Measuring Anglo-American corporate governance influences on
CEO compensation for listed Dutch companies
July 2008
Martijn
Verburg
1467638
University of Groningen
Measuring Anglo-American corporate governance influences on
CEO compensation for listed Dutch companies
Abstract:
In this thesis the Anglo-American influences on CEO compensation in the Netherlands are examined. Based on the rationale developed by Oxelheim and Randoy, two measures are used as proxies for the Anglo-American influences: an Anglo-American exchange listing of the firm and the presence of an Anglo-American board member. Using panel data for Dutch firms on CEO remuneration for the period 2002-2006, I test whether there is a significant influence of these two measures on remuneration. In line with previous studies, I find that there is a significant effect of Anglo-American effect of Anglo-American influences on CEO compensation; however, in contrast to the literature I find that CEO compensation is larger for firms with an Anglo-American board member compared to firms without Anglo-American board member, but firms with an Anglo-American exchange listing pay their CEOs significantly less than firms without Anglo-American exchange.
JEL-codes: G15,G30, G34
1.
Introduction
The topic of this study is the effect of the Anglo-American corporate governance system on CEO compensation in Dutch firms. In the last decennium, corporate governance has
developed into a large research field, and within this field CEO compensation has become a hot topic for research. In the Netherlands a few cases of ‘excessive’ (in the eyes of the
majority of the Dutch’) rewarding of CEOs have triggered public attention to this topic. In the Dutch debate about CEO compensation, the emphasis has been on good corporate governance to prevent the alleged excessive rewarding of CEOs. This emphasis was enhanced by the introduction of the ‘Code Tabaksblat’, the Dutch corporate governance code. The legislator designated the Dutch Corporate Governance Code as a code of conduct to which listed companies should refer to in their annual report. Firms should indicate to what extent they have complied with the principles and best practice provisions ("the comply or explain principle"). These provisions create a set of standards governing the conduct of management board and supervisory board members (also in relation to the external auditor) and
shareholders.They reflect the national and international 'best practices' and may be regarded as an elaboration of the general principles of good corporate governance. Companies may depart from these best practice provisions, but have to explain why they do so; this obligation increases the transparency about executive compensation for Dutch listed companies.
A topic that can be related to corporate governance issues, is the increasing
internationalisation of many companies. To this extent, the current credit crisis provides evidence that the international financial markets are more entangled than ever. Therefore, it might be interesting to use an internationalisation-perspective to study the CEO compensation in the Netherlands, since this country depends largely on imports and exports to account for the size of its GDP and hosts many internationalised firms. To my knowledge, there has not been any attempt to link the CEO compensation of Dutch firms to foreign influences they face. It would be interesting to take a look at these influences, and in particular at the influences of Anglo-American countries, because Anglo-American countries have a very different standard of corporate governance than countries in Continental Europe (Aguilera and Jackson, 2003). These foreign influences on CEO compensation are intuitive; the
Anglo-American firm. To illustrate this example of the ‘institutional contagion effect’ (Oxelheim and Randøy, 2006), let us compare table I, which contains the top 10 of best earning CEOs of listed Dutch companies in 2006, with table II, which contains the top 10 of best earning CEOs of listed American companies in 2006.
Table I
Top 10 - total CEO remuneration of listed Dutch companies 2006
The average CEO remuneration for American companies listed at the New York Stock exchange in 2006 was 15.2 million dollar. Even considering the weak dollar of these days, they clearly earned more than their colleagues of Dutch companies. This difference in pay has also to do with the size of the firms, American companies are also larger than Dutch firms, and other firm characteristics but nonetheless it illustrates that there are huge opportunity costs for CEOs working in the Netherlands. However, strangely enough it is common practice to compare CEO numeration on a domestic level, while the actual firms operate on an
international level.
Table II
Top 10 - total CEO remuneration of listed American companies 2006
Company CEO Remuneration
1 Apple Jobs € 646.600.000
2 Occidental Petroleum Irani € 321.640.000 3 Interactive Corporation Diller € 295.140.000 4 Fidelity National Foley € 179.560.000
5 Yahoo Semel € 174.200.000
6 Dell Dell € 153.230.000
7 Countrywide Financial Mozilo € 141.980.000 8 Abercrombie & Fitch Jeffries € 114.640.000 9 Bank of America Lewis € 99.800.000
10 First Data Duques € 98.210.000
Source : Forbes
Company CEO Remuneration
1 Numico Bennink € 11.644.557
2 Royal Dutch Shell Van der Veer € 9.114.399
3 Reed Elsevier Davis € 7.439.911
4 Aegon Shepard € 7.312.230
5 ING Groep Tilmant € 5.752.886
6 Wolters Kluwer McKinstry € 5.600.000
7 Unilever Cescau € 4.934.435
8 Fortis Votron € 4.665.009
9 ABN AMRO Groenink € 4.614.397
10 Ahold Moberg € 3.733.058
Foreign influences on Dutch CEO compensation are important because of the ‘huge variation in corporate governance among economies’ (Aguilera and Jackson, 2003). Researchers contrast the two models of Anglo-American and Continental European corporate governance (La Porta et al, 1998). The Anglo-American corporate governance system is stylized in terms of financing through equity, dispersed ownership, active markets for corporate control, and flexible labour markets while the Continental Europe corporate governance system is stylized in terms such as long-term debt finance, ownership by large blockholders, weak markets for corporate control and rigid labour markets (Aguilera and Jackson, 2003).
