Anglo-Saxon influence on CEO
compensation: An empirical analysis of
European panel data
By: Sjoukje Tadema
Student number: s1791370
Date: 21th June 2013
Supervisor: Prof. dr. C.L.M Hermes
Master thesis MSc Finance and
MSc Business Administration: Organization and Management Control Faculty of Economics and Business
University of Groningen
I would like to thank prof. dr. C.L.M Hermes and dr. R.B.H. Hooghiemstra for their excellent research assistance and useful comments. In addition, I would like to thank Gerard Land, Wouter van Marle and my family for their support.
ABSTRACT: This paper tests the relationship between direct and indirect Anglo-Saxon influence on CEO compensation, by using panel data for 740 CEO-year observations gathered from 2001-2011. Direct Anglo-Saxon influence is measured by Anglo-Saxon board membership or by having and Anglo-Saxon member in the compensation committee. Indirect Anglo-Saxon influence is measured by a cross listing on an Anglo-Saxon exchange. I conclude that CEO compensation is subject to direct Anglo-Saxon influence, but only when there is an Anglo-Saxon member in the compensation committee. Furthermore, I find that CEO compensation is positively associated with indirect Anglo-Saxon influence.
JEL classification: G34
2
1. Introduction
Within the field of corporate governance, Chief Executive Officer (CEO) compensation has been addressed very extensively. A specific field has focused on the internationalization of the company and its impact on CEO compensation (see: Sanders and Carpenter, 1998; Chitra, Reije and Rajaram, 2010). Internationalization may among other things take place through international board membership. However, it seems that the impact of international board membership on CEO compensation has focused solely on simply having foreign directors (Masculis, Wang and Xie, 2012). The impact of having directors from specific countries or corporate governance systems has generally been ignored. Internationalization can also take place via the capital market. Companies residing in one country might enter the capital market, via a cross–listing, in another country. Baker and McKenzie (2012) have witnessed an increase in cross-border exchange listings and expect this to continue in the future. Therefore, it is important to study both the effect of international board membership and cross-border exchange listings on CEO compensation.
In the corporate governance literature a distinction between corporate governance systems is often made. In most research the Anglo-Saxon system is contrasted with the Continental Europe system (see: La Porta, Lopez-de-Silanes, Shleifer and Vishny, 1998; Rubach and Sebora, 1998; Becht and Röel, 1999; Cernat, 2004). Lucier, Schuyt and Handa (2004) identify the Anglo-Saxon corporate governance system as the most demanding system. The Anglo-Saxon system is known for its reliance on a strong market for corporate control and a focus on performance-dependent compensation (Weimer and Pape, 1999). Since these characteristics have an impact on CEO compensation, it is especially interesting to focus on Anglo-Saxon influence on CEO compensation in Europe. In this research having an Anglo-Saxon board member is a measure of direct Anglo-Saxon influence, while indirect Anglo-Saxon influence is measured by having a cross-listing in an Anglo-Saxon country.
It is hypothesized, that direct as well as indirect Anglo-Saxon influence results in higher CEO compensation and higher performance-dependent compensation. The higher compensation reflects that the Anglo-Saxon corporate governance system is less tolerant of poor performance. CEOs might require a risk premium for the increased risk of dismissal, and therefore CEO compensation is higher (Oxelheim and Randøy, 2005). In addition, Anglo-Saxon board members might bring with them Anglo-Saxon corporate governance practices which are common in the Anglo-Saxon corporate governance system.
3 However, it might be the composition of the compensation committee that determines the level and composition of the CEO’s compensation package, since this is where the majority of the decision making regarding CEO compensation takes place (Kesner, 1994). The results from earlier research are very inconclusive about the influence of the compensation committee on CEO compensation. Therefore, this research especially takes into account the impact of having an Anglo-Saxon member in the compensation committee, as opposed to Oxelheim and Randøy (2005) who only focus on Anglo-Saxon influence in the board as a whole. It is one of the major contributions of this paper that especially the composition of the compensation committee is taken into account.
Another major contribution of this paper regards the large sample size used and the panel feature of the dataset. Past research on this topic by Oxelheim and Randøy (2005) is limited by relying on cross-sectional data. The panel data analysis allows me to get more robust parameter estimates, and therefore improves upon the method used by Oxelheim and Randøy (2005). Moreover, this research does not only focus on the level of CEO compensation, but also on the composition of the compensation package, by examining performance-dependent compensation. In addition, I am able to distinguish between countries, since the dataset contains information for thirteen European countries. It is often assumed that the law system has an impact on corporate governance (Aguilera and Jackson, 2003). Therefore, I also test whether Anglo-Saxon influence on CEO compensation differs between different law systems within Europe.
The empirical results indicate that being subject to direct and indirect Anglo-Saxon influence leads to an increase in CEO compensation. However, after controlling for firm-fixed effects the results with respect to direct Anglo-Saxon influence only hold when there is an Anglo-Saxon member in the compensation committee. No strong evidence is found with respect to Anglo-Saxon influence on the composition of the compensation package, however this could be due to the fact that long term incentive plans are not taken into account. Taking a broader perspective, this paper provides evidence of corporate governance convergence. Due to corporate actions, like international board membership and cross-listings, corporate governance practices which are common in one country migrate to other countries.
4
2. Literature
2.1.1 Agency Theory
The theoretical basis for CEO compensation lies in agency theory developed by Jensen and Meckling (1976). They state that a separation of ownership and control might result in a so called agency problem. The principal (owner) delegates decision making to the agent (manager), but the agent might not always act in the best interest of the principal. As Fama and Jensen (1983) highlight, a separation of ownership and control might lead to suboptimal decision making by managers in order to serve their own interests and or goals. In addition, as emphasized by Eisenhardt (1989), there is a problem with the different attitudes to risk of the principal and the agent. It is generally believed that the agent is more risk averse than the principal. The agency costs that arise from these agency problems are described as: “the costs of structuring, monitoring, and bonding a set of contracts among agents with conflicting interests” (Fama and Jensen, 1983, p.306). Agency costs could, for instance, arise from managers holding excess cash to pursue their own objectives, and thus not acting in the principal’s best interest (Jensen, 1986). Within the literature of agency theory CEO compensation is seen as an effective mechanism to align the interests of shareholders and management.
5 From this it can be concluded that, CEO compensation is a mechanism which is used to align the interest of agents and principals. Principals, who are generally less risk averse than principals, try to minimize the agency costs by setting appropriate compensation packages.
2.1.2 The managerial power approach
Even though not the major focus of this paper, I introduce the managerial power approach in this section. It is an important compensation theory which will be referred to later and leads to the formulation of the most important control variables.
