CEO compensation in Continental-European
countries: The Anglo-American Influence
By Marco Hilbrink
March, 2013
University of Groningen
Faculty of Economics and Business
ABSTRACT
This paper examines the impact of Anglo-American influences on the level and structure of CEO compensation within Continental-European companies, employing a hand-collected unbalanced panel data set with 336 firm-year observations of French, German and Dutch non-financial companies for the period 2005-2007. Different from previous studies on this subject, this paper takes, among others, into account the presence and role of an Anglo-American director within the remuneration committee of a Continental-European company as measure of Anglo-American influence. The empirical results indicate that there is a positive relationship between cross-listing activities of Continental-European companies and the level of CEO compensation. This finding suggest that Continental-European companies are willing to pay a higher compensation and risk premium for CEOs that possess the skills and are prepared to deal with the complex Anglo-American corporate governance environment. Moreover, the results reveal that not the presence, but the role as chairman of an Anglo-American director within the remuneration committee of a Continental-European company is associated with CEO compensation. However, this relationship is weak and not generalizable. Relating this outcome with results of previous studies, that found a significant relationship between Anglo-American board membership and CEO compensation, one may implicate that the remuneration committee has more an advising role than it has a decisive vote on CEO compensation.
Key words: Corporate governance, CEO compensation, Anglo-American influences. JEL-Classification: G30; G34; J33; M12; M52
1. INTRODUCTION
CEO compensation is widely discussed, especially during the financial crisis. The discussion both in public as in the academic world mainly focuses on the relative pay levels of CEOs (Bebchuk and Grinstein, 2005; Frydman and Jenter, 2010), transparency of pay structures (Bhagat and Romano, 2010), the relationship between pay and performance (Murphy, 1985; Jensen and Murphy, 1990; Ozkan, 2011) and more recently studies focusing on shareholder activism and CEO compensation schemes are popular, the so called ‘Say on Pay’ (Conyon and Sadler, 2010; Ferri and Maber, 2012). The studies that try to explain the variance in CEO compensation mainly focus on classical theories. On the one hand these theories state that CEO compensation is the result of companies who design compensation schemes that provides managers with efficient incentives in order to maximize shareholder value and reduce agency costs. On the other hand these theories emphasize the significant discretionary power CEOs possess and describe that compensation is the result of the potency of CEOs to influence their own remuneration contracts.
This study takes a different approach. Next to the classical theories, this paper also takes in account alternative models that try to explain compensation practices. These alternative theories focus on the market driven aspect of CEO compensation. In this view globalization and the supply and demand of talented CEOs are important determinants of compensation practices. In order to find evidence for this explanation this research aims at analysing whether Anglo-American influences effect CEO compensation practices within Continental-European companies. More specifically, the central question in this study is: ‘What is the impact of Anglo-American influences on both the level and structure of CEO compensation within Continental-European companies?’ To test this research question, data that covers a three year period of publicly listed companies from France, Germany and the Netherlands will be used. This dataset includes, among others, the following three explanatory variables as proxies for Anglo-American influence within Continental-European companies: Anglo-American cross-listing activities, the presence of an Anglo-American director within the remuneration committee and the role of this director within the committee.
These compensation practices contain a specific set of characteristics that differ significantly from the practices used within the Continental-European system. For example, compensation within the Anglo-American system is much more focused on short and long term incentives in order to reduce the agency problem. Besides, information requirements to increase monitoring, transparency and accountability are more stringent in the Anglo-American model. In this study we explain that these requirements may raise CEO pay levels within a Continental-European company.
Previous studies that focused on the effect of Anglo-American influences within non-Anglo-American companies (Oxelheim and Randøy, 2005; Hermes and Schulenberg, 2008) found that both cross-listing activities and the presence of an Anglo-American director in the board positively significantly influence executive compensation. This paper contributes to both the corporate governance literature and the literature on CEO compensation by testing two new measures of Anglo-American influence that focus on the remuneration committee of a Continental-European company and by using a larger hand collected unbalanced panel data set from three countries over three years.
From the findings it appears that CEO compensation within Continental-European companies is positively associated with two specific Anglo-American influences. Firstly, this study shows that cross-listing activities of Continental-European companies within Anglo-American countries positively impact the level of CEO compensation. Besides, the results of this study also indicate that not the presence, but the role as chairman of an Anglo-American director within a remuneration committee of a Continental-European company may influence the structure of CEO compensation. However, the latter relationship is considered to be weak and not completely generalizable.
2. THEORETICAL FRAMEWORK & HYPOTHESIS DEVELOPMENT
2.1 Compensation Theories
There are several theories that try to explain variances in executive compensation. These theories will be discussed and in the next sub-chapter related to the specific topic of this study in order to derive the hypotheses. The two main models used to explain the primary forces shaping executive pay advocate for the primacy of market based explanations and the other for the importance of executive power over the board. Market based explanations are based on the relationship between principals (shareholders) and agents (managers). Berle and Means (1932) are the first to describe the situation where there is a separation of ownership and control within a company. This separation of ownership and control can lead to a conflict of interest between the two parties, in which the managers favour their own interests over those of the shareholders. Consequently managers will use organizational sources for their own benefits instead of following strategies that maximizes shareholders return (Jensen and Meckling, 1976). These actions are considered to be part of the so called agency problem. The main reason for this conflict of interest is the widely dispersed ownership within some companies and the information asymmetry between the shareholders and the managers. Both reasons make it difficult for small shareholders to monitor and control the activities of the agents. With these characteristics of the agency problem in mind, the ‘optimal contracting approach’ (Jensen and Meckling, 1976) state that companies should design an efficient remuneration package that motivates executives to work in shareholders’ interest. The main focus when designing this remuneration package is on aligning interest of both parties using incentives from different sources, e.g. financial incentives or portfolio incentives, and making sure that pay and performance are positively correlated. The different corporate governance codes support this view and stress the importance of a strong pay-performance relationship.
great part are de-coupled from managers’ performance and shareholders’ interest, but mainly focuses on value creation for themselves. According to Bebchuk and Fried (2003) companies will use pay practices that are non-transparent and difficult to value in order to camouflage this rent seeking behaviour for shareholders. These are the so called stealth compensation practices. Arrangements used by companies to camouflage the compensation are for example pension plans, deferred compensation and stock options.
Alternative models that try to explain compensation practices do not directly relate executive compensation to the agency problem, but are more focused on the market driven aspect of compensation practices. This market driven aspect is important to this study where globalization and with this the interchange of employees, knowledge and capital is the central item to be discussed. These alternative theories focus on the requirements for a CEO candidate and the skills of a CEO, herewith discussing both the demand and supply side of a CEO market, respectively the managerial discretion and the human capital theory (Oxelheim and Randøy, 2005; Conyon, 2006). These theories state that compensation is not the result of CEO power over the board or as a solution to the agency problem, but that compensation is a consequence of the limited supply of CEOs for a job and the willingness of companies to pay for specific skills of a CEO, i.e. a function of supply and demand.