The internationalisation of the financial structure and governance of a firm is of particular significance to companies outside what are commonly considered the most prestigious capital markets (Oxelheim and Randøy (2006)). This means that, in general, CEOs of non-Anglo-American firms subject to Anglo-non-Anglo-American financial influences have to handle more complex tasks than their colleagues who are not subject to these influences. Oxelheim and Randøy (2006) suggest four channels through which Anglo-American influence can increase CEO pay: (1) institutional contagion or spillover effects (2) demand conditions (3) supply conditions (4) a risk premium. Institutional contagion or spillover effects refer to the cost implications with respect to the catch-up of CEO compensation of continental European CEOs compared to their Anglo-American colleagues. There is still a significant difference between the earnings of those two groups (compare table I and table II). Nowadays, when more and more companies are forced by law to disclose CEO compensation, the awareness of the non-Anglo-American CEOs that they are actually underpaid when compared to their Anglo-American colleagues has improved dramatically. The supply and demand conditions (2 and 3) deal with the fact that CEOs that face Anglo-American influences have to be able to handle new, complex tasks resulting from this exposure. As a result, there are less candidates capable of doing the job (supply effects) and firms are willing to pay the CEO more because of their more difficult job (demand effects). Finally, as the Anglo-American corporate governance system is less tolerant of poor performance (Lucier et al., 2004), CEOs that are aware of this greater risk will demand a premium in compensation.
compensation. Previous literature provides evidence that the following factors influence CEO compensation: firm size (e.g. Baumol, 1959), firm performance (accounting and stock price performance)(e.g. Jensen and Murphy, 1990), board size (e.g. Core et al, 1999), board independence and blockholder ownership (e.g. Core et al, 1999). Although the main topic of the research is the Anglo-American effects, the influence of these control variables will also be tested for significance. Second, my research will also focus on the Anglo-American effects on specific compensation elements: stock and stock options. Can Anglo-American influences explain the usage of stock and stock options as reward for CEOs in Dutch firms?
Foreign influences have been excluded from the list of determinants of executive
compensation for a long time, maybe due to the limited availability of cross-border data about executive compensation. However, I think the recent wave of research about differences in corporate governance systems (e.g. La Porta et al, 1998 and Aguilera and Jackson, 2003) should be incorporated in the research about CEO compensation because CEO compensation is also an element of corporate governance. Therefore, linking the Anglo-American corporate governance system to CEO compensation of other countries provides also a test for the actual differences between corporate governance systems. As a result this thesis can serve as an indication whether different corporate governance systems are a cause for differences in CEO compensation.
This thesis is structured as follows. The next section reviews the relevant previous literature. The third section deals with the data I used in my research and the fourth section is about the methodology of our research. The empirical results are reported in section five and finally section six summarizes the main findings of this thesis.
2. Literature review
depends among other things on statutory and common law and human ingenuity in devising contracts”. Bebchuk and Fried (2003) conclude that among financial economists, the
dominant approach to the study of executive compensation views managers’ pay
arrangements as a remedy to the agency problem. Under this approach, which is called the ‘optimal contracting approach’, boards are assumed to design compensation schemes to provide managers with efficient incentives to maximize shareholder value. However,
Bebchuk and Fried argue that executive compensation also have elements that actually result in agency costs. They explain these elements with a new theory called ‘managerial power approach’, which is not a substitute for the optimal contract approach but rather
complementary. Compensation arrangements are likely to be shaped by market forces, which push toward value-maximizing outcomes, and by managerial influence, which pushes toward departures from optimal outcomes in directions favourable to managers. The managerial power approach simply claims that these departures are substantial and that compensation practices thus cannot be adequately explained by optimal contracting alone (Bebchuk and Fried, 2003)
Starting in the 1950s, for a long time only two predictors of executive compensation were included in studies about CEO compensation: company size and profitability. In the studies performed before the 1980s, company size appeared to be far more closely related to the size of the firm than to its profitability (see e.g. Baumol (1959)). The fact that company size is more related to executive compensation than the firm profitability is an indication of the agency problem because the income of the owners of the firm is tied to firm performance (Jensen and Meckling, 1976). As a result of the connection between firm size and executive compensation, executives became motivated to maximize firm size instead of firm
profitability.
the CEO would result in more CEO compensation, and large companies have more management levels in the hierarchy than smaller companies.
The paper by Agrawal (1980) was one of the first to solely focus on the determinants of cross-section differences in executive compensation. The topic was analyzed from the perspective of the human capital theory. Agrawal viewed executive compensation as a function of three basic factors: job complexity, employer’s ability to pay and executive human capital. Interestingly, Agrawal and previous studies defined executive compensation as only direct cash payments, meaning salary and bonus, since “the process of computing the present income equivalent of stocks and options is bound to be very complex and prone to errors” (Agrawal, 1980). Agrawal (1980) finds a significant influence of job complexity on executive compensation and therefore suggests that any job requirements that limit the supply of CEO candidates would increase CEO pay. Jensen and Murphy (1990) find a small but significant relationship between firm performance and CEO compensation. They also conclude that the pay-performance relationship as well as the level of CEO pay is declining since the 1930s. Jensen and Murphy hypothesize that public and private political forces impose constraints that reduce the pay performance sensitivity. Although their paper is focused on stock price
performance measures, Jensen and Murphy reckon that changes in accounting income are also an additional important determinant of pay changes.
Jensen (1993) argues that board size affects corporate governance; independent of other board attributes. His argument focuses on the productivity losses that arise when work groups grow large; he argues that as boards grow larger they become less effective because the
coordination, communication and process problems overwhelm the advantages of having more people to draw on. Core et al (1999) find that measures of board and ownership
structure explain a significant amount of cross-sectional variation in CEO compensation, after controlling for standard economic determinants of pay(the firm’s demand for a high-quality CEO, prior firm performance, and risk). Core et al. explain this result by arguing that firms with weaker corporate governance structures have greater agency problems and that CEOs at firms with greater agency problems receive greater compensation. More specific, the paper by Core et al (1999) provides evidence that board size is positively related to CEO compensation and furthermore they conclude that concentrated ownership of the firm, in the form of a blockholder, reduces CEO remuneration.
countries, especially the United States and the United Kingdom. Furthermore they characterise two models of corporate governance: the Anglo-American model and the Continental European model. The Anglo-American model is stylized in terms of financing through equity, dispersed ownership, active markets for corporate control and flexible labour markets. The Continental Europe model is characterized by long-term debt finance, ownership by large blockholder, weak markets for corporate control, and rigid labor markets. The results of the study by La Porta et al. provide evidence that the corporate governance system in the Netherlands and in Anglo-American countries is quite different.
Ozkan (2006) empirically examines the influence of corporate governance mechanisms on the level of CEO compensation for a sample of 414 listed UK companies for the fiscal year 2003/2004. His results show that corporate governance mechanisms (in the form of board and ownership structure) explain a significant amount of cross-sectional variance in the total CEO compensation, the sum of cash and equity-based compensation, after controlling for other firm characteristics. The study by Core et al. (1999), mentioned earlier, also analyzes the relationship between corporate governance and CEO compensation and Core et al. also finds that corporate governance mechanisms influence CEO compensation.