Bebchuk and Fried (2003) have identified several compensation package characteristics which are not consistent with optimal contracting theory. They for instance, find that executive compensation is only weakly related to managerial performance. This is in line with findings from Jensen and Murphy (1990), who studied performance compensation for over 2,000 CEOs. They state that the relationship between shareholder wealth and CEO wealth is small and has fallen dramatically in the last 50 years. On average, CEO wealth changes with $3.25 for every $1,000 change in shareholder wealth. Morse, Nanda and Seru (2011) show that some powerful CEOs in the U.S. rig incentive pay, by shifting the weight of performance pay towards the better performing measures, thereby deviating what would be expected from an optimal contract. Tosi, Werner, Katz and Gomez-Meija (2000) found that indicators of company size are significantly more important in explaining CEO pay than performance. More than 40% of the variance of CEO compensation can be explained by company size, while company performance accounts for less than 5% of the variance.
The managerial power approach which Bebchuk and Fried (2003) embrace, assumes that executives have substantial power over boards which are responsible for the compensation package. CEOs have control over the pay-setting process and are therefore able to have a large influence on their own compensation. As a result, CEO compensation should not only be considered as a remedy against the principal-agent problem described earlier, but also as part of the agency problem itself. The failure to find a link between compensation and performance has motivated researchers to develop new models in which compensation is set by CEOs instead of shareholders (e.g. Kuhnen and Zwiebel, 2008).
6 compensation outcomes, especially those compensation packages that link pay to performance. Furthermore, Edmans and Gabaix (2009) show that the results from Bebchuk and Fried might be the result of the simplicity of their model. Even though, some aspects of the compensation package might seem not to be in line with optimal contracting theory, several complexities are not taken care of which might when included in the model, be fully consistent with efficient compensation packages.
From this it can be concluded that, despite CEO compensation is often used to align the interests of principals and agents, CEOs might have an influence on their own compensation packages. A CEO who has substantial power over the board might be able to set compensation packages in such a way which is beneficial to him.
2.2 The compensation committee
Focusing on CEO compensation, some corporate governance studies emphasize the role of the compensation committee. The compensation committee is a subcommittee of the board of directors and responsible for determining CEO compensation (Baker, Jensen and Murphy, 1988). Particularly the relationship between compensation committee quality, measured by committee independence (e.g. the fraction of independent directors on the compensation committee), and CEO compensation and performance-dependent compensation are examined. However, the results from these studies are mixed. Newman and Mozes (1999) cannot find a relationship between CEO compensation and compensation committee independence. However, under certain circumstances, incentive based pay is favored to the CEO when there are insiders in the compensation committee. Daily et al. (1998) do not find a relationship between the nature of the compensation committee and the level and composition of CEO compensation. Anderson and Bizjak (2003) find that more independent compensation committees do not lead to a higher association between CEO compensation and stock returns. In contrast to these findings, Conyon and Peck (1998) find that the ratio of outside directors in a compensation committee is positively related to the sensitivity of pay to performance. Sun and Cahan (2009) show by using a comprehensive measure of compensation committee quality1, a positive relationship between compensation committee quality, and the relationship between accounting earnings and CEO compensation. From this it can be concluded that, the results with respect to the influence of the composition of the compensation committee on CEO compensation are very inconclusive.
1
7 It can be argued that the composition of the compensation committee has an impact on CEO compensation, since this is where the majority of the decision making regarding CEO compensation occurs (Kesner, 1994). However, to date no papers have focused on whether having an Anglo-Saxon board member in the compensation committee has an impact on CEO compensation. This could be the reason why the results from earlier studies are inconclusive. It can be assumed that a board member which is a member of the compensation committee has more involvement in setting CEO compensation than a board member which is not a member of the compensation committee.
2.3 Hypotheses
This paper tests several hypotheses. The first analysis focuses on direct Anglo-Saxon influence and uses an Anglo-Saxon board member as the unit of observation. The impact on CEO compensation is examined. Subsequently, this paper analyses the impact of having a cross-listing in an Anglo-Saxon country on CEO compensation. As stated before, this paper focuses on Anglo-Saxon corporate governance influence. Therefore, the independent variables capture influence from all countries which are characterized by an Anglo-Saxon corporate governance system: United States, United Kingdom, Canada and Australia (Weimer and Pape, 1999). So, a company is subject to indirect or direct Anglo-Saxon influence, when it is has one or more board members from, or is listed on an exchange, from one of these countries.
8 Cernat, 2004). So, in the Anglo-Saxon corporate governance system agency problems are partly solved by a strong external market for corporate control. While in Anglo-Saxon countries acquisitions are viewed as an efficient corporate governance mechanism which substitutes inefficient teams for more efficient ones, in Continental Europe acquisitions are often considered as being based on greed and myopia (Tirole, 2006). Being exposed to Anglo-Saxon corporate governance increases the risk of dismissal and therefore from an agency perspective the risk averse CEO requires a risk premium. In addition, performance-dependent compensation is an important characteristic of the Anglo-Saxon corporate governance system. Hall and Murphy (2002) show that the increased risk associated with performance-based pay might also result in CEOs demanding a risk premium.
Via these two channels agency theory provides the following hypotheses:
Hypothesis 1a: There is a positive relationship between Anglo-Saxon board membership and the level of CEO compensation in European companies.
Hypothesis 1b: There is a positive relationship between Anglo-Saxon board membership and performance-dependent CEO compensation in European companies.
As mentioned earlier, it might not be the board as a whole that has a major impact on CEO compensation, but the compensation committee who determines CEO compensation. Therefore, the sources of influence which have been identified before are also tested for having an Anglo-Saxon board member in the compensation committee. The following hypotheses are tested:
Hypothesis 2a: There is a positive relationship between having an Anglo-Saxon board member in the compensation committee and the level of CEO compensation in European companies.
Hypothesis 2b: There is a positive relationship between having an Anglo-Saxon board member in the compensation committee and performance-dependent compensation in European companies.
9 The Anglo-Saxon investors might also simply bring with them corporate governance practices which are common in the Anglo-Saxon system. Therefore, due to institutional spillover effects it can be expected that CEOs serving on a company with an Anglo-Saxon exchange listing earn higher and more performance-dependent compensation. So, from agency theory it can be expected that, CEOs serving on a company with an Anglo-Saxon exchange listing receive higher and more performance-dependent compensation.
However, the reasons why a cross-listing impacts upon CEO compensation go beyond agency theory. Labour market dynamics could also play a role with respect to the level of CEO compensation. As Finkelstein and Hambrick report: “CEO compensation can be thought of as a response to the marketplace for CEO talent. Depending on economic conditions and the relative supply and demand for CEO-caliber executives, a CEO’s compensation may be bid up by potential suitors” (1988, p. 546). Murphy and Zabojnik (2004) also take a market based view, and propose that the large increases in CEO compensation over the past three decades is not due to increased managerial power as proposed by Bebchuk and Fried (2003), but due to a change in the composition in the managerial skills needed to manage a modern corporation. The internationalization of the company might be one of the factors that affects the skills needed by CEOs, and alter the demand and supply conditions that determine the level of CEO compensation (Oxelheim and Randøy, 2005).