2.2 Anglo-American influences
2.2.1 Corporate governance systems
Jean Tirole (2001) described corporate governance as the “design of institutions that induce or force management to internalize the welfare of stakeholders”. The central theme within corporate governance is the presence of the agency problem. Different measures are used to improve monitoring, accountability and transparency in order to reduce the agency problem and with this aligning the interest of managers and shareholders. Denis and McConnell (2003) broadly characterize these measures as being internal or external to the firm. The internal mechanisms of primary interest are the composition of the board of directors and the ownership structure within firms. On the other hand the external mechanisms are the external markets for corporate control, i.e. the takeover market and the legal system of a country. Aguilera and Jackson (2003) emphasize the role of the legal system as one of the external corporate governance mechanisms. They state that legal systems shape the nature of corporate governance, especially the patterns of corporate ownership. La Porta et al. (1998, 1999) relate the legal tradition of a country to different corporate governance measures. In their studies they classify two legal families, i.e. civil law and common law. Where civil law is mainly based on statutes and comprehensive codes and common law is based on earlier judicial decisions. La Porta et al. (1998, 1999) indicated two important differences between a countries’ legal system and their corporate governance practices. First they analysed the laws of different countries with respect to shareholder and creditor rights. They conclude that common-law countries generally have stronger shareholder and creditor protection than civil law countries. A second difference between the legal systems is the ownership structure of companies. La Porta et al. (1999) state that ownership concentration is a substitute for weak legal protection, because large owners are better able to exercise their rights, since they have stronger voting power, more incentives to monitor and sometimes direct access to decision making. In their study they find that ownership of companies within common law countries is widely dispersed. This is in line with the image of Berle and Means (1932) of the widely held company. However, as expected, ownership within civil law countries is concentrated and the stock market within these countries is less developed. This concentration of ownership lead to a new agency problem: the one between the large controlling (family) owners and the minority shareholders. Based on the studies discussed above and the paper by Aguilera and Cuervo-Cazurra (2004) two corporate governance systems are distinguished: the Anglo-American/Anglo-Saxon model and the Continental-European model, respectively dominant in the common and civil law countries.
are publicly listed in the Anglo-American model and ownership is widely dispersed. In sum the conflict of interest between managers and shareholders here is potentially larger than within the Continental-European model, where ownership is more concentrated and shareholders are normally better able to monitor managers and get a return on their investment. To reduce the potential agency problems information requirements, to increase monitoring, transparency and accountability are more stringent in the American model. This causes that the Anglo-American corporate governance system is regarded to be one of the most demanding and unforgiving corporate governance systems in the world (Lucier et al., 2004; Oxelheim and Randøy, 2005). Within the Anglo-American corporate governance system there is less tolerance for poor performance of CEOs, more demand of their capacity and with this a higher risk of dismissal compared to the Continental-European model (Lucier et al., 2004). These higher efforts are reflected in CEO remuneration, which is generally higher in the Anglo-American model. Oxelheim and Randøy (2005) explain the difference in CEO compensation between the two corporate governance models as the risk premium, which is associated with the harsher monitoring of owners and board members within the Anglo-American model.
2.2.2 Anglo-American cross listing
To understand why Anglo-American cross listing may influence CEO compensation within Continental-European companies, it is first important to get insight into the reasons why companies choose to cross-list. Pagano et al. (2002) studied the trends in cross listing in the period 1986-1997 and came up with several motivations why companies list their shares on foreign stock markets. Most important reason to cross-list is to get access to external financing in foreign countries. However, the choice of a company to cross-list may not always have a finance objective. Other reasons are merely based on cultural and geographical principles. For example, enhancing visibility in (potential) export markets (Baker et al., 2002) or to get in contact with specialised investors who are better able to value business growth opportunities. Another motivation that is more aligned with the topic of this study results from a paper by Pagano et al. (2001). They state that companies cross-list to get exposed to stricter corporate governance practices in order to alleviate the agency problem. This may explain the finding of Pagano et al. (2002) of an increasing number of European companies that cross list, mainly aiming at US stock markets.
the stricter corporate governance system of the country. These factors make the job of a CEO more demanding. Besides, Anglo-American investors are less tolerant towards underperforming CEOs, resulting in higher risk of dismissal. The higher efforts and the risk of dismissal are reasons for a CEO to ask for a higher compensation and risk premium. Secondly, a CEO must possess a scarce set of skills in order to deal with the changing environment after the cross listing. According to the managerial discretion and the human capital theory, both discussed in chapter 2.1, this set of specific skills result in a limited supply of CEOs for the job and the willingness of companies to pay for these skills. Both influencing CEO compensation positively. Finally, as discussed in the previous sub-chapter, one of the differences between the Anglo-American model and the Continental European model is the dispersion of ownership within Anglo-American companies. A risk of highly dispersed ownership is the conflict of interest between managers and shareholders. According to the optimal contracting approach shareholders must reward CEOs with incentive based compensation in order to align the interest between the two parties. This is another reason that explains cross-listing as one of the factors that influences CEO compensation. Taken together, it is to be expected that there is a positive relation between Anglo-American cross-listing of Continental-European companies and CEO compensation. This results in the following hypothesis:
Hypothesis 1a: There is a positive association between Anglo-American cross-listing of
Continental-European companies and total CEO compensation.
Besides, since incentive based compensation is, according to the optimal contracting approach, one of the key variables to align the interest between principals and agents. It is to be expected that when a Continental-European company lists on an Anglo-American stock market the share of short term incentives within total CEO compensation increases. This results in the following hypothesis:
Hypothesis 1b: There is a positive association between Anglo-American cross-listing of
2.2.3 Anglo-American directors
The process of globalization not only removes barriers to get access to external financing in foreign countries, it also reduces firm level barriers to cross-border knowledge and skills. One of the effects of this is the increase of board globalisation (Staples, 2007), where a foreigner take place in the board of a company. Research show that the inclusion of at least one non-national board member can have positive impact on a company. For example, Oxelheim and Randøy (2003) show that Anglo-American board membership enhances firm value among Scandinavian firms. Besides, Oxelheim and Randøy (2005) also found a positive relationship between Anglo-American board membership and CEO compensation for their sample of Norwegian and Swedish firms. The latter relationship is directly aligned with the topic of this study. However, this study will focus on the Anglo-American directors that sit on the remuneration committee of a Continental-European firm. Using this variable gives better insight in a possible relationship, since the remuneration committee is a subcommittee of the board with the task of determining CEO compensation schemes.
Hypothesis 2a: There is a positive association between an Anglo-American director
sitting on a remuneration committee of a Continental-European company and total CEO compensation.
Besides, based on the expectation that Anglo-American directors bring in compensation practices of the Anglo-American corporate governance system, one can also expect that the share of incentive based compensation within the total CEO compensation increases. Since incentive based compensation is, according to the optimal contracting approach, used to motivate executives to work in shareholders’ interest where the agency problem is potentially larger within the Anglo-American model. The following hypothesis is developed:
Hypothesis 2b: There is a positive association between an Anglo-American director
sitting on a remuneration committee of a Continental-European company and the share of short-term incentives within total CEO compensation.
2.2.4 Role of Anglo-American directors
Hypothesis 3a: There is a positive association between an Anglo-American chairman of
a remuneration committee, within a Continental-European company, and total CEO compensation.