The studies mentioned above will tempt to let one conclude that CEO compensation should be different in the Netherlands compared to CEO compensation Anglo-American countries, since Anglo-American firms differ from Dutch firms in the dimensions that influence CEO
compensation. Anglo-American firms are usually bigger than Dutch firms, for example. However, despite regional differences and country-based pay, for multinationals a trend of global wide converging compensation for all workers (including executives) is observed (the Economist, 1999)Internationalization is creating a single, global market for top-talent. Executives in any of the three time zones are increasingly rewarded the same way. As New York drags up London, London drags up Europe (the Economist, 1999). Conyon and Schwalbach (1997) find that country effects on pay are partially eradicated by the effects of internationalization of capital and labour markets. Sanders and Carpenter (1996) find that the complexity resulting of a firm’s degree of internationalization is managed through higher, longer-term CEO pay, larger top management teams and the separation of chairperson and CEO positions.
different corporate governance systems and differences in factors such as firm size and firm performance. However, this difference in CEO compensation should decrease because of internationalization effects. This is the rationale developed in the paper by Oxelheim and Randøy (2004). In their paper, they suggest channels through which Anglo-American influence affect the CEO compensation in non-Anglo-American firms: (1) institutional
contagion or spillover effects (2) demand conditions (3) supply conditions (4) a risk premium. Furthermore, they point out two specific firms activities that signal compliance with an
Anglo-American standard of corporate governance: (1) cross-listing in an Anglo-American stock market and (2) the recruitment of at least one independent board member representing the American system. In addition, they provide empirical evidence for this Anglo-American influence in Norwegian and Swedish firms. They argue that further research in other countries is needed to validate the generalisation of their results. This thesis will also deal with a question that follows indirectly from the rationale by Oxelheim and Randøy: Can Anglo-American influences explain the usage of stock and stock options as reward for CEOs in Dutch firms? In Anglo-American countries, executive pay in stocks and options is more common than in Continental Europe (for example, Jensen and Murphy (2006) talk about the “US-led option explosion”). Therefore, it can be argued that Anglo-American influences, in the form of a Anglo-American exchange listing or Anglo-American board member, will result in compensation schemes that are more closely tied to performance, in other words: the compensation schemes will include more stock- and option-based pay.
Research objectives and hypotheses
The aim of my research is to examine the impact of Anglo-American financial markets on CEO compensation in Dutch firms. Therefore, my research question is:
Does the Anglo-American corporate governance system influence the CEO compensation in listed Dutch firms?
These research objectives will be tested by the following hypotheses. The first two hypotheses deal with Anglo-American influences on CEO remuneration of listed Dutch firms in absolute terms. The third and fourth hypothesis deal with the Anglo-American influence on the
remuneration components stock and stock options.
Hypothesis 1:The CEO remuneration will be different for Dutch firms with an
Anglo-American board member compared to Dutch firms that do not have an Anglo-Anglo-American board member
Hypothesis 2: The CEO remuneration will be different for Dutch firms that have an Anglo-American exchange listing compared to Dutch firms that do not have an Anglo-Anglo-American exchange listing.
Hypothesis 3: The relative size of the components stocks and options in CEO remuneration will be different for Dutch firms with an Anglo-American board member compared to Dutch firms that do not have an Anglo-American board member.
Hypothesis 4: The relative size of the components stocks and options in CEO remuneration will be different for Dutch firms with an Anglo-American exchange listing compared to Dutch firms that do not have an Anglo-American exchange listing.
For all four hypotheses, the null hypothesis is that there is a no difference between CEO compensation for Dutch firms with Anglo-American influence compared to Dutch firms that do not have an Anglo-American influence.
3. Methodology
The model that I use in this thesis is according to the specification of Oxelheim and Randøy (2006) because they introduced the measures for Anglo-American influences in their paper and therefore their model is well-suited to test my hypotheses.
1 1 10 1 9 1 8 1 7 1 6 1 5 1 4 1 3 1 2 1 1 − − − − − − − − − − − + + + + + + + + + + + = t, i t, i t , i t, i t, i t, i t, i t, i t, i t, i t, i t, i Age * r Blockholde * size Board * ) Employees ( Ln * ) Sales ( Ln * Q Tobins * ROE * turn Re Stock Annual * Member Board American Anglo * Listing Exchange American Anglo * ) muneration (Re Ln
ε
β
β
β
β
β
β
β
β
β
β
α
(1)It is well-known that ‘size’ variables are often heavily skewed. To justify the use of OLS, the residuals must be normally distributed, however. To increase the likelihood that the error term in the model is normally distributed, for all ‘size’ variables the natural logarithm is taken. To test Hypothesis 1, I check whether the coefficient β1 in equation (1) is significantly different
from zero in the estimation results. To test Hypothesis 2, I check whether the coefficient β2 of
equation (1) is significantly different from zero in the estimation results.
For the hypotheses 3 and 4, to test whether the null hypothesis is true, an Ordinary Least Squared regression is performed on the following equation (2):
(2)
Again, the error term in the model is normally distributed, because for all ‘size’ variables the natural logarithm is incorporated in the model. To test Hypothesis 3, I check whether the coefficient β1 in equation (2) is significantly different from zero in the estimation results. To
test Hypothesis 4, I check whether the coefficient β2 of equation (2) significantly different
from zero in the estimation results.
In the model, the dependent variable is lagged 1 year with respect to the independent variables. However, this issue is subject to debate. One could argue that the real CEO
remuneration is determined at the end of the year and therefore that the independent variables should not lag behind. However, one could also argue that the remuneration is set at the beginning of the year and afterwards it’s only a matter of judging whether the CEO actually did attain the milestones that were set at the beginning of the year. Previous literature can also not provide a definitive answer to this issue. However, since this study is based on the
rationale by Oxelheim and Randøy, and they chose to lag the dependent variable, I will use the model above.
In addition to the two variables that serve as proxies for the Anglo-American spill-over effect, a large amount of control variables are included in the model to minimise specification bias in the hypothesis testing. From previous literature, various control variables are known to have impact on CEO remuneration. By including these variables, I will be able to extract the pure Anglo-American effect on CEO remuneration from the model.