10 Human capital theory provides the basis for how the supply of potential CEOs impacts on CEO compensation. International companies require more capable CEOs and therefore the supply of CEOs is reduced. In other words, there is a limited supply of CEOs who are capable of dealing with the international financial environment in which companies operate who have a cross-border exchange listing. From this it follows that supply conditions are another channel through which CEO compensation is impacted.
It can be stated that both influences have a positive impact on the level of CEO compensation. The CEO has to meet higher standards when it wants to serve on a company which has an exchange listing in an Anglo-Saxon country, therefore CEO compensation is higher. Hence, agency theory as well as labour market dynamics are able to explain why CEOs subject to indirect Anglo-Saxon influence receive higher compensation. The reason why CEOs subject to indirect Anglo-Saxon influence receive higher performance-dependent compensation focus solely on agency theory. From this discussion the following hypotheses can be formulated:
Hypothesis 3a: There is a positive relationship between Anglo-Saxon exchange listing and the level of CEO compensation in European companies.
Hypothesis 3b: There is a positive relationship between Anglo-Saxon exchange listing and performance-dependent compensation in European companies.
3. Data
This study examines CEO compensation from 2001 to 2011 for thirteen European countries (Austria, Belgium, Denmark, France, Germany, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden and Switzerland) by the means of a panel analysis. The sample consists of an unbalanced panel with 740 CEO-year observations.
3.1 Construction of the dataset
BoardEx is used as the major source for obtaining the relevant governance data. BoardEx is a large dataset which includes in-depth profiles of board members of quoted companies and large private companies. The procedure to come up with a workable dataset for this research is structured as follows. First, all director-years2 in which the board member held a directorship at a company in which financial transactions are the most important source of business are removed from the dataset. This means that companies in the banking, insurance, private equity, real estate and specialty and other finance industries are excluded. Second, all director years are removed for which
11 the stated board size for a given year does not correspond to the actual number of board members available in the dataset. Since this study focuses on the impact of Anglo-Saxon board members on compensation I need the nationalities of all the supervisory directors. Consequently, all director years for which for one of the supervisory directors the nationality was not available, are removed from the dataset. Then I identify the CEO of each board. All board members indicated by “Acting CEO”, “CEO”, “Group CEO” or “Chairman Executive” are automatically recognized as CEO. For the board-years3 in which still no CEO is identified, I manually try to identify the CEO. In some instances this is not possible, and the board-years are removed from the dataset. Then, for each board-year I indicate, by the means of a dummy variable, whether there is an Anglo-Saxon supervisory director on the board and whether there is an Anglo-Saxon supervisory directors in the remuneration or compensation committee.
Then, I remove all the data for non-CEOs. For the CEOs for which relevant data was missing I try to fill the gaps. For instance, age is not available for some CEOs. I search for the missing information in annual reports and on the internet (for instance, LinkedIn). In some instances I contact the company by e-mail. Missing compensation data is searched for in annual report or corporate governance statements. For compensation which is not quoted in euros, the exchange rate effective at the closing date of the annual report year is taken to calculate compensation in euros. If no compensation data is found the CEO is removed from the dataset.
In addition, I check for abnormalities in the dataset. For instance, CEO-years in which the CEO appeared to earn total compensation less than 50 thousand euros or more than 5 million euros, the numbers are checked and adjusted if necessary. I adjust CEO compensation when the CEO joined his role in a given company-year. Compensation is then scaled to (or added up to) one year, as if the CEO has served on the board for the whole year. So, if a CEO is in his role for 0.3 years, compensation is multiplied by 3,33 to determine total annual compensation.
The remaining dataset is merged with financial data from Compustat. For financial data not stated in euros, the exchange rate effective at the closing date of the financial year is used to calculate euro amounts. Lastly, the dataset is extended with cross-listing data from Orbis. In all the available databases at my disposal only listing data for the last available report year is available (2012). I add a dummy variable whether in 2012 the company on which the CEO serves is listed on an Anglo-Saxon exchange or not. After this procedure the dataset contains a final sample size of n=740 (CEO-years).
12 These CEOs serve on 231 different European companies in the period 2001-2011. This means that on average there are 3.2 observations per company, an indication of an unbalanced dataset.
3.2 Variables
3.2.1 Direct Anglo-Saxon influence
Direct Anglo-Saxon influence is measured by Anglo-Saxon board member. The dummy explanatory variable takes the value 1 if one or more supervisory board member(s) hold(s) an Anglo-Saxon citizenship, and 0 otherwise. In addition, direct Anglo-Saxon influence is measured by the dummy variable Anglo-Saxon board member in compensation committee.
3.2.2 Indirect Anglo-Saxon influence
Indirect Anglo-Saxon influence is measured by Anglo-Saxon exchange listing. The dummy explanatory variable takes the value 1 if the company is listed on a stock exchange in a capital market with an Anglo-Saxon background, and 0 otherwise. The category of cross-listed companies include companies which trade on the Australian Securities Exchange (ASX), New York Stock Exchange (NYSE), Nasdaq, the National Stock Exchange of Australia, Toronto Stock Exchange (TSE) and the London Stock Exchange (LSE).
3.2.3 CEO compensation
I define total CEO compensation as the sum of salary, bonus, pension and other compensation. Long term incentive plans (LTIP) are left out, because the dataset only contains information on LTIPs for 89 (12.03%) CEOs. This is a major restriction of the research. Besides a measure of total compensation, this paper also focuses on the composition of the compensation package. In this case the focus is on the performance-dependent part of CEO compensation. Performance-dependent compensation is measured as the ratio between variable and fixed pay. In some studies having CEO compensation as the dependent variable, the independent variables are lagged one year. However, since this study takes direct compensation into consideration it does not seem necessary to lag the independent variables with one year.
3.3.3 Control variables
13 of larger companies receive higher compensation than CEOs of smaller companies. Finkelstein and Hambrick (1989), also following the marginal product argument, and state that CEOs of larger companies receive higher compensation because they oversee relatively more resources and therefore their actions have a greater dollar impact. Several empirical studies have shown that company size has a significant impact on CEO compensation (e.g. Baker et al., 1988; Kostiuk, 1990; Murphy, 1999). Furthermore, Finkelstein and Hambrick (1988) report that pay consultants use company size as a major determinant of CEO pay. Since the relationship between company size and CEO compensation has been one which is most widely recognized in the literature, this paper controls for company size as a determinant of CEO compensation. Company size is expected to positively impact CEO compensation.
Much of the management compensation literature focuses on company performance as a determinant of CEO compensation. From a theoretical as well as empirical point of view there is ample support for a positive relationship between company performance and CEO compensation. The theoretical background for the proposed positive relationship lies, as explained earlier, in agency theory. It is asserted that pay-for-performance structures of compensation align the interests of shareholders with agents, and therefore the agency problem can be solved. Empirically, the results have been mixed. Jensen and Murphy (1990) show that overall a small but positive relationship between pay and performance. Contrasting to what would be expected from agency theory, it seems that company size is able to explain much more of the variance of CEO compensation than company performance (Lewellen and Huntsman, 1970; Murphy, 1985; Tosi et al., 2000). This is a strong indication of the existence of an agency problem, since it seems that CEOs are more rewarded by growing their business and maximizing company size than by maximizing shareholder value.