Besides, as discussed in the layout to hypothesis 2b, it is also expected that the Anglo-American chairman of the remuneration committee brings in compensation practices of the Anglo-American corporate governance model that are especially focused on incentive based compensation. This results in the following hypothesis:
Hypothesis 3b: There is a positive association between an Anglo-American chairman of
a remuneration committee, within a Continental-European company, and the share of short-term incentives within total CEO compensation.
3. METHODOLOGY
3.1 Research Method
The research question ‘What is the impact of Anglo-American influences on both the level and structure of CEO compensation within Continental-European companies?’ will be tested with empirical research. In the following sub-chapters the different variables will be discussed. Firstly, the dependent variables are further explained, secondly the independent variables and lastly the ten different control variables. These variables will form the basis of the empirical model that will be designed in the final sub-chapter.
3.1.1 The dependent variables
measured by the natural logarithm, because of a possible skewed distribution which can lead to normality problems. The second dependent variable will measured as the percentage of short term incentives within the variable discussed before; total CEO compensation. This way the Anglo-American influences on both the level and composition of CEO compensation will be measured.
3.1.2 The explanatory variables
To find evidence for the hypotheses set out in chapter two there are three key explanatory variables of interest in this study. First a dummy variable will be created for Anglo-American
cross-listing. The variable is equal to 1 if a Continental-European company from the sample is
listed on an official Anglo-American stock exchange and 0 otherwise.
In order to be able to find evidence for hypothesis two a second dummy variable will be created. This explanatory variable, Anglo-American remuneration committee membership, will assign a value of 1 if one (or more than one) Anglo-American director(s) is member of the remuneration committee of a Continental-European company and 0 otherwise. A dummy variable will be used instead of the share of Anglo-American members, because according to Oxelheim and Randøy (2005) “the signalling effect of Anglo-American corporate governance influence would be achieved by the inclusion of even one Anglo-American board member”. In line with the latter variable a third dummy will be set up that equals 1 if the Anglo-American member of the remuneration committee is also the chairman of the committee.
Related to the first two variables a value of 1 is assigned when a company is cross-listed or the director is born in a country with an Anglo-American corporate governance model, i.e. United States, Canada, United Kingdom and Australia (Fernando, 2009).
3.1.3 The control variables
Following prior literature a number of control variables will be included in the empirical model in order to account for possible influences on CEO compensation other than Anglo-American influences. The control variables that will be used are classified in three categories: firm characteristics, governance characteristics and individual CEO characteristics.
governance codes used by countries, that both mainly stimulate the use of incentives to create a strong pay performance relationship, one can say that performance should be an important determinant of CEO compensation. This study will use two proxies for firm performance, one backward-looking accounting measure and one forward-looking mixed accounting and market based measure, respectively ROE and Tobin’s Q. The accounting variable ROE is measured as the net income of continuing activities divided by the average shareholders’ equity. Besides, Tobin’s Q is measured as the sum of the year-end book value of total debt and market capitalization divided by the book value of total assets at the end of fiscal the year. The variable market capitalization, also known as the market value of common stock, is calculated as the number of ordinary outstanding shares times the market price of a share, both at the end of the fiscal year.
The second firm characteristic that will be used as a control variable is firm size. Prior studies found firm size to be positively correlated with CEO compensation. Firstly, Simon (1957) argued that CEO compensation is greater in larger firms because of the number of levels within the organization in combination with the efforts of larger companies to maintain pay differentials between the levels. Secondly, Gomez-Meija et al. (1987) state that larger firms are more complex and difficult to manage. Based on the human capital theory this leads to a company that is willing to pay higher levels of compensation for managers who possess the specific managerial skills to lead the company. Finally, in their meta-analysis Tosi et al. (2000) found that firm size explains 40% of the variance in CEO compensation. In this study firm size is measured by the sales of a company and the average number of employees in the fiscal year. To reduce the impact of a possible skewed distribution the natural logarithm of both variables will be used. Using these proxies for firm size is line with previous studies (Duffhues and Kabir, 2008; Becker-Blease, 2010; Haleblian and Finkelstein, 1993; among others).
relationship may also be positive (Duffhues and Kabir, 2008), because higher debt is associated with a higher risk of the firm, which may induce higher compensation levels for CEOs. Leverage ratio is calculated as the book value of debt divided by the book value of total assets at the end of the fiscal year.
To control for corporate governance effects the variable board size will be included in the model. Previous research shows that the relation between board size and CEO compensation has two sides. One the one hand there is evidence for a negative relationship (Hwang and Kim, 2009). This is based on the idea that larger boards have a greater level of expertise and tasks can be more easily divided, which facilitates a higher level of monitoring. On the other hand several studies found a positive relationship (Core et al. 1999; Sapp, 2008) between board size and CEO compensation, based on the belief that boards become oversized and therefore are less effective in fulfilling their role of monitoring the management. This is in line with the research of Jensen (1993) who argued that it is likely that boards become ineffective when it has more than seven or eight members. Board size is measured as the total number of executive and non-executive directors in the board or supervisory board of the company. Also the squared variable will be included in the model to allow for the non-linear relationship described above.
A second governance characteristic that will be included in the empirical model is board
independence. According to the managerial power approach (Bebchuk and Fried, 2004) managers
Finally, CEO is Chairman is the last governance characteristic that will be added as a control variable. This variable is an indicator that equals 1 if the CEO is also chairman of the board of directors and zero otherwise. Based on the managerial power theory, discussed in chapter two, it is expected that a CEO who is also chairman of the board has a significant influence on its own compensation scheme, since the CEO is entrenched and there is no arm length bargaining with the other directors anymore. One of the papers that found evidence for this positive association is the paper by Fahlenbrach (2008). He states that: “CEOs who are also chairman have both higher pay and higher pay-for-performance sensitivity”.
To control for personal characteristics of the CEO, CEO tenure will be added to the model. One of the first academic papers that found evidence for this relationship is the paper by Hill and Phan (1991). In their study they state that if tenure increases CEOs get more entrenched and are better able to influence the board, which allows them to help design their own compensation packages. Johnston (2002) sets out another line of reasoning to explain the connection between the variables. In his paper he states that change in compensation is a reflection of CEOs who develop skills, become more experienced and having a higher productivity over the length of tenure. Research so far provided mixed results. For example, Deckop (1988) and Johnston (2002) found a positive relationship between CEO tenure and compensation, whereas Finkelstein and Hambrick (1989) and McKnight and Tomkins (2004) found no association. Tenure will be measured as the number of years an individual is CEO as of at end of the fiscal year.
TABLE 1 Variable definitions.
Variable name Description Dependent Variables
Ln_CEOtotal Total CEO compensation will be measured as the sum of base salary and annual bonus tied to accounting performance for the fiscal year. The natural logarithm will be used in order to deal with normality problems. Expressed in Euros (€).
CEOshort The share of short-term incentives will be measured as the percentage of short term incentives within the total CEO compensation. Short term incentives are measured as the annual cash bonuses tied to accounting performances.
Independent Variables Explanatory Variables
Dummy_Crosslist A dummy variable to measure Anglo-American cross-listing; it will be equal to 1 if the firm is listed on an offical Anglo-American stock exchange and 0 otherwise.