The variables that are used in the model are straightforward but all need some additional remarks (the expected sign of the coefficient is in brackets). All variables concerning remuneration are explained in the section ‘data’. The data about remuneration are incorporated in the model by taking the natural logarithm. The variable Anglo-American exchange listing (+) is a dummy variable that is assigned a one if the firm is listed on a stock exchange in an Anglo-American country. The variable Anglo-American board member (+) is also a dummy and will receive a value of one when one or more directors of the board are from Anglo-American origin. Although I reckon that the influence of a single board member on the remuneration can be limited, I chose to use this definition because the signalling effect of Anglo-American corporate governance influence would be achieved by the inclusion of even one Anglo-American board member (Oxelheim and Randøy, 2006).
governance. In my model, good corporate governance is proxied by the presence of a
Blockholder (-) or a small board of directors (Jensen, 1993). The variable Age (+) is a proxy for the surviving power of the firm. Older firms pay their CEO more than new firms, and firms that internationalize are usually firms that have proven to be ‘survivors’ domestically. Therefore the model controls for the age of the firm.
4. Data
The database consists of panel data with yearly data from the period 2002-2006 from 120 companies. The decision to use panel data is based on the limited availability of data about remuneration. In the Netherlands, firms are obliged to disclose CEO remuneration from 2002. This fact eliminates the possibility of using time series data. Also, the number of listed Dutch firms is quite limited, so there is not enough cross-section data for a sample that is large enough to be able to come to statistical conclusions. Therefore, panel data is a natural choice for this research topic.
All data concerning remuneration (total remuneration as well as the individual components) were obtained from a database put together and published by the Dutch Investors Association (in Dutch: Vereniging van Effectenbezitters). The data in this database are collected from annual reports by the Dutch Investors Association. Data is available for the current period starting at 2002, since from this year Dutch companies are forced by legislation to disclose information about CEO remuneration in their annual reports. This rather unique database about CEO compensation is very suitable for research, because it provides standardized values of the different parts of CEO compensation. For example, for all companies the value of options is determined by the Black and Scholes option-model.
Table III
Average composition of CEO compensation
Total remuneration is divided into six categories: salary, bonus, pension, stocks, options and other remuneration. Salary deals with the fixed part of the remuneration and the category bonus deals with the variable remuneration. Pensions costs are calculated as the company’s contribution to the pension plan of the CEO in the concerning year. The value of stocks and options is calculated at the date of issuance. Both absolute and relative size of the remuneration components are reported in the table.
(1000 of euros) Salary 427 36% 392 44% 400 44% 429 37% 443 34% 453 31% Bonus 271 23% 196 22% 193 21% 245 21% 306 23% 369 25% Pensions 124 11% 96 11% 120 13% 127 11% 144 11% 126 9% Other 29 2% 29 3% 13 1% 36 3% 33 3% 32 2% Stock 185 16% 37 4% 56 6% 186 16% 260 20% 313 22% Options 138 12% 141 16% 118 13% 142 12% 125 10% 163 11% Total 1.174 100% 890 100% 900 100% 1.164 100% 1.311 100% 1.456 100%
Source: Dutch Investors Association
2002
2002-2006 2003 2004 2005 2006
The independent variables are collected from three sources: Amadeus, Datastream and annual reports. Table XVII in Appendix 1 shows the source of data for each variable individually. For each variable, only one source of data was used to remain consistent for each variable, and annual reports were used as back-up (except when noted as main source of data). The total sample consists of 496 company-year observations for 120 companies between 2002-2006, this means that it is not a balanced panel because not for every firm, all data of the period 2002-2006 were available. Table IV presents the systematic manner by which this sample was selected. Obtaining the ROE appeared to be especially problematic for companies in the banking sector.
Table IV Data selection
Filter Observations
Data from database VEB 530
Data from foreign companies listed in the Netherlands 12 Annual stock return unavailable 12
ROE unavailable 8
-32
Data description
Table V reports the descriptive statistics of total remuneration, the dependent variable. The average CEO compensation in the period 2002-2006 was 1,17 million euros. However, note that the median compensation is only 606.000 euros, which suggests that CEO compensation is highly skewed and because of that, the mean compensation is determined by the largest rewards. In addition, also note that the standard deviation is very large. These two factors are important warnings not to rely too much on the average compensation for a reliable picture of CEO remuneration in Holland. However, both the median and the mean show a trend of increasing compensation during the period 2002-2006, in nominal terms (not corrected for inflation). Furthermore, the large numbers for kurtosis and skewness point to a non-normal distribution of the CEO compensation. One remarkable feature of the table is the negative compensation as minimum in the year 2005. This observation deals with the remuneration of Gerard Kleisterlee from Phillips. In 2005, he retired and returned some of his pension cost to the company. Finally, data is distributed relatively similarly across all years, except for 2002. In 2002, less firms were put in the database of the Dutch Investors Association than other years, possibly because it was the first year in which Dutch firms were forced to disclose their executive remuneration.
Table V
CEO compensation of listed Dutch firms
The descriptive statistics for total CEO compensation are reported for the whole sample period and for each year.
(1000 of euros) 2002-2006 2002 2003 2004 2005 2006 Mean 1.174 890 900 1.164 1.311 1.455 Median 606 479 509 616 700 767 Stand. Dev. 1.562 1.006 1.151 1.603 1.726 1.869 Min -104 87 82 63 -104 68 Max 11.644 4.425 8.823 11.511 10.826 11.644 Kurtosis 14 4 24 18 10 10 Skewness 3 2 4 4 3 3 N 498 76 94 101 107 120
Table VI compares the average CEO compensation of different industries. The CEOs that earn the most are located in the industries post and telecommunication, financial
activities, paper products and computer activities would like to link their pay to the pay level of their colleagues in other industries, their remuneration is the lowest.
Table VII reports the same descriptive statistics for the firm characteristics excluding the dummy variables. The average company in the sample has an annual stock return of 11%, a market capitalisation of about three times the book value of the firm, 9,7 billion of sales, a return on equity of 5%, a board of directors consisting of 5 persons, 17682 employees and is 73 years old. The negative Tobin’s Q-value belongs to the company De Vries en Robbe Groep, that was declared bankrupt in that year.