14 strategic vision and solutions, and therefore the CEO’s efficacy declines. This finding is consistent with a curvilinear relationship between CEO tenure and CEO compensation.
Also the board characteristic board size is widely recognized as impacting CEO compensation. Larger boards are often associated with weaker governance structures, because of lower monitoring effectiveness and higher CEO power. Yermack (1996) finds that smaller boards are more effective and provide CEOs with more incentive based pay. Core, Holthausen and Larcker (1999) find that CEOs receive higher compensation when boards are larger. Lastly, research focusing on CEO compensation often controls for CEO gender (Brick, Palmon and Wald, 2006; The findings by Bertrand and Hallock (2001) suggest that women in executive positions earn on average 45% less than men.
From this discussion I identify the following control variables to minimize specification bias. Company size is measured by total assets. To ensure normality a log transformation is undertaken. Company performance, or Return on Assets (ROA), is measured as the ratio of the accounting performance measure, Earnings before Interest and Taxes (EBIT) to the book value total assets4. EBIT is included as an absolute measure of performance. Since EBIT can take non positive numbers, an inverse hyperbolic sine transformation is undertaken5. CEO tenure is measured as the number of years the CEO has served as CEO of the company. Because of the expected curvilinear nature of the relationship between tenure and CEO compensation as advocated by Hambrick and Finkelstein (1995), this paper also controls for tenure squared. CEO age is used as a measure of CEO experience. Board size is measured by adding all executive directors and non-executive directors of the company. Lastly, CEO gender is a dummy variable taking the value of 1 if the CEO is a male, and 0 if the CEO is a female. All variables used in this study are defined in table I.
4 I recognize that from an agency theory perspective, a stock market performance measure should be included
as well, but the dataset does not allow me to include such a measure since this data is not available for all of the companies in the dataset.
5
15
Table I: Definition of the variables
Variable Description
Anglo-Saxon board member This variable indicates whether the board includes at least one supervisory board member with an Anglo-Saxon nationality; value 1 if such board member (s) exist (s), 0 otherwise.
Anglo-Saxon board member in compensation committee
This variables indicates whether the compensation committee includes at least one independent board member with an Anglo-Saxon nationality; value 1 if such board member (s) exist(s), 0 otherwise.
Anglo-Saxon exchange listing This variables shows whether the company has an Anglo-Saxon exchange listing; value 1 if such exchange listing exists, 0 otherwise.
Company size Company size is measured by the natural log of total assets . Company performance
(EBIT)
This variable indicates the natural log of Earnings before Interest and Taxes and is a measure of accounting performance which reflects operating income. Company performance
(ROA)
Company performance is also measured by accounting return on assets. The ratio of Earnings before interest and taxes over total assets.
CEO tenure CEO tenure is measured by the time the CEO has served in his role. It is measured absolute and squared.
CEO age This variable indicates the age of the CEO at the closing date of the annual report year.
CEO gender This is a dummy variable indicating whether the CEO is a male or female; value 1 if the CEO is a male, and 0 otherwise.
Board size Board size is measured by the total number of executive and non-executive board members.
CEO compensation
Salary Salary indicates the fixed part of annual direct compensation.
Bonus This variable indicates the variable part of annual direct compensation. Pension Pension is the contribution of the company to pension plan of the CEO. Other Other indicates compensation in other kinds, the so called “perks”. Performance-dependent
compensation
16
3.3 Descriptive statistics
Total compensation is divided into four categories: salary, bonus, pension and other. Average compensation for the whole sample can be found in table II. It is demonstrated that average total CEO compensation over the period 2001-2011 is 834 thousand euros per year6. On average, bonus payments account for 31.15% of total compensation. Table III presents the descriptive statistics for all the variables used in this research. Average total assets is 3,313.95 million euros, while average ROA is 0.250. The average CEO is 52.44 years old and serves 6.35 years in his role as CEO. Furthermore, the average board size in the sample is 8.37.
Total CEO compensation has increased, in nominal terms, by 70.84% over time, rising from 631 thousand euros in the period 2001-2005 (n=404) to 1,078 thousand euros in the period 2006-2011 (n=336). Figure 1 in the Appendix shows the evolution of annual compensation and performance-dependent compensation from 2001 to 2011. Both variables show a positive trend. Average CEO compensation decreases in 2008.
The proportion of companies having an Anglo-Saxon board member has increased over time, from 89 (28.25%) for the period 2001-2005 to 94 (38.52%) for the period 2006-2011. In the period 2001-2005, 35 (14.34%) CEOs were serving on a board with an Anglo-Saxon board member in the compensation committee, while in the period 2006-2011 this number was 47 (19.26%).
Table II: Average CEO compensation
Average direct CEO compensation. Salary is the fixed part of direct annual compensation. Bonus is the variable part of direct annual compensation. Pension is the annual contribution of the company to the pension plan of the CEO for a year. Other is all annual direct compensation in other kind, the so called “perks”. Average compensation is indicated in thousand euros. Average % of total compensation shows the percentage of each compensation component as part of total direct compensation.
Average compensation Average % of total compensation
Salary 474.356 56.876%
Bonus 259.826 31.153%
Pension 56.438 6.767%
Other 43.401 5.204%
Total direct compensation 834.021 -
6
17 Table III: Summary statistics
This table reports the number of observations, mean, median, standard deviation, minimum and maximum for each of the variables. The sample consists of an unbalanced panel of 740 CEO-company-years from 231 European companies for the period 2001-2011. All CEO level and board level data are retrieved from BoardEX or annual reports. BoardEx contains in-depth profiles of board members of quoted companies and large private companies. The dataset is merged with financial data from COMPUSTAT. Anglo-Saxon exchange listing is a dummy variable indicating whether the company on which the CEO serves has an Anglo-Saxon exchange listing in 2012 or not. Company size is measured by total assets, and is shown in million euros. Company performance is shown as EBIT (in millions) and as Return on Assets (ROA). CEO age indicates the age of the CEO at the closing date of the annual report year. CEO gender is a dummy variable indicating 1 for male and 0 for female. CEO tenure is measured by the time the CEO has served in his role. Anglo-Saxon board member, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member on the board. Anglo-Saxon board member in compensation committee, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member in the compensation committee. Board size is measured by the total number of executive and non-executive directors on the board. Other refers to compensation in other kinds, the so called “perks”. Total compensation is the sum of salary, bonus, pension and other compensation. All compensation components are reported in thousand euros. Performance-dependent compensation is measured as the ratio between variable and fixed (salary) compensation.