Dummy_Membership A dummy variable to measure if an Anglo-American director is member of the remuneration committee; it will be equal to 1 if one (or more than one) Anglo-American director(s) is member of the remuneration committee of Continental-European company and 0 otherwise.
Dummy_Chairman A dummy variable to measure if an Anglo-American member of the remuneration committe is also chairman of the committee; it will be equal to 1 if the Anglo-American member is also the chairman of the committee and 0 otherwise.
Definition Anglo-American A company is Anglo-American cross-listed or has an Anglo-American director when the company is cross-listed or the director is born in a country with an Anglo-American corporate governance model i.e. United States, Canada, United Kingdom and Australia (Fernando 2009).
Control Variables
Tobinsq Tobin's Q is a measure of firm performance; it will be measured as the sum of the year-end book value of total debt and market capitalization divided by the book value of total assets at the end of fiscal the year. Market capiltalization will be calcuated as as the number of ordinary outstanding shares times the market price of a share, both at the end of the fiscal year. ROE Return on Equity is a measure of firm perfromance; it will be measured as the net income of
continuing activities divided by the average shareholders’ equity.
Ln_Sales Sales is a measure of firm size; it will be measured as the natural logarithm of the sales. Expressed in Euros (€).
Ln_Employees Number of employees is a measure of firm size; it will be measured as the natural logarithm of the average number of employees.
Leverage Leverage ratio will be calculated as the book value of debt divided by the book value of total assets at the end of the fiscal year.
Boardsize Board size will be measured as the total number of executive and non-executive directors in the board or supervisory board of the company. To allow for a possible non-linear relationship the variable will be squared in the regression model.
Independence Board independence will be measured as the percentage of independent outside directors of the total board of directors/supervisory board.
Dummy_CEOchairman A dummy variable to measure if the CEO is also chairman of the board; it will be equal to 1 if the CEO is also chairman and 0 otherwise.
CEOtenure CEO tenure will be measured as the number of years an individual is CEO as of at end of the fiscal year.
CEOage CEO age will be measured as the age of the CEO in the fiscal year. To allow for a possible non-linear relationship the variable will be squared in the regression model.
Dummy_Industries 5 Dummy variables to measure the industry of the observation; Dummy_Manu , Dummy_Cons , Dummy_Trad , Dummy_Serv , Dummy_Oth .The dummy will be equal to 1 if the industry of the obseveration is respecitively manufacturing, construction, trade, service or other and 0 otherwise Dummy_Years 3 Dummy variables to measure the year of the observation; Dummy_2005, Dummy_2006,
3.1.4 Empirical model
The independent variables identified before are the key explanatory variables in the regression model that will be used to estimate the impact of Anglo-American influences on CEO compensation within Continental-European companies. To test whether there is a positive association between Anglo-American cross-listing of Continental-European companies and CEO compensation (H1a), the following regression model is constructed:
Ln_CEOtotalit = α + β1Dummy_Crosslist it-1 + β2Tobinsq it-1 + β3ROE it-1 + β4Ln_Sales it-1 +
β5Ln_Employees it-1 + β6Leverage it-1 + β7Boardsize it-1 + β8Boardsize 2
it-1
+ β9Independence it-1 + β10 Dummy_CEOchairman it-1 + β11CEOtenure it
+ β12CEOage it + β13CEOage2it
In relation with the previous empirical model, the following model is constructed to find out if Anglo-American cross-listing of Continental-European companies not only positively influence the level of CEO compensation, but also the share of short-term incentives within the total CEO compensation (H1b):
CEOshortit = α + β1Dummy_Crosslist it-1 + β2Tobinsq it-1 + β3ROE it-1 + β4Ln_Sales it-1 +
β5Ln_Employees it-1 + β6Leverage it-1 + β7Boardsize it-1 + β8Boardsize2 it-1
+ β9Independence it-1 + β10 Dummy_CEOchairman it-1 + β11CEOtenure it
+ β12CEOage it + β13CEOage2it
To test if and how Anglo-American members of a remuneration committee in a Continental-European company determine CEO compensation (H2a), the following model is constructed:
Ln_CEOtotalit = α + β1Dummy_Membership it-1+ β2Tobinsq it-1 + β3ROE it-1 + β4Ln_Salesit-1
+ β5Ln_Employees it-1+ β6Leverage it-1+ β7Boardsize it-1+ β8Boardsize 2
it-1
+ β9Independence it-1 + β10 Dummy_CEOchairman it-1 + β11CEOtenure it
+ β12CEOage it + β13CEOage 2
To test whether these Anglo-American remuneration committee members also influence the share of short-term incentives within the total CEO compensation (H2b), the following equation will be estimated:
CEOshortit = α + β1Dummy_Membership it-1+ β2Tobinsq it-1 + β3ROE it-1 + β4Ln_Salesit-1 +
β5Ln_Employees it-1+ β6Leverage it-1+ β7Boardsize it-1+ β8Boardsize 2
it-1 +
β9Independence it-1 + β10 Dummy_CEOchairman it-1 + β11CEOtenure it +
β12CEOage it + β13CEOage 2
it
Related to hypothesis 2 the following model is constructed, to find out if an Anglo-American chairman of the remuneration committee within a Continental-European company even has more influence than a normal member on the total CEO compensation (H3a):
Ln_CEOtotalit = α + β1Dummy_Chairman it-1 + β2Tobinsq it-1 + β3ROE it-1 + β4Ln_Sales it-1
+ β5Ln_Employees it-1+ β6Leverage it-1+ β7Boardsize it-1+ β8Boardsize 2
it-1
+ β9Independence it-1 + β10 Dummy_CEOchairman it-1 + β11CEOtenure it
+ β12CEOage it + β13CEOage 2
it
The last hypothesis (3b) will be estimated using the following model, where the focus is not on the influence of the relevant chairman on total CEO compensation, but on the share of short-term incentives within this compensation scheme:
CEOshortit = α + β1Dummy_Chairman it-1 + β2Tobinsq it-1 + β3ROE it-1 + β4Ln_Sales it-1 +
β5Ln_Employees it-1+ β6Leverage it-1+ β7Boardsize it-1+ β8Boardsize 2
it-1 +
β9Independence it-1 + β10 Dummy_CEOchairman it-1 + β11CEOtenure it +
β12CEOage it + β13CEOage2it
variables tenure and age, are lagged with one year with respect to the explanatory variables. The one year lag of the variables is in line with the research of Oxelheim and Randøy (2005) and other studies on CEO compensation (Kerr and Bettis, 1987; Coombs and Gilley, 2005). In these studies they argue that levels of CEO compensation are evaluated based on past results and therefore are a function of last year changes in the independent variables, e.g. performance, board independence, board size , etc. Based on this knowledge, in this study, the variables tenure and age are not lagged, since these are ascending variables that grow constantly with one year.