Table VIII reports the descriptive variables for the dummy variables and it can be observed that 10% of the observations contained a firm that was listed on an Anglo-American stock exchange. Furthermore 28% of the observations belonged to companies with an Anglo-American director. Since the values for the firms are stable over years, it is also possible to report the descriptive statistics for the firms. Panel B of table VIII shows that 20 of the 120 firms (17%) in the sample are listed on an American stock exchange and that 35 firms (29%) have an Anglo-American board member. Furthermore, the high percentage (54%) firms with a blockholder is rather remarkable .
Table IX shows the correlation between the variables. There is one pair of variables that is statistically correlated, when 0,6 is used as cut-off point: Board size and Total Remuneration are positively correlated. However, Board size is not a proxy for firm size or firm
performance, since Board size is not significant correlated to other variables. In some cases, the results of the correlation table seem counterintuitive. For example, Total Remuneration is expected to be positively correlated with Annual Stock Return, since part of the remuneration is in shares. Also, in this sample total remuneration, and all parts of the remuneration
including shares, are insignificantly correlated. Furthermore, the presence of a large
Table VI
Average CEO compensation per industry
The definition of each industry is based on the NACE classification. The average compensation per year of each year is reported.
(1000 of euros) 2002 2003 2004 2005 2006
Oil and gas extraction 1.709 1.780 5.294 4.329 5.816
Food and beverages 1.785 1.284 1.546 3.897 3.399
Textiles 589 662 731 924 1.053
Paper products 242 338 453 272 259
Petreoleum products 1.042 1.119 2.696 1.733 2.223
Chemicals 854 684 1.145 1.233 1.156
Rubber and plastic products N/A 357 319 315 518
Non-metallic mineral products 122 124 160 197 292
Fabricated metal products 447 498 502 617 1.949
Machinery and equipment 565 511 854 1.061 1.868
Office machinery and computers 1.047 1.252 928 1.405 1.644
Electrical machinery 934 1.170 1.134 1.126 1.143
Medical and optical instruments 416 426 393 -104 337
Motor vehicles N/A N/A 336 351 222
Other transport equipment 385 654 555 583 614
Furniture 1.627 N/A 63 66 323
Construction 533 685 777 874 1.218
Sale of motor vehicles 295 446 454 467 453
Wholesale trade 643 661 676 784 864
Retail trade 1.964 2.600 1.684 1.653 1.597
Transport 206 406 428 548 689
Post and telecommunication 2.161 1.804 1.850 2.213 2.053
Financial intermediation 4.176 2.654 3.084 2.286 2.608
Insurance and pension funding 2.687 2.320 3.235 4.877 4.340
Activities auxiliary to financial intermediation 705 525 560 723 607
Real estate activities 238 252 252 292 219
Computer activities 297 419 361 395 553
R&D N/A N/A N/A N/A 346
Other business activities 696 550 914 1.109 1.389
Table VII
Relevant firm characteristics (t-1)
Annual stock return is calculated as the difference between the stock price on 31 December of the year and the opening stock price on 1 January of the year, divided by the stock price on 1 January. The variable Tobin’s Q is calculated by dividing the market capitalisation per share by the book value per share. The return on
shareholder’s funds is used as a proxy for the ROE. The dummy blockholder is rewarded a one when a shareholder of the company owns more than 20% of the shares, according to the definition by La Porta (1999). The variable age is calculated as the difference between the founding year of the company and the current year.
Annual stock Tobin's Q Sales ROE Board size Employees Age
return (in mln) Mean 0,11 2,92 9.805 0,05 5 17.682 73 Median 0,05 1,79 594.382 0,16 4 3.204 45 Stand. Dev. 0,52 8,49 40.366 0,72 2 43.826 74 Min -0,99 -10,25 0 -9,04 1 2 1 Max 4,97 154,17 345.100 4,40 15 300.006 390 Kurtosis 18,54 231,30 43 94,24 4 17 4 Skewness 2,63 14,26 6 -7,59 2 4 2 Table VIII
Descriptive statistics of dummy's (t-1)
The descriptive statistics of the dummies in the model are reported in the table. For each value of the dummy, the absolute and relative size of the sample that contains that value is reported. In Panel A, the statistics of the total sample are presented (including different observations for one company). In Panel B, the statistics for the firms are presented, this means all companies are counted only once.
Panel A: Observations
Anglo-American exchange listing 449 90% 49 10% 498 100%
Anglo-American board membership 359 72% 139 28% 498 100%
Blockholder ownership 277 56% 498 44% 498 100%
Panel B: Firms
Anglo-American exchange listing 100 83% 20 17% 120 100%
Anglo-American board membership 85 71% 35 29% 120 100%
Blockholder ownership 55 46% 65 54% 120 100% N YES NO N YES NO
CEO significant more than the Dutch companies without Anglo-American board member, in total and in separate components of remuneration. Panel B shows that Dutch companies with an Anglo-American exchange listing on average reward their CEOs more than the Dutch companies without an Anglo-American exchange listing on all components, but only the difference in mean for total compensation, salary, bonus and pension is statistically
significant. So without controlling for other variables, table IX provides evidence for the first two null hypotheses to be rejected. Of course, CEO compensation is influenced by other factors and one must control for those other factors to observe the conditional Anglo-American influence, since these differences could be due to sub-sample bias. Therefore an OLS regression is also performed (see section results).
Table X
Comparing the average remuneration with and without Anglo-American influence A statistic mean t-test is performed to observe whether two sub-samples, based on the two Anglo-American dummies. In the first two columns, the outcomes for the t-test for the dummy Anglo-American board member are presented, and in the last two columns this is done for the dummy Anglo-American exchange listing.
1000 of euro's Listed Not listed Board No board member member Average total remuneration 668 2483*** 1094 1908**
Average salary 326 687*** 406 614***
Average bonus 144 597*** 238 569***
Average pension 90 212*** 115 205**
Average value of shares 43 552*** 173 301
Average value of options 50 366*** 134 178
Average other remuneration 13 69*** 28 41
Anglo-American exchange listing
Anglo-American board member
Table IX Correlation matrix
This table reports the correlations between all variables, independent and dependent, without the dummies. The first six variables are all variables about remuneration, these are the dependent variables. The other variables are all control variables, which are included to minimize specification bias in the hypothesis testing. The cut-off point for significant correlation is 0,60.