Variable N Mean Median St. dev. Min Max
Company characteristics
Anglo-Saxon exchange listing 740 0.368 0 0.482 0 1
Company size (total assets) 740 3,312.949 410.724 9,453.085 2.815 74714.130 Company performance (EBIT) 740 367.506 29.177 1188.053 -2,452.564 13,222.000 Company performance (ROA) 740 0.250 0.070 1.247 -7.984 7.445
CEO characteristics
CEO Age 740 52.441 52.000 8.011 33.000 76.000
CEO gender 740 0.977 1.000 0.150 0 1
CEO tenure (years) 740 6.347 4.200 6.673 0.200 42.900
Board characteristics
Anglo-Saxon board member 740 0.246 0.000 0.431 0 1
18 In table IV average compensation is shown separately for each country and law system. For Austria, Luxembourg and Spain only two CEO-year observations are available. It can be seen that CEO compensation in the group of countries characterized by German Civil law (1,177 thousand euros) is higher than CEO compensation in Scandinavian Civil law (669 thousand euros) and French Civil law (798 thousand euros). Furthermore, performance-dependent compensation is higher in German Civil law countries than in Scandinavian and French Civil law countries.
Table IV: Average total compensation for each country and law system
The sample consists of an unbalanced panel of 740 CEO-company-years from 231 European companies for the period 2001-2011. N indicates the number of CEO-year observations per country or law system. Average total compensation is measured by the sum of salary, bonus, pension and other compensation, and is measured in thousand euros. Performance-dependent compensation is measured by the ratio of bonus to bonus plus salary.
Country N Average total compensation Performance-dependent compensation
Austria 2 511 0.191
Germany 61 1,324 0.467
Switzerland 55 1,037 0.277
Average German Civil law 118 1,177 0.374
Denmark 7 943 0.188
Norway 31 667 0.075
Sweden 103 651 0.161
Average Scandinavian Civil law 141 669 0.143
Belgium 10 403 0.100 Italy 23 1,089 0.112 France 256 771 0.161 Luxembourg 2 542 0.589 Netherlands 182 802 0.260 Portugal 6 758 0.374 Spain 2 2,883 0.506
Average French Civil law 481 798 0.201
I test whether the explanatory variables are orthogonal to one another, or whether multicollinearity exists, by constructing a correlation matrix of the independent variables. The correlation matrix of the independent variables is shown in Table I in the Appendix. The correlation matrix shows the direction and strength of the linear association of two variables. Not surprisingly, the dummy variables Anglo-Saxon board member and Anglo-Saxon member in compensation committee are highly correlated with each other (0.618). However, this is not problematic since these variables are not regressed simultaneously. The variables EBIT and Board size show near multicollinearity7 with
7
19 total assets (0.571 and 0.572). Since in the regression EBIT is only used as a robustness check for a measure of Company performance it is not an issue in the main analysis. A robustness check will test whether the correlation between Board size and total assets has a significant impact on the results.
3.4 Are companies with direct Anglo-Saxon influence different?
For 182 CEO-years (24.59%) there is a member with an Anglo-Saxon nationality on the board and for 82 CEO-years (11.08%) there is a member with and Anglo-Saxon nationality on the compensation committee. A bivariate analysis is undertaken to test whether CEO total compensation and performance dependent compensation is significantly larger for the CEO-years in which there is an Saxon supervisory director on the board. Average CEO compensation for boards with an Anglo-Saxon supervisory director on the board (subsample I) is 1,256 thousand euros, while average total compensation for boards without an Anglo-Saxon supervisory board member (subsample II) is 622 thousand euros. It can be concluded that average total CEO compensation is significantly different for the two subsamples (1% significance level), and there for the null hypothesis that average total compensation for the two subsamples are the same can be rejected. Also, performance dependent compensation are significant different for the two subsamples. These results are preliminary evidence of a positive relationship between Anglo-Saxon board membership and CEO compensation. However, I cannot expect the two groups to be similar on other aspects than Anglo-Saxon influence. Table V shows the statistical results for the comparison of CEO compensation with and without direct Anglo-Saxon influence.
Table V: Comparison of CEO compensation with and without direct Anglo-Saxon influence
This table shows a comparison of the means of the compensation variables for CEO-years in which the boards have an Anglo-Saxon member and those without an Anglo-Saxon member. The sample consists of an unbalanced panel of 740 CEO-company-years from 231 European companies for the period 2001-2011. Total compensation is measured by the sum of salary, bonus, pension and other compensation, and measured in thousand euros. Performance-dependent compensation is defined as the ratio between variable and fixed (salary) compensation. All tests are performed one-sided. *** indicates significance at the 1% level.
Compensation variable
Mean for CEO-years with Anglo-Saxon board member,
N= 182
Mean for CEO-years without Anglo-Saxon board member,
20
3.5 Are companies with indirect Anglo-Saxon influence different?
For 272 CEO-years (36.76%) the company on which the CEO serves is subject to indirect Anglo-Saxon influence, through an Anglo-Saxon cross-listing. Again an bivariate analysis is undertaken to whether CEO total compensation and performance-dependent compensation is larger for the CEO-years in which the company is listed on an Anglo-Saxon exchange and the CEO-years for which such listing is non-existent. As demonstrated in table VII, total compensation and performance-dependent compensation is higher for CEO-years in which the company is listed on an Anglo-Saxon exchange. These results are statistically significant at the 10% level. Again, this is preliminary evidence of a relationship between indirect Anglo-Saxon influence and CEO compensation. Table VI shows the statistical results for the comparison of CEO compensation with and without indirect Anglo-Saxon influence.
Table VI: Comparison of CEO compensation with and without indirect Anglo-Saxon influence
This table shows a comparison of the means for the compensation variables for CEO-years in which the companies have an Anglo-Saxon cross-listing and companies without such a listing. The sample consists of an unbalanced panel of 740 CEO-company-years from 231 European companies for the period 2001-2011. Total compensation is measured by the sum of salary, bonus, pension and other compensation, and measured in thousand euros. Performance-dependent compensation is defined as the ratio between variable and fixed (salary) compensation. All tests are performed one-sided. * indicates significance at the 10% level.
Compensation variable
Mean for CEO-years with Anglo-Saxon cross-listing,
N= 272
Mean for CEO-years without Anglo-Saxon cross listing,
N= 468
Difference
Total direct compensation (thousands)
901.964 794.533 107.431*
Performance-dependent compensation
21
4. Methodology
4.1 Empirical specification
The relationship between direct and indirect Anglo-Saxon influence and CEO compensation is assessed by the means of a regression analysis. To limit the effect of outliers, a log transformation is undertaken to CEO compensation. The following regression equations are estimated:
+ + ) + + )
22 The regressions test whether having an Anglo-Saxon board member, an Anglo-Saxon board member in the compensation committee or having and Anglo-Saxon exchange listing is associated with higher CEO compensation or more performance-dependent compensation. Then coefficient 1 in the six regression analysis should be positive and significant. To ensure normality and limit the effect of outliers a log transformation is applied to both total assets and total compensation. In all specifications of the model, I include year dummies to account for year-specific shocks that might influence CEO compensation. Furthermore, in all specifications the standard errors are adjusted for heteroskedasticity.