3.2 Sample Selection and Data Sources
The dataset that will be used covers a three year period of publicly listed companies from France (Euronext Paris), Germany (Frankfurt Stock Exchange) and the Netherlands (Euronext Amsterdam). In order to prevent possible influences of the credit crisis, which started in 2008, on the empirical model, data for the years 2005 to 2007 will be used. In line with Oxelheim and Randøy (2005) financial firms are excluded, because these companies are subject to special corporate governance rules and regulations. Per year 55 companies from each country are selected. For Germany the selected companies are listed on the DAX or MDAX, for France on the CAC 40 or the CAC Next 20 and the Dutch companies are listed on the AEX, AMX or ASCX. The dataset is for 80% hand collected, using annual and financial reports from the different selected companies. Besides, this study will use data collected by Hermes and Schulenburg (2008) from their research on the same topic and it will use some data that was available in the Orbis and Amadeus databases. Finally, following Hermes and Schulenburg (2008), companies not headquartered in the aforementioned countries or companies that are dual-listed with also a headquarter in an Anglo-American corporate governance system will be excluded from the dataset.
4. EMPIRICAL RESULTS AND DISCUSSIONS
4.1 Descriptive Statistics
Table 2 reports the descriptive statistics for the variables that will be used in the regression analyses. Firstly, with respect to panel A, that presents the descriptive statistics of the dependent variables and the variables that compose the total CEO compensation measure, the following two insights emerges. The mean of the total CEO compensation over three years is €1,593,324. Further specified it stands out that this figure not remained stable over the three year period but increased significantly with 14.97%. However, it is important to notice that long-term incentives, that become more important within the design of the compensation scheme, are not included within this total CEO compensation measure. That is why this growth in total compensation may in reality be much stronger. Besides, from the descriptive statistics it can be concluded that over the three year sample period the base salary is still the main component within the CEO compensation scheme, whereas the mean of the variable CEOshort is 46.1%. However, short term incentives become more important within the total CEO compensation. Over the three year period the share of short-term incentives increased, related to the 2005 figure, with 1.33% to 47.09%.
explanatory variables (Dummy_Membership, Dummy_Chairman). This difference arises because some companies did not use a specific committee to design remuneration packages for the executives within the firm. Another reason is that some companies did not include any information on the remuneration committee within the annual report.
With respect to the descriptive statistics of the control variables, discussed in panel C, the following insights emerge. Firstly, the boards of the companies in the dataset are on average quite independent, with 73.5% of all directors in the board being independent. However, more striking for this variable is the low number of firm-year observations. This number is low due to German companies, which are not required to include any information on independency of the directors in their annual reports. Therefore, this specific variable will be excluded from the analyses in order to prevent distorted results. Secondly, the variables sales and employees, that control for firm size, are skewed to the right, since the median of these variables differs substantially from the mean. This seems to indicate that a few companies within the dataset present high values for these firm characteristics. Lastly, on average, in 16.3% of the companies the CEO was also the chairman of the board. This figure is relatively high, since Dutch and German companies are obliged by the corporate governance code to have two-tier board and French companies also can choose to work with a one-tier board structure. The difference between the two board structures is that within the one-tier board all the directors, both executive and non-executive, are united in the so called board of directors. In contrast to a two-tier board structure, where there is a separate supervisory board that is fully represented by non-executives and logically the CEO cannot be the chairman.
TABLE 2 Descriptive Statistics.
4.2 Correlations
Table 3 presents the results of the performed Pearson correlation test among the different regression variables, from these results the following four insights emerge. Firstly, in line with results of the same test performed by Oxelheim and Randøy (2005), there is a significant positive correlation, at the 1% level, between cross-listing of Continental-European companies and total CEO compensation (0.200). A similar significant correlation is found between the same explanatory variable and the share of short-term incentives (0.l80). Secondly, the results show that there is a significant correlation between the three explanatory variables of this study. This makes sense, since Continental-European companies that choose to cross-list to an
Anglo-Variable N Mean StDev Min 25% Median 75% Max
Panel A: Depedent variables
CEOtotal in € 336 1,593,324 1,025,426 0 766,737 1,415,900 2,250,250 4,850,680
CEO Base salary in € 336 725,737 360,764 0 440,298 674,000 931,250 1,873,933
CEO Bonus in € 335 865,798 781,290 0 240,000 669,940 1,337,021 3,383,187
CEOshort 335 0.461 0.194 0 0.333 0.497 0.589 1.0
Panel B: Explanatory variables
Dummy_Crosslist 332 0.283 0.451 0 0 0 1.0 1.0
Dummy_Membership 266 0.226 0.419 0 0 0 0 1.0
Dummy_Chairman 266 0.086 0.282 0 0 0 0 1.0
Panel C: Control variables
Tobinsq 325 1.560 0.958 0.710 1.086 1.296 1.623 9.550 ROE 336 0.136 0.224 -2.550 0.090 0.148 0.212 1.080 Sales (000s €) 335 14,857,530 20,136,619 147.0 2,009,084 6,926,500 16,016,000 87,341,432 Employees 336 64,465 83,551 20.0 9,149 26,544 86,377 346,783 Leverage 336 0.642 0.148 0.170 0.556 0.664 0.751 1.210 Boardsize 333 11.466 5.908 3.0 6.0 11.0 16.0 28.0 Independence 204 0.735 0.251 0.130 0.556 0.800 1.0 1.0 Dummy_CEOchairman 332 0.163 0.370 0.0 0.0 0.0 0.0 1.0 CEOtenure 307 6.707 6.573 1.0 2.330 4.670 8.0 28.800 CEOage 318 54.802 6.873 36.0 50.0 55.0 60.0 75.0
Panel D: Extra information on dummy and other variables
Freq. % total Freq. 2005 Freq. 2006 Freq. 2007
American country may attract Anglo-American directors that have knowledge of the rules and corporate governance code of the specific stock market. Thirdly, there is no significant correlation between the two control variables that both measure firm performance. This indicates that, as discussed in chapter 3.1.3, the two variables measure different dimensions of firm performance, i.e. one backward-looking accounting based measure (ROE) and one forward-looking mixed accounting and market based measure (Tobinsq). Finally, the other variables that control for firm characteristics and governance characteristics are significantly correlated with the dependent variables. This shows their consistency with prior literature and the relevance of including them in the regression analyses.
