, Does agency theory have universal relevance? A reply to Lubatkin, Lane, Collin, and Very, Journal of Organizational Behaviour 28, 81-88.
Greenbury, S. R., 1995, Directors’ Remuneration ( London : Gee & Co. Ltd.).
Hall, Brian J., and Jeffrey B. Liebman, 1998, Are CEO’s really paid like bureaucrats?, Quarterly
Journal of Economics 113, 653-691.
Haubrich, Joseph, G., 1991, Risk Aversion, Performance Pay, and the Principal-Agent Problem,