The regressions are first estimated by using ordinary least squares. Then, to exploit the panel feature of the data and to avoid the omitted variable problem I include random or fixed effects. Since Anglo-Saxon exchange listing is a time-invariant variable (or constant across time) no fixed effects can be used in this model because this variable would be multicollinear with the constant.
4.2 Sensitivity analysis and robustness checks
In a sensitivity analysis I test whether outliers drive the results. To reduce the effects of outliers the extreme values are limited by the means of a winsorising procedure. All data below the 5th percentile are set to the 5th percentile and all data above the 95th percentile are set to the 95th percentile.
As stated earlier, board size and total assets are highly collinear (0.572). Therefore, I test whether by removing board size the values of the coefficients on the other variables change. Since it appears that the gender dummy shows very little variation (98% male) and since it can be expected that this variable is a near time-invariant, I also test whether excluding this variable impacts on the regression results. Furthermore, I test whether including country dummies change the results for the variables I want to study.
Lastly, I check whether the results interact with the country the company is located in. So the effect of direct or indirect Anglo-Saxon influence on CEO compensation in a specific group of countries is measured by the means of an interaction variable. The advantage of this procedure is that I can still use the whole sample for this analysis.
23 Continental Europe corporate governance. However, Aguilera and Jackson (2003) acknowledge that differences exist between countries in Europe. They were among the first who were able to conceptualize national diversity in corporate governance. If one wants to study Anglo-Saxon influence on European countries, as is done in this research, it might be necessary to take these differences into account. Aguilera and Jackson (2003) show in their model that legal systems shape the nature of corporate governance and that agency theory fails to take differences between countries into account. Therefore, a law system can be acknowledged as an important external corporate governance mechanism.
Law systems broadly are defined based on their legal origin into Civil law and Common law. European countries are commonly assigned to the Civil law group. Three groups of Civil law systems in Europe are identified: 1) French; 2) German and 3) Scandinavian. La Porta et al. (1998) show that different law systems and the quality of its enforcement lead to different rights for the owners of companies and that Common law countries generally have the strongest legal shareholder protection. The rights of owners become critical when managers act in their own interest, for instance when shareholders with large voting power can more easily vote out directors who are not acting in their best interest. From this it can be concluded that investors rights might help explain corporate governance variables. Also when one studies Anglo-Saxon influence on CEO compensation in Europe, it might be necessary to take the law system into account. Table VII shows to which law system the countries in the dataset belong.
Table VII: A distinction between European countries based on their law system
German Civil law Scandinavian civil law French civil law
Austria Denmark Belgium
Germany Norway Italy
Switzerland Sweden France
24
5. Results
Table VIII shows the results of the ordinary least squares regression. By examining table VIII it can be concluded that hypothesis 1.a, 2.a and 3.a are supported. In line with expectations, it is shown that direct as well as indirect Anglo-Saxon influence is positively associated with CEO compensation, after controlling for other factors which explain CEO compensation. CEOs who serve on boards with an Anglo-Saxon board member earn on average 10.4% more compensation than CEOs who serve on boards without an Anglo-Saxon board member. In addition, having an Anglo-Saxon member in the compensation committee increases compensation with 40.4%. Having an Anglo-Saxon exchange listing results in 15.7% higher compensation. The coefficients of Anglo-Saxon board member in compensation committee and Anglo-Saxon exchange listing are significant at the 1% level, while Anglo-Saxon board member is significant at the 10% level.
As suggested by hypotheses 2.b and 3.b, I find a positive and significant relationship between having an Anglo-Saxon compensation committee member and having an Anglo-Saxon exchange listing, and performance-dependent compensation. I also find that having an Anglo-Saxon board member is positively associated with performance-dependent compensation, however this result is not statistically significant. Therefore, I do not find support for hypothesis 1.b.
As expected, the control variables Company size and Board size, show a positive and significant relationship with CEO compensation. The significant negative relationship between Company performance and CEO compensation is rather surprising. In an unreported analysis I use EBIT (IHS) as a performance measure instead of ROA. This does not change the main results. I also test the robustness of the results, by excluding the variables Board size and CEO gender. This also does not have an impact on the major results. The results for the regression analysis with country dummies are reported in table II in the Appendix. As can be seen in this table, including country dummies does not change the results of the regression analysis.
25 Table VIII: Ordinary least squares regressions of direct and indirect Anglo-Saxon influence on CEO compensation
The table shows results from ordinary least squares regressions of total CEO compensation and performance-dependent compensation for European companies for the period 2001-2011. Total CEO compensation is measured as the sum of salary, bonus, pension and other compensation. A log transformation is applied to total compensation. Performance-dependent compensation is defined as the ratio between variable and fixed pay. Anglo-Saxon board member, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member on the board. Anglo-Saxon board member in compensation committee, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member in the compensation committee. The variable Anglo-Saxon exchange listing shows whether in a given CEO-year, the company on which the CEO serves, is listed on an Anglo-Saxon exchange in 2012. The coefficients of the control variables are also given in the table. Company size is measured by the natural logarithm of total assets. ROA is used as an indicator of Company performance. Board size is measured by the total number of executive and non-executive members on the board. CEO gender is a dummy variable. CEO tenure is measured by the time the CEO served in the role of CEO. All regressions include year dummies, but they are not reported. The robust standard errors of the variables are reported in parentheses. Statistical significance at the 10%, 5% and 1% level is indicated by *, ** and ***, respectively.
Total CEO compensation Performance-dependent compensation
Independent variable (1) (2) (3) (4) (5) (6)
Anglo-Saxon board member 0.104* 0.010
(0.059) (0.018)
Anglo-Saxon board member
in compensation committee 0.404*** 0.079***
(0.075) (0.014)
Anglo-Saxon exchange listing 0.157*** 0.081***
(0.030) (0.011) Company size (ln) 0.249*** 0.233*** 0.246*** 0.028*** 0.025*** 0.024*** (0.020) (0.020) (0.020) (0.004) (0.004) (0.004) Company performance -0.051*** -0.050*** -0.048*** -0.012** -0.011** -0.009* (0.010) (0.009) (0.010) (0.005) (0.005) (0.005) Board size 0.025*** 0.029*** 0.022*** 0.014*** 0.015*** 0.012*** (0.007) (0.007) (0.007) (0.001) (0.002) (0.002) CEO age -0.004 -0.003 -0.004 -0.001 -0.001 -0.001 (0.003) (0.003) (0.003) (0.002) (0.002) (0.002) CEO gender 0.240 0.204 0.211 0.147*** 0.138*** 0.121** (0.184) (0.174) (0.174) (0.052) (0.052) (0.049) CEO tenure -0.004 0.001 -0.006 0.005* 0.006* 0.004 (0.006) (0.006) (0.006) (0.003) (0.003) (0.003)
CEO tenure squared 0.000 0.000 0.000 0.000** 0.000* 0.000**
(0.000) (0.000) (0.000) (0.000) (0.001) (0.000)
Constant 4.868*** 4.891*** 4.927*** -0.115 -0.109 -0.077
(0.191) (0.183) (0.191) (0.077) (0.075) (0.076)
Adjusted R-squared 0.484 0.500 0.488 0.245 0.254 0.267
26
4.2 Regression with firm-fixed effects
To exploit the panel feature of the data, I also include random or fixed effects in the model. In the models with Anglo-Saxon exchange listing as an independent variable no fixed effects can be used, because this variable is time-invariant. For the other models, I perform the Hausman test to assess whether the fixed effects estimator is more consistent than the random effects model. The significance (1%) of the Hausman test for all models indicates that the fixed effects estimator is more consistent than the random effects estimator. Furthermore, I determine whether the fixed effects are necessary by performing a redundant fixed effects test. These tests shows that I need to use the fixed effects model8. The fixed effect estimator captures the idiosyncratic features of each company. So the effect of company time invariant variables, like company culture, are removed by including firm fixed effects. This method is different than the method used by Oxelheim and Randøy (2005), who only use ordinary least squares.