Total Salary Bonus Pension Stock Options Other Annual stock Tobin's Q Sales ROE Board size Blockholder Employees Age
remuneration remuneration return
Total remuneration 1,00 Salary 0,59 1,00 Bonus 0,54 0,72 1,00 Pension 0,19 0,38 0,27 1,00 Shares 0,43 0,56 0,48 0,17 1,00 Options 0,40 0,46 0,37 0,08 0,54 1,00 Other 0,16 0,23 0,36 0,42 0,12 0,08 1,00
Ann. stock return -0,01 -0,10 -0,06 -0,02 -0,06 -0,09 0,02 1,00
Tobin Q 0,00 -0,03 0,02 -0,03 0,02 0,01 0,00 0,04 1,00 Sales 0,24 0,21 0,18 0,04 0,19 0,06 0,03 -0,08 -0,02 1,00 ROE 0,07 0,07 0,06 0,05 0,02 0,03 0,04 0,08 0,02 0,03 1,00 Board size 0,63 0,56 0,44 0,21 0,23 0,21 0,18 -0,05 -0,06 0,15 0,14 1,00 Blockholder -0,26 -0,25 -0,15 -0,11 -0,18 -0,16 -0,09 0,01 0,07 -0,05 0,00 -0,21 1,00 Employees 0,58 0,55 0,46 0,17 0,26 0,27 0,27 -0,10 -0,01 0,23 0,07 0,56 -0,15 1,00 Age 0,09 0,09 0,06 0,06 0,08 0,03 0,02 0,01 -0,05 0,31 0,08 0,07 0,01 0,09 1,00
5. Results
OLS regressions were performed, first, for the total remuneration, second, for the components separately, and third, for the ratio ‘performance pay’. The outcomes are reported the tables XI-XVI. Performing OLS regression is justified by the fact that the residuals of the regression are normally distributed, according to a post-estimation diagnostic test (Jarque Bera).
Total CEO remuneration
CEO remuneration appears to be influenced by a number of factors. First, the two measures of American influence are both significant. The coefficient of the variable
Anglo-American board member is positive, but the coefficient for the variable Anglo-Anglo-American exchange listing is negative. Since both variables are measures for Anglo-American influence, this difference in sign of those variables is rather strange. When one of the two variables is omitted from the model, the results remain the same (table XI). So, both the relative size and the significance of these effects seem fairly robust. The coefficients of the control variables are also reported in table XI. The Annual stock return in the year before has a significant positive effect on total CEO compensation. Remarkably, the other performance measure, Return on equity, seems to be unrelated to CEO compensation – at least the relation is not very apparent - since the effect is insignificant. Both measures of size, Sales and Employees, have a strong significant positive effect on CEO compensation. The significant positive coefficient of the variable Board size is in line with previous literature, a larger board of directors usually decreases the productivity of the board. The other variable about
corporate governance, Blockholder, is significant negatively related to CEO compensation, as expected. Overall, the correlation coefficient shows that the model is able to explain about 76% of the variation in CEO compensation of Dutch firms.
Components of remuneration
The different components of remuneration are regressed against the same variables, for two reasons. First, this provides additional insight into the results on total remuneration. Second, these regressions can be useful to test the robustness of the previous outcomes. The
samples for the components shares, options and other remuneration, but these samples remain large enough to perform statistical analysis.
Overall, the component salary is best explained by the model and the component ‘other
remuneration’ the least (table XII). On average, the component salary accounts for 36% of the
Table XI
Anglo-American influences on total CEO remuneration (lagged control variables) The estimated model is: Ln (compensation) = a + b1*Anglo-American board memberi,t-1 + b2*Anglo-American exchange listingi,t-1 + b3*Annual Stock Returni,t-1 + b3*Return on Equityi,t-1 + b4*Tobin_Qi,t-1 + b5*Ln (Sales)i,t-1 + b6*Ln (Employees)i,t-1 + b7*Board sizei,t-1 + b8*Blockholderi,t-1 + b9*Agei,t-1
The dependent variables, total remuneration, is not directly put into the Panel-OLS regression, but by taking the natural logarithm of the variables. The independent variables are lagging behind one year.
Variable Coefficient Coefficient Coefficient (t-statistic) (t-statistic) (t-statistic) Anglo-American 0,3489*** 0,3173*** -board member (5,73) (5,27) -Anglo-American -0,2371*** - -0,1582** exchange listing (-2,95) - (-1,93) Control variables Annual stock 0,1712*** 0,1748*** 0,1826*** return (3,75) (3,80) (3,87) Return on equity -0,0134 -0,0147 -0,0201 (-0,41) (-0,44) (-0,59) Tobin_Q 0,0007 0,0010 0,0017 (0,25) (0,39) (0,61) Log Sales 0,1435*** 0,1326*** 0,1359*** (9,21) (8,70) (8,48) Log Employees 0,065*** 0,0748*** 0,0819*** (3,36) (3,90) (4,16) Board size 0,1492*** 0,1465*** 0,1745*** (10,91) (10,68) (13,07) Blockholder -0,2041*** -0,2032*** -0,2455*** (-4,22) (-4,18) (-4,97) Age -0,0005 -0,0003 -0,0004 (-1,56) (-1,05) (-1,36) R-squared 0,76 0,76 0,74 N 491 492 492
Table XII
Anglo-American influences on CEO remuneration components (lagged control variables)
The estimated model is: Ln (compensation) = a + b1*Anglo-American board memberi,t-1 + b2*Anglo-American exchange listingi,t-1 + b3*Annual Stock Returni,t-1 + b3*Return on Equityi,t-1 + b4*Tobin_Qi,t-1 + b5*Ln (Sales)i,t-1 + b6*Ln (Employees)i,t-1 + b7*Board sizei,t-1 + b8*Blockholderi,t-1 + b9*Agei,t-1
The dependent variables, the remuneration components, are not directly put into the Panel-OLS regression, but by taking the natural logarithm of the variables. The regressions concerning the different parts of the
remuneration are performed without the observations with zeros on that specific remuneration component. The independent variables are lagging behind one year.