The Pearson correlation test is also performed to control for possible multicollinearity problems between the different variables, which may significantly influence the results of the final regression analyses. Cohen (1988) suggests that a correlation coefficient greater than 0.5 is an indication of large correlation. First it is important to notice that the explanatory variables
Dummy_Membership and Dummy_Chairman are highly correlated (0.582). In order to avoid
collinearity bias the two variables will not be included within one final full regression model. Therefore two full models will be set up, both including two of the three explanatory variables. Besides, table 3 shows that the control variables Ln_Sales and Ln_Employees are highly correlated, i.e. 0.859. This is explained by the fact that these control variables are both measures of firm size. However, the same table also show that these two variables have a high correlation with the variables Leverage (0.456, 0.512) and Boardsize (0.620, 0.593). Consequently, in order to prevent biased results, the two variables for firm size will not be employed in the main regression models. However, as a robustness test, Ln_Employees will replace Leverage and
TABLE 3
Pearson Correlation Matrix. Pearson correlations among independent variables
Variable A B C D E F G H I J K L M N (A) Ln_CEOtotal 1.000 (B) CEOshort 0.514 *** 1.000 (C) Dummy_Crosslist 0.200 *** 0.180 *** 1.000 (D) Dummy_Membership 0.005 -0.018 0.220 *** 1.000 (E) Dummy_Chairman 0.058 0.063 0.131 ** 0.582 *** 1.000 (F) Tobinsq -0.028 0.047 0.029 0.128 ** 0.167 *** 1.000 (G) ROE -0.011 0.085 -0.019 -0.064 -0.097 0.080 1.000 (H) Ln_Sales 0.271 *** 0.363 *** 0.285 *** 0.014 -0.026 -0.441 *** 0.033 1.000 (I) Ln_Employees 0.229 *** 0.301 *** 0.262 *** 0.008 -0.061 -0.432 *** -0.014 0.859 *** 1.000 (J) Leverage 0.163 *** 0.153 *** -0.096 * -0.059 0.020 -0.452 *** -0.079 0.456 *** 0.512 *** 1.000 (K) Boardsize 0.319 *** 0.462 *** 0.220 *** -0.194 *** -0.089 -0.265 *** -0.104 * 0.620 *** 0.593 *** 0.278 *** 1.000 (L) Dummy_CEOchairman -0.030 -0.023 0.021 0.101 0.150 ** -0.108 * -0.130 ** 0.236 *** 0.232 *** 0.124 ** 0.159 *** 1.000 (M) CEOtenure 0.090 0.009 -0.017 -0.057 -0.026 -0.022 0.081 -0.090 -0.041 -0.007 0.006 0.273 *** 1.000 (N) CEOage 0.018 -0.007 0.171 *** 0.073 -0.092 -0.102 * -0.009 0.117 ** 0.129 ** 0.083 0.204 *** 0.247 *** 0.404 *** 1.000
4.3 Main Results
In table 4-7 the results of the performed multiple regressions tests are presented. Firstly, in table 4 and 6 regressions are performed without including any control variables. Excluding the control variables in these regressions give insight in a possible standalone relationship between the dependent and the explanatory variables central in this study. In each model the explanatory variables are regressed individually against the dependent variable. Because of the possible collinearity problems, between the explanatory variables Dummy_Membership and
Dummy_Chairman, two instead of one full model is used. These full models both include two of
the three explanatory variables. Secondly, table 5 and 7 present the results of the performed regressions including the control variables and show what happen to the possible relationship described in table 4 and 6. These two tables are made up of six models. In order to measure the model fit, in both table 5 and 7, the first model demonstrate the relation between only the dependent variable and the control variables of this study. Afterwards the regression is performed in five steps to find out whether a particular explanatory variable is a significant predictor of the dependent variable, over and above the other control variables tested in model one. The results are presented in models two to six and are in line with the hypotheses set out in chapter 3. The first step examines the impact of Anglo-American cross-listing on CEO compensation practices (model 2). The second step estimates the relation between Anglo-American remuneration committee members and CEO compensation practices (model 3). Thirdly, the regression is performed to measure the impact of an Anglo-American chairman of a remuneration committee on CEO compensation practices (model 4). Finally, also in table 5 and 7 two full-model regressions are performed (model 5 and 6). The most remarkable results of the analyses will be discussed in the upcoming sub-chapters.
4.3.1 Anglo-American influences on total CEO compensation
Table 4 and 5 summarizes the outcomes of the regressions performed for hypothesis 1a, 2a and 3a. First it is important to notice that the F-value of the different models in table 5 is statistically significant at a significance level of 1%. This suggests that the models are valid and have explanatory power. However, this is not the case with all models in table 4. The F-value is statistically significant, at a significance level of 1% or 5%, only when the explanatory variable
Dummy_Crosslisting is included in the model. Without taking into account the other statistics and
where all the control and explanatory variables are included, the explanatory power increases with 4.1% and 4.4% to respectively 15.8% and 16.2%. To gain further insight in this explanatory power first the impact of the explanatory variables relevant in this study will be discussed. Secondly, the statistics of the control variables will be explored further.
Hypothesis 1a suggests that there is positive relationship between Anglo-American cross-listing and total CEO compensation. In model one of table 4 and model two of table 5 the results for this hypothesis are summarized. As discussed in the previous paragraph, the models in table 4 are only statistically valid and have explanatory power when the dummy for Anglo-American cross-listing is included. Also the coefficients in table 4 indicate the relevance of this variable and show that there is positive significant relationship between Anglo-American cross-listing and total CEO compensation at the 1% significance level. This result is also supported in the full models, i.e. model four and five. However, to better understand the existence of this relationship, the regressions presented in table 5 control for other variables that may influence the level of total CEO compensation. Again, in model two of table 5 statistically significant evidence, at the 1% significance level, is found for the relationship between Anglo-American cross-listing and total CEO compensation. This result remains about the same in the full model, i.e. model five and six, but at the 5% significance level. Hence, hypothesis 1a is accepted. This finding is in line with the paper by Oxelheim and Randøy (2005) who also found a statistically significant relationship between this specific Anglo-American influence and total CEO compensation. Also Hermes and Schulenberg (2008) found a similar association, however, they not only focused on CEO compensation, but also included the total compensation of other executives. Correctly interpreting this outcome, it means that there is a significant relationship between the cross-listing of a Continental-European company on an Anglo-American stock exchange and the total level of CEO compensation.
No statistical significant evidence was found for hypothesis 2a, presented in model two and four of table 4 and model three and five in table 5. This result indicates that an Anglo-American member of the remuneration committee does not significantly influence the CEO compensation level within a Continental-European company. This result also applies to the full models of both tables 4 and 5. Previous studies (Oxelheim and Randøy, 2005; Hermes and Schulenberg, 2008) focused not on Anglo-American members of the remuneration committee, but on the presence of Anglo-American board members in general. These studies found a significant relationship at a significance level of 1%.
more influence on the decision making process than a normal member. However, the F-value and the unstandardized beta in model three of table 4, that both are not statistically significant, already indicate that there is no significant evidence for this relationship. This outcome is supported by the results presented in model four and six of table 5. From this can be concluded that both Anglo-American members and chairman’s, of the remuneration committee in a Continental-European company, have no significant influence on the total level of CEO compensation.
The regression results show that only two control variables are significantly associated with total CEO compensation in all the six models. Firstly, the dependent variable is statistically significant related with the board size variable at the 5% significance level. As discussed in the literature review some researchers suggest that the relationship between board size and CEO compensation in non-linear. This non-linear relationship is based on the belief that at a certain level boards become oversized and therefore are less effective in fulfilling their monitoring role, which will positively influence the CEO compensation level. However, the results presented in table 5 are not in line with these studies. No significant evidence for a non-linear relationship is found, since the squared board size variable is not significant in all the models. On the other hand, in line with Core et al. (1999) and Sapp (2008) the results indicate that there is a positive relationship instead of a negative relationship. Concluding, that larger boards with greater level of expertise not facilitates a higher level of monitoring, but become oversized and therefore are less effective in fulfilling their monitoring role. Secondly, consistent with prior research, CEO tenure also positively significantly influences total CEO compensation at the 5% level. As discussed in the literature review, this may indicate that if tenure increases CEOs get more entrenched and are better able to influence the board. Or it may indicate that the higher CEO compensation is a reflection of the developed skills, experience and higher productivity over the length of the tenure.
TABLE 4
Regression results for total CEO compensation (Ln_CEOtotal), control variables excluded.