In table IX the results of the regression analysis with fixed effects are shown. As indicated by the adjusted R-squared the predictive power of the model increases significantly by including firm fixed-effects. Only the variable Anglo-Saxon board member in compensation committee remains significant on the level of CEO compensation. The interpretation of this variable is that by having an Anglo-Saxon member in the compensation committee, total CEO compensation increases by 33.33%. So after including firm-fixed effects I only find support for hypothesis 2.a. CEO compensation is influenced through direct Anglo-Saxon influence, however only when there is an Anglo-Saxon member in the compensation committee. The negative coefficients for Anglo-Saxon board member on total CEO compensation and Anglo-Saxon board member in compensation committee on performance-dependent compensation seem counterintuitive. However, the results are not statistically significant, so these variables are unable to explain variation in total CEO compensation and performance-dependent compensation.
As expected, Company size and Company performance have a positive and significant relationship with total CEO compensation. CEO gender is only marginally negatively significant when regressing de dummy variable Anglo-Saxon member in the compensation committee against total CEO compensation. The control variables Board size, CEO age do not have a significant impact CEO compensation.
8
27 CEO tenure and CEO tenure squared are the only control variables which are significant in models 4 and 5. These two variables show a positive relationship with performance-dependent compensation. All results are robust to excluding CEO gender. Including or excluding this dummy variable, a variable which is near time-invariant, does not change the results. Also, excluding CEO gender from the regression does not have an impact on the results. In addition, the results are robust against the Company performance measure ROA. If I exclude this measure, or replace it by EBIT (IHS) the conclusions from the regression analysis do not change. In this specification of the model I am not able to check whether including country dummies would change the results. Country dummies in combination with fixed effects would produce multicollinearity. This results from the fact that for some countries the dataset only contains a very limited number of observations.
Table IX: Ordinary least squares and firm fixed effects regressions of direct Anglo-Saxon influence on CEO compensation
The table shows results from ordinary least squares regressions including firm fixed effects of total CEO compensation and performance-dependent compensation. The sample consists of an unbalanced panel of CEO level data from 231 European companies for the period 2001-2011. The total number of observations is 740. Total CEO compensation is measured as the sum of salary, bonus, pension and other compensation. A log transformation is applied to total compensation. Performance-dependent compensation is defined as the ratio between variable and fixed pay. Anglo-Saxon board member, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member on the board. Anglo-Saxon board member in compensation committee, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member in the compensation committee. The coefficients of the control variables are also given in the table. Company size is measured by the natural logarithm of total assets. ROA is used as an indicator of Company performance. Board size is measured by the total number of executive and non-executive board members. CEO gender is a dummy variable. CEO tenure is measured by the time the CEO served in the role of CEO. All regressions include year dummies. The robust standard errors of the variables are reported in parentheses. Statistical significance at the 10%, 5% and 1% level is indicated by *, ** and ***, respectively.
Total CEO compensation Performance-dependent compensation
Independent variable (1) (2) (4) (5)
Anglo-Saxon board member -0.130 0.003
(0.104) (0.059)
Anglo-Saxon board member in
compensation committee 0.333* -0.012 (0.177) (0.076) Company size (ln) 0.162*** 0.165*** -0.001 -0.001 (0.060) (0.058) (0.015) (0.015) Company performance 0.050*** 0.047*** 0.005 0.005 (0.018) (0.018) (0.011) (0.011) Board size 0.019 0.017 0.002 0.002 (0.023) (0.023) (0.008) (0.008) CEO age 0.007 0.008 -0.001 -0.001 (0.008) (0.008) (0.002) (0.002) CEO gender -0.490 -0.523* 0.081 0.082 (0.310) (0.311) (0.117) (0.113) CEO tenure -0.031*** -0.032*** 0.005** 0.005** (0.011) (0.011) (0.002) (0.002)
CEO tenure squared 0.001 0.001* 0.000*** 0.000***
(0.000) (0.000) (0.000) (0.000)
Constant 5.762*** 5.660*** 0.232 0.235
(0.690) (0.677) (0.170) (0.164)
Adjusted R-squared 0.802 0.803 0.656 0.656
28
4.2 Robustness check based on location
29 Table XI: Ordinary least squares and firm random effects regressions of direct Anglo-Saxon influence on CEO
compensation with law system interaction variables
The table shows results from ordinary least squares regressions including firm fixed effects of total CEO compensation and performance-dependent compensation. The sample consists of an unbalanced panel of CEO level data from 231 European companies for the period 2001-2011. The total number of observations is 740. Total CEO compensation is measured as the sum of salary, bonus, pension and other compensation. A log transformation is applied to total compensation. Performance-dependent compensation is defined as the ratio between variable and fixed pay. Anglo-Saxon board member, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Anglo-Saxon board member on the board. Anglo-Saxon board member in compensation committee, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member in the compensation committee. The coefficients of the control variables are also given in the table. Company size is measured by the natural logarithm of total assets. ROA is used as an indicator of Company performance. Board size is measured by the total number of executive and non-executive board members. CEO gender is a dummy variable. CEO tenure is measured by the time the CEO served in the role of CEO. Interaction group 1 is an interaction variable between Anglo-Saxon board member or Anglo-Saxon board member in compensation committee and the countries characterized by German Civil law (group 1). Interaction group 2 is an interaction variable between Anglo-Saxon board member or Anglo-Saxon board member in compensation committee and the countries characterized by French Civil law (group 2). The robust standard errors of the variables are reported in parentheses. Statistical significance at the 1% and 10% level is indicated by *** and *, respectively.