Salary Bonus Pension Shares Options Other Variable Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) Anglo-American 0,1289*** 0,0954** 0,0302 0,6195* 0,4570** -0,0280 board member (3,43) (2,02) (0,25) (1,96) (2,08) (-0,15) Anglo-American -0,0354 -0,0867 0,0608 -0,6905* -0,6441** -0,0660 exchange listing (-0,72) (-1,30) (0,39) (-1,82) (-2,35) (-0,37) Control variables Annual stock 0,0528* 0,1196*** 0,0143 0,5278* 0,4053** -0,0595 return (1,86) (3,37) (0,16) (1,78) (2,08) (-0,59) Return on equity -0,0204 -0,069 0,1989* -0,1561 -0,1451 0,0282 (-1,01) (-1,58) (1,86) (-0,41) (-0,58) (0,45) Tobin_Q 0,0006 0,0226*** -0,0051 0,0498 0,1693*** 0,0345 (-0,36) (3,09) (-0,92) (1,52) (3,53) (1,19) Log Sales 0,0989*** 0,0946*** 0,0578* 0,1901** 0,2107*** -0,0434 (10,27) (7,03) (1,79) (2,05) (2,82) (-1,26) Log Employees 0,0678*** -0,0057 0,1874*** -0,0065 -0,0561 0,0721* (5,68) (-0,36) (4,97) (-0,04) (-0,58) (1,74) Board size 0,0792*** 0,0878*** 0,0686** 0,0655 0,1031** 0,0663** (9,42) (8,62) (2,52) (1,23) (2,37) (2,27) Blockholder -0,0718** (-0,0870)** -0,1986** -0,3352 -0,3953** -0,0019 (-2,41) (-2,33) (-2,11) (-1,24) (-2,09) (-0,01) Age -0,0005** -0,0002 0,0005 0,0006 0,0033** 0,0001 (-2,40) (-0,76) (0,77) (-0,34) (2,32) (0,11) R-squared 0,78 0,57 0,41 0,33 0,42 0,21 N 488 401 387 100 196 123
Table XIII
Anglo-American influences on Performance pay (lagged control variables) The estimated model is: Ln (compensation) = a + b1*Anglo-American board memberi,t-1 + b2*Anglo-American exchange listingi,t-1 + b3*Annual Stock Returni,t-1 + b3*Return on Equityi,t-1 + b4*Tobin_Qi,t-1 + b5*Ln (Sales)i,t-1 + b6*Ln (Employees)i,t-1 + b7*Board sizei,t-1 + b8*Blockholderi,t-1 + b9*Agei,t-1
The dependent variables, performance pay, is not directly put into the Panel-OLS regression, but by taking the natural logarithm of the variable. The independent variables are lagging behind one year.
Variable Coefficient Coefficient Coefficient (t-statistic) (t-statistic) (t-statistic) Anglo-American 0,1191*** 0,0900*** -board member (4,28) (4,61) -Anglo-American -0,1193*** - -0,1025*** exchange listing (-3,41) - (-3,56) Control variables Annual stock 0,1268*** 0,0940*** 0,0962*** return (5,01) (4,98) (5,00) Return on equity -0,0340 -0,0331 -0,0286 (-1,08) (-0,90) (-1,06) Tobin_Q 0,0105*** 0,0076*** 0,0097*** (2,65) (2,76) (2,66) Log Sales 0,0244** 0,0266** 0,0233*** (2,48) (2,53) (2,63) Log Employees -0,0230 -0,0198 -0,0228 (-1,77) (-1,66) (-1,59) Board size 0,0102 0,0091 0,0099 (1,77) (1,63) (1,68) Blockholder -0,0171 -0,0085 -0,0154 (-0,71) (-0,83) (-0,77) Age 0,0001 0,0002 0,0002 (0,37) (0,46) (0,52) R-squared 0,28 0,27 0,27 N 242 242 242
*p < 0.10 (two-tailed), **p < 0.05 (two-tailed) and ***p < 0.01 (two-tailed) Performance pay
The variable Anglo-American board member has a significant positive effect on the
remuneration components, except for pension and other remuneration. The fact that those two components are not influenced by Anglo-American influences comes as no surprise, those two remuneration components are generally not heavily related to performance, and therefore will be less influenced by a higher standard of corporate governance. Interestingly, the effect of an Anglo-American exchange listing on CEO compensation appears to be only channelled through remuneration in stocks and options. Therefore, in this first robustness check of the results, the effect of an Anglo-American board member on CEO compensation seems to be more robust than the effect of an Anglo-American exchange listing.
The positive effect of the Annual stock return is also an influence for the separate components of CEO compensation, except for pension and other remuneration (again). The influence of Return on equity on the salary, bonus, shares and options is insignificantly negative. Pension and other remuneration are higher when the Return on equity of the year before is higher, and for pension reward this effect is light significant. The variable Tobin’s Q, is positively related to CEO salary, bonus, shares, options and other remuneration, but only the effect on the CEO annual bonus and options is (heavily) significant.
The size effect, measured by the variable Employees and Sales, is also a major influence for the remuneration components separately. The coefficient of the variable Sales is positive and significant for all components, except for other remuneration. The second measure of the sign effect, Employees, has a significant positive effect on CEO salary, pension and other
remuneration.
The two measures of corporate governance have different signs for their coefficients, as observed with the results for total remuneration. The effect of a larger board of directors on the CEO salary, bonus, pension costs, options and other remuneration is significant positive. The effect of a blockholder is positive significant for CEO salary, bonus, pension costs and options.
Finally, the variable Age, a proxy for the surviving power of the company, is significant negatively related to CEO salary and significant positively related to the value of the remunerations in options.
Performance pay
compensation is regressed against all variables. Table XIII shows the results of that regression. One can observe a positive influence of an Anglo-American board member on ‘performance pay’ and a negative influence of an Anglo-American exchange listing. Both influences are highly significant, again. Only Annual stock return, Tobin’s Q and Sales have a significant effect on ‘performance pay’, the other control variables are not significant.
Robustness
To test the robustness of the previous results, all regressions are performed again, but this time the independent variables are of the same year as the dependent variables. The results of this robustness check can be found in Table XIV, XV and XVI. The results for the Anglo-American variables do not change when the independent variables are not lagged.