Expected sign 1 2 3 4 5
Explanatory variables
Dummy_Crosslist + (H1a) 0.753 *** (0.203) 0.622 *** (0.218) 0.577 *** (0.215)
Dummy_Members hip + (H2a) 0.019 (0.239) -0.143 (0.244)
Dummy_Chairman + (H3a) 0.334 (0.355) 0.202 (0.356)
Cons tant 13.695 *** (0.108) 14.016 *** (0.114) 13.988 *** (0.105) 13.857 *** (0.128) 13.820 *** (0.123)
Number of firm-year observations 332 266 266 263 263
R-squared 0.040 0.000 0.003 0.030 0.030
Adj. R-squared 0.037 -0.004 0.000 0.023 0.023
F-value 13.746 *** 0.006 0.885 4.075 ** 4.018 **
TABLE 5
Regression results for total CEO compensation (Ln_CEOtotal), control variables included.
4.3.2 Anglo-American influences on the share of CEO short-term incentives
Table 6 and 7 summarizes the outcomes of the regressions performed for hypothesis 1b, 2b and 3b. Again it is important to notice that the regression models that take into account the control variables, presented in table 7, are valid and have explanatory power, since the F value is statistically significant. Similar to results of table 4, in table 6 only the models that include the
Dummy_Crosslisting variable have a statistically significant F-value at the 1% and 10%
significance level. This value together with the positive unstandardized beta, that is also statistically significant in both model one and the full models (model 4 and 5), may already indicate the relevance of this specific explanatory variable. In comparison to the previous models in table 5, where total CEO compensation was the dependent variable, the models in table 7 show a significantly higher explanatory power. The adjusted R-Squared for model one is 0.286, stating that the variation in the share of short-term incentives of CEOs is for 28.6% explained by the control variables of this study. In the full regression models, presented in model five and six, the explanatory power increases with 3.1% and 3.2% to respectively 31.7% and 31.8%. In the upcoming paragraphs first the statistics of the explanatory variables will be discussed. Afterwards the regression results of the control variables will be examined.
Model one of table 6 and model two of table 7 present the results of the regression performed to test hypothesis 1b with CEOshort as the dependent variable. As discussed in the previous paragraph, only the models in table 6 that include the dummy for Anglo-American cross-listing are statistically valid and have explanatory power. Besides, both the betas in model one and the two full models are statistically significant at respectively the 1% and 5% level. In table 7 the same regressions are performed, this time with the inclusion of the control variables
Expected sign 1 2 3 4 5 6
Explanatory variables
Dummy_Cros slist + (H1a) 0.561 *** (0.173) 0.373 ** (0.163) 0.332 ** (0.161)
Dummy_Members hip + (H2a) 0.107 (0.175) -0.008 (0.181)
Dummy_Chairman + (H3a) 0.366 (0.257) 0.246 (0.263) Control variables Tobinsq + 0.104 (0.086) 0.098 (0.085) 0.104 (0.079) 0.088 (0.081) 0.107 (0.079) 0.093 (0.080) ROE + 0.042 (0.321) 0.039 (0.317) 0.123 (0.381) 0.174 (0.378) 0.085 (0.379) 0.134 (0.377) Leverage +/- 0.313 (0.587) 0.585 (0.586) 0.325 (0.565) 0.227 (0.573) 0.617 (0.575) 0.513 (0.586) Boardsize + 0.166 *** (0.063) 0.151 ** (0.063) 0.167 *** (0.061) 0.157 ** (0.060) 0.164 *** (0.061) 0.156 ** (0.060) Boardsize² - -0.003 (0.002) -0.003 (0.002) -0.004 (0.002) -0.003 (0.002) -0.004 (0.002) -0.003 (0.002) Dummy_CEOchairman + 0.081 (0.248) 0.135 (0.246) 0.096 (0.233) 0.066 (0.234) 0.140 (0.233) 0.102 (0.234) CEOtenure + 0.033 ** (0.014) 0.036 *** (0.014) 0.027 * (0.014) 0.028 ** (0.014) 0.027 * (0.014) 0.028 ** (0.013) CEOage + 0.120 (0.136) 0.158 (0.135) 0.106 (0.124) 0.111 (0.123) 0.116 (0.123) 0.120 (0.123) CEOage² - -0.001 (0.001) -0.002 (0.001) -0.001 (0.001) -0.001 (0.001) -0.001 (0.001) -0.001 (0.001) Constant 9.561 *** (3.596) 8.558 ** (3.570) 9.785 *** (3.220) 9.755 *** (3.205) 9.437 *** (3.203) 9.410 *** (3.196)
Number of firm-year observations 285 282 225 225 223 223
R-squared 0.381 0.176 0.180 0.187 0.200 0.203
Adj. R-squared 0.117 0.146 0.141 0.149 0.158 0.162
F-value 5.199 *** 5.799 *** 4.687 *** 4.918 *** 4.787 *** 4.890 ***
that may influence the share of short-term incentives within the total CEO compensation. What strikes from this table is that, although the presented positive significant relationship in model two, there is no significant relationship found in the two full models (model 5 and 6). Hypothesis 1b is not supported. This indicates that cross-listing of a Continental-European company on an Anglo-America stock market not significantly influence the share of short-term incentives within total CEO compensation.
To find evidence for hypothesis 2b the variable Dummy_Membership is included in the regressions as an explanatory variable. This dummy is included to test the association between Anglo-American members of the remuneration committee of a Continental-European company and the share of short term incentives within total CEO compensation. The results presented show that a significant relationship, at the 10% significance level, is only found in model three of table 6. However, all the other results expressed in table 6 and 7 show that, overall, there is no significant relationship between the two variables. This means that hypothesis 2b is also rejected.
What is most striking in table 6 and 7 is the relationship with the variable
Dummy_Chairman. Equal to the results of hypothesis 3a, also no significant evidence was found
for the relationship between an Anglo-American chairman of the remuneration committee and the share of CEOs’ short-term incentives in table 6. Besides, model three of table 6 is, according to the data, not statistically valid and has no explanatory power. However, when controlling for the fixed set of variables in table 7 it seems that there is a small positive significant relationship between the two variables at the 10% significance level. Correctly interpreting this result, it means that an Anglo-American chairman of a remuneration committee has a significant influence on the level of short-term incentives within his or her compensation scheme. The robustness tests performed in the next sub-chapter should tell more about the value of this remarkable outcome.
boards grow they become oversized and therefore are less effective in monitoring the actions of executive directors.
TABLE 6
Regression results for the share of short-term incentives (CEOshort), control variables excluded.
TABLE 7
Regression results for the share of short-term incentives (CEOshort), control variables included.