Total CEO compensation Performance-dependent compensation
Independent variable (1) (2) (4) (5)
Anglo-Saxon board member -0.079 -0.012
(0.171) (0.087)
Anglo-Saxon board member in
compensation committee 0.329* -0.013 (0.175) (0.076) Company size 0.159*** 0.160*** -0.002 -0.002 (0.058) (0.058) (0.016) (0.015) Company performance 0.051*** 0.050*** 0.005 0.005 (0.017) (0.017) (0.010) (0.011) Board size 0.020 0.018 0.002 0.002 (0.022) (0.022) (0.008) (0.007) CEO age 0.007 0.008 -0.001 -0.001 (0.008) (0.009) (0.003) (0.003) CEO gender -0.469 -0.479 0.087 0.087 (0.300) (0.301) (0.119) (0.120) CEO tenure -0.031*** -0.031*** 0.005* 0.005* (0.011) (0.011) (0.002) (0.003)
CEO tenure squared 0.001 0.001 0.000 0.000
30
5. Discussion and conclusion
This paper is aimed at studying the relationship between CEO compensation and direct and indirect Anglo-Saxon influence, as first proposed by Oxelhiem and Randoy (2005). Direct Anglo-Saxon influence is measured by Anglo-Saxon supervisory board membership and indirect Anglo-Saxon influence is measured by an Anglo-Saxon exchange listing. The analysis suggest two broad conclusions. Firstly, I find that CEO compensation is influenced through direct Anglo-Saxon influence. However, after controlling for firm-fixed effects it seems that this influence is only present when there is an Anglo-Saxon board member in the compensation committee. This is in line with Daily et al. (1998) who supports the view that when studying CEO compensation one should focus on the composition of the compensation committee, since this is where the majority of the decision making regarding CEO compensation occurs. So, even though an Anglo-Saxon board member might bring with him practices which are common in Anglo-Saxon countries, he is probably only able to express these practices when he is a member of the compensation committee.
Secondly, I find evidence that indirect Anglo-Saxon influence has an effect on CEO compensation. Having a cross-listing in an Anglo-Saxon country results in having significant higher total CEO compensation and having more performance dependent compensation. However, in this research I have not been able to include firm-fixed effects in the models with indirect Anglo-Saxon influence. Future research, should test whether after including firm-fixed effects the results are the same.
31 The results indicate that the nationality of supervisory board members might have an impact on CEO compensation. One important question remains: What is the broader impact of the relationship found between direct and indirect Anglo-Saxon influence and CEO compensation? The paper provides evidence that, corporate governance practices which are common in the Anglo-Saxon corporate governance system might migrate to the Continental Europe corporate governance system. Therefore, corporate governance convergence takes place. Coffee (2002) identifies two routes through which this process might take place. On the one hand corporate governance convergence may occur through institutions and legislation (e.g. corporate governance codes), and on the other hand corporate governance convergence may occur through actions from corporations (e.g. cross border exchange listings and hiring board members from other corporate governance systems). It is exactly this latter route on which this research focuses on by examining the effects of Anglo-Saxon board membership and by examining Anglo-Saxon cross-listings. The results are in line with Coffee’s (2002) findings that corporate governance convergence takes place via corporate actions.
The dataset allows me to study the Anglo-Saxon influences for a broad range of Continental European countries. However, the dataset has some drawbacks as well. The sample did not allow me to include a stock market performance measure, which from an agency theory perspective should have been included as a control variable. As a proxy I used the accounting performance measure ROA. Furthermore, I was not able to study the Anglo-Saxon effect on long-term incentive schemes, like option compensation. The weak link found between Anglo-Saxon influence and performance-dependent compensation might be the result of the fact that I rely on total cash compensation as a measure of total CEO compensation. The equity based components of compensation (e.g. options and stock awards) are omitted. Future research should include these components of compensation in the analysis, because these schemes are important mechanisms to align the interest of managers and investors.
33
Appendix
0 0,05 0,1 0,15 0,2 0,25 0,3 0,35 0,4 0 200 400 600 800 1000 1200 1400 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011Figure I: Evolution of total compensation and performance-dependent compensation from 2001 to 2011. Total compensation is the sum of salary, bonus, pension and other compensaiton. Performance-dependent compensation is measured by the ratio of bonus over s
35 Appendix Table II: Ordinary least squares regressions of direct and indirect Anglo-Saxon influence on CEO compensation
The table shows results from ordinary least squares regressions of total CEO compensation and performance-dependent compensation for European companies for the period 2001-2011. Total CEO compensation is measured as the sum of salary, bonus, pension and other compensation. A log transformation is applied to total compensation. Performance-dependent compensation is defined as the ratio between variable and fixed pay. Anglo-Saxon board member, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member on the board. Anglo-Saxon board member in compensation committee, is a dummy variable indicating whether in a given CEO-year there is an Anglo-Saxon board member in the compensation committee. The variable Anglo-Saxon exchange listing shows whether in a given CEO-year, the company on which the CEO serves, is listed on an Anglo-Saxon exchange in 2012. The coefficients of the control variables are also given in the table. Company size is measured by the natural logarithm of total assets. ROA is used as an indicator of Company performance. Board size is measured by the total number of executive and non-executive members on the board. CEO gender is a dummy variable. CEO tenure is measured by the time the CEO served in the role of CEO. All regressions include year dummies and country dummies, but they are not reported. The robust standard errors of the variables are reported in parentheses. Statistical significance at the 10%, 5% and 1% level is indicated by *, ** and ***, respectively.
Total CEO compensation Performance-dependent compensation
Independent variable (1) (2) (3) (4) (5) (6)
Anglo-Saxon board member 0.022* 0.010
(0.068) (0.018)
Anglo-Saxon board member
in compensation committee 0.303*** 0.065***
(0.074) (0.014)
Anglo-Saxon exchange listing 0.124*** 0.051***
(0.039) (0.006) Company size (ln) 0.246*** 0.235*** 0.237*** 0.034*** 0.031*** 0.029*** (0.034) (0.033) (0.032) (0.007) (0.007) (0.007) Company performance -0.048*** -0.047*** -0.043*** -0.009** -0.008* -0.007 (0.010) (0.009) (0.009) (0.005) (0.004) (0.005) Board size 0.026*** 0.025** 0.024** 0.012*** 0.011*** 0.011*** (0.010) (0.011) (0.010) (0.003) (0.003) (0.003) CEO age -0.001 0.000 0.000 0.000 0.000 0.000 (0.003) (0.003) (0.003) (0.001) (0.001) (0.001) CEO gender 0.155 0.106 0.141 0.088** 0.073* 0.077** (0.170) (0.165) (0.162) (0.043) (0.042) (0.043) CEO tenure 0.000 0.002 -0.002 0.006** 0.006** 0.005** (0.005) (0.005) (0.005) (0.002) (0.002) (0.002)
CEO tenure squared 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 5.057*** 5.006*** 5.160*** -0.168* -0.178* -0.124
(0.308) (0.285) (0.314) (0.102) (0.099) (0.103)
Adjusted R-squared 0.522 0.530 0.525 0.330 0.335 0.338
Year dummies Yes Yes Yes Yes Yes Yes