4.4 Robustness Tests
In this section a number of additional tests will be performed in order to test the robustness of the findings that are discussed in the previous sub-chapters. Firstly, as discussed in chapter 4.2, there is a high correlation among the regression variables that proxy for firm size, board size and leverage ratio. Therefore, the regressions presented in table 5 and 7 are replicated using Ln_Employees as a control variable instead of Boardsize and Leverage. Table 8 and 9 present the results for these regressions. Although one can detect slight differences in the level of the coefficients of the explanatory and control variables, using total CEO compensation as the dependent variable, the main findings are consistent and robust to those reported earlier. Furthermore, the new control variable that proxy for firm size (Ln_Employees) is a significant predictor of total CEO compensation at the 1% significance level. This is in line with prior studies
Expected sign 1 2 3 4 5 Explanatory variables Dummy_Crosslist + (H1b) 0.076 *** (0.023) 0.057 ** (0.025) 0.052 ** (0.025) Dummy_Members hip + (H2b) -0.008 (0.028) -0.026 (0.028) Dummy_Chairman + (H3b) 0.042 (0.041) 0.027 (0.041) Cons tant 0.441 *** (0.012) 0.479 *** (0.013) 0.473 *** (0.012) 0.468 *** (0.015) 0.461 *** (0.014)
Number of firm-year observations 331 264 264 262 262
R-squared 0.032 0.000 0.004 0.021 0.021
Adj. R-squared 0.030 -0.004 0.000 0.013 0.013
F-value 11.050 *** 0.081 1.028 2.719 * 2.720 *
The dependent variable in all regressions is total CEO compensation (Ln_CEOtotal). The numbers in parentheses are the standard errors Notes:***, **, * Coefficient is statistically significant at the 1%, 5%, and 10% level, respectively (2-tailed).
Expected sign 1 2 3 4 5 6 Explanatory variables Dummy_Cros slist + (H1b) 0.046 ** (0.021) 0.021 (0.023) 0.022 (0.023) Dummy_Members hip + (H2b) 0.047 * (0.026) 0.038 (0.026) Dummy_Chairman + (H3b) 0.083 ** (0.038) 0.073 * (0.038) Control variables Tobinsq + 0.052 *** (0.011) 0.051 *** (0.010) 0.042 *** (0.012) 0.039 *** (0.012) 0.042 *** (0.011) 0.039 *** (0.012) ROE + 0.086 ** (0.040) 0.082 ** (0.039) 0.130 ** (0.056) 0.140 ** (0.055) 0.122 ** (0.054) 0.132 ** (0.054) Leverage +/- 0.186 ** (0.073) 0.209 *** (0.072) 0.124 (0.083) 0.088 (0.084) 0.144 * (0.082) 0.110 (0.084) Boardsize + 0.025 *** (0.008) 0.025 *** (0.008) 0.029 *** (0.009) 0.027 *** (0.009) 0.031 *** (0.009) 0.029 *** (0.009) Boardsize² - 0.000 (0.000) 0.000 (0.000) 0.000 (0.000) 0.000 (0.000) -0.001 (0.000) 0.000 (0.000) Dummy_CEOchairman + -0.010 (0.031) -0.009 (0.030) 0.019 (0.034) 0.016 (0.034) 0.016 (0.033) 0.013 (0.034) CEOtenure + 0.004 ** (0.002) 0.004 ** (0.002) 0.003 (0.002) 0.003 (0.002) 0.003 (0.002) 0.003 (0.002) CEOage + 0.021 (0.017) 0.026 (0.017) 0.024 (0.018) 0.025 (0.018) 0.026 (0.018) 0.027 (0.018) CEOage² - 0.000 (0.000) 0.000 * (0.000) 0.000 (0.000) 0.000 (0.000) 0.000 * (0.000) 0.000 * (0.000) Constant -0.443 (0.448) -0.597 (0.439) -0.501 (0.471) -0.507 (0.470) -0.568 (0.459) -0.577 (0.458)
Number of firm-year observations 284 282 224 224 223 223
R-squared 0.309 0.332 0.329 0.331 0.351 0.352
Adj. R-squared 0.286 0.307 0.298 0.299 0.317 0.318
F-value 13.602 *** 13.441 *** 10.446 *** 10.517 *** 10.355 *** 10.422 ***
that also found a positive association between firm size and CEO compensation. The human capital theory is one of the theories that is used to explain this relationship. This theory states that larger companies are willing to pay higher levels of compensation for managers who possess the specific managerial skills to lead a company that is, based on the size, complex to manage.
Besides, in table 9 the results of regression are presented, using the share of short-term incentives as a dependent variable. In contrast to the results of the robustness test discussed before, the results of table 9 are not similar to those found for the sample presented in table 7. Firstly, the explanatory variable Dummy_Chairman seems to be not associated with the share of short term incentives when Ln_Employees is added to the model as a control variable. This result is completely in line with table 6, that described that the variable Dummy_Chairman is not significantly related with the share of short-term incentives and has no explanatory power at all. Secondly, with respect to the control variables, one can see that Ln_Employees is significantly related, at the 1% significance level, with the share of short-term incentives within total CEO compensation. However, with exception of the Tobinsq measure, other control variables seem not to be robust and consistent with the results discussed in table 7. For example, the performance measure ROE is not statically significant anymore and the results specify that there is a non-linear relationship with CEOage. The coefficients of the latter relationship indicate that the age of a CEO is first positively related to the share of short-term incentives within the compensation scheme and that at a certain age this share remains the same. However, since the results of the last two relationships discussed are uncertain over the different regression models, further study is necessary to generalize the outcomes.
TABLE 8
Robustness test results for total CEO compensation (Ln_CEOtotal), replacing correlated variables.
Expected sign 1 2 3 4 5 6
Explanatory variables
Dummy_Cross list + (H1a) 0.563 *** (0.177) 0.396 ** (0.161) 0.331 ** (0.160)
Dummy_Membership + (H2a) -0.121 (0.173) -0.210 (0.176) Dummy_Chairman + (H3a) 0.330 (0.258) 0.240 (0.261) Control variables Tobinsq + 0.148 * (0.086) 0.104 (0.086) 0.182 ** (0.081) 0.157 * (0.080) 0.157 * (0.081) 0.136 * (0.081) ROE + -0.174 (0.327) -0.137 (0.324) -0.237 (0.394) -0.212 (0.392) -0.212 (0.391) -0.195 (0.391) Ln_Employees +/- 0.255 *** (0.056) 0.202 *** (0.057) 0.259 *** (0.054) 0.256 *** (0.054) 0.234 *** (0.055) 0.235 *** (0.055) Dummy_CEOchairman + 0.168 (0.235) 0.256 (0.235) 0.180 (0.227) 0.071 (0.231) 0.248 (0.227) 0.128 (0.232) CEOtenure + 0.035 ** (0.014) 0.037 *** (0.014) 0.032 ** (0.014) 0.034 ** (0.014) 0.031 ** (0.014) 0.034 ** (0.014) CEOage + 0.158 (0.136) 0.208 (0.135) 0.129 (0.121) 0.131 (0.120) 0.157 (0.121) 0.157 (0.121) CEOage² - -0.002 (0.001) -0.002 * (0.001) -0.001 (0.001) -0.001 (0.001) -0.002 (0.001) -0.002 (0.001) Constant 7.267 * (3.691) 6.576 * (3.660) 7.831 ** (3.252) 7.738 ** (3.243) 7.398 ** (3.234) 7.340 ** (3.240)
Number of firm-year obs ervations 287 284 225 225 223 223
R-squared 0.097 0.128 0.139 0.143 0.162 0.158
Adj. R-squared 0.075 0.103 0.107 0.111 0.126 0.123
F-value 4.304 *** 5.041 *** 4.364 *** 4.502 *** 4.564 *** 4.454 ***