Remarkably, the variable Annual stock return is not significant anymore for total
remuneration, salary, shares and options. So apparently Annual stock return of the forgoing year has more influence on CEO compensation than the Annual stock return of that actual year, This fact is a strong argument for the theory that CEO compensation is determined at the beginning of the year, and not at the end of the year (section Methodology).
The outcomes for the control variables again provide evidence that CEO compensation is strongly influenced by a size effect and a corporate governance effect in the form of a negative effect of the presence of a Blockholder and a positive influence of the Board size. Overall, the main results of this study seem to be robust.
Discussion
The negative effect of the variable Anglo-American exchange listing is puzzling, that is an Anglo-American exchange listing of a firms reduces CEO compensation. Oxelheim and Randøy (2006) point out the variables Anglo-American exchange listing and Anglo-American board member as signals for compliance with an Anglo-American standard of corporate governance. Therefore, Oxelheim and Randøy expect positive coefficients for both variables, and they have provided empirical evidence for this relationship. The negative coefficient of the variable Anglo-American exchange listing is contradictory to the findings of Oxelheim and Randøy. The different signs of the two Anglo-American variables is of course also
Table XIV
Anglo-American influences on total CEO remuneration
The estimated model is: Ln (compensation) = a + b1*Anglo-American board memberi,t + b2*Anglo-American exchange listingi,t + b3*Annual Stock Returni,t + b3*Return on Equityi,t + b4*Tobin_Qi,t + b5*Ln (Sales)i,t + b6*Ln (Employees)i,t + b7*Board sizei,t + b8*Blockholderi,t + b9*Agei,t
The dependent variable, total remuneration, is not directly put into the OLS regression, but by taking the natural logarithm of the variable.
Variable Coefficient Coefficient Coefficient (t-statistic) (t-statistic) (t-statistic) Anglo-American 0,3432*** 0,3114*** -board member (5,63) (5,14) -Anglo-American -0,2431*** - -0,1646* exchange listing (-3,01) - (-2,01) Control variables Annual stock 0,0738 0,0778 0,0768 return (1,56) (1,63) (1,57) Return on equity -0,0150 -0,0048 -0,0101 (-0,33) (-0,10) (-0,21) Tobin_Q 0,0046 0,0053 0,0067 (1,02) (1,17) (1,45) Log Sales 0,1485*** 0,1371*** 0,1413*** (9,75) (9,22) (9,03) Log Employees 0,0596*** 0,0704*** 0,0766*** (3,18) (3,80) (4,02) Board size 0,1433*** 0,1393*** 0,1668*** (10,57) (10,25) (12,54) Blockholder -0,2157*** -0,2139*** -0,2555*** (-4,45) (-4,38) (-5,17) Age -0,0004 -0,0003 -0,0004 (-1,33) (-0,82) (-1,16) R-squared 0,76 0,75 0,74 N 495 495 495
Table XV
Anglo-American influences on CEO remuneration components
The estimated model is: Ln (compensation) = a + b1*Anglo-American board memberi,t + b2*Anglo-American exchange listingi,t + b3*Annual Stock Returni,t + b3*Return on Equityi,t + b4*Tobin_Qi,t + b5*Ln (Sales)i,t + b6*Ln (Employees)i,t + b7*Board sizei,t + b8*Blockholderi,t + b9*Agei,t
The dependent variables, total remuneration and its components, are not directly put into the OLS regression, but by taking the natural logarithm of the variables. The regressions concerning the different parts of the
remuneration are performed without the observations with zeros on that specific remuneration component Salary Bonus Pension Shares Options Other Variable Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) Anglo-American 0,1248*** 0,2264** 0,0597 0,5533* 0,5143** -0,2473 board member (3,32) (2,12) (0,50) (1,94) (2,26) (-0,58) Anglo-American -0,0413 -0,1900 0,0304 -0,3536 -0,8360*** -0,0933 exchange listing (-0,83) (-1,26) (0,19) (-0,94) (-3,05) (-0,23) Control variables Annual stock 0,0204 0,2503*** 0,1297 0,2013 0,0854 -0,1064 return (0,70) (2,92) (1,21) (0,48) (0,41) (-0,48) Return on equity -0,0795*** 0,1308 0,0864 -0,4159 -0,0252 -0,0515 (-2,84) (0,88) (1,02) (-0,62) (-0,07) (-0,81) Tobin_Q 0,0012 0,0585*** -0,0500 0,2499*** 0,0466 0,0703 (0,45) (3,43) (-1,74) (3,75) (1,61) (0,92) Log Sales 0,1017*** 0,2132*** 0,0805** 0,1735** 0,2708*** -0,0953 (10,84) (7,28) (2,57) (2,13) (3,69) (-1,27) Log Employees 0,0633*** -0,0062 0,1605*** -0,0387 -0,0852 0,2251** (5,48) (-0,18) (4,48) (-0,30) (-0,87) (2,48) Board size 0,0078*** 0,1907*** 0,0746*** 0,0582 0,1055** 0,1548** (9,44) (8,34) (2,79) (1,17) (2,42) (2,27) Blockholder -0,0772*** -0,02127** -0,1994** -0,3737 -0,4019** 0,0333 (-2,60) (-2,50) (-2,13) (-1,49) (-2,08) (0,10) Age -0,0004** -0,0002 0,0003 0,0004 0,0026* 0,0003 (-2,16) (-0,42) (0,50) (0,02) (1,81) (0,12) R-squared 0,78 0,58 0,42 0,42 0,39 0,24 N 494 405 391 100 197 124
Table XVI
Anglo-American influences on Performance pay
The estimated model is: Ln (compensation) = a + b1*Anglo-American board memberi,t + b2*Anglo-American exchange listingi,t + b3*Annual Stock Returni,t + b3*Return on Equityi,t + b4*Tobin_Qi,t + b5*Ln (Sales)i,t + b6*Ln (Employees)i,t + b7*Board sizei,t + b8*Blockholderi,t + b9*Agei,t
The dependent variables, total remuneration, is not directly put into the OLS regression, but by taking the natural logarithm of the variable.
Variable Coefficient Coefficient Coefficient (t-statistic) (t-statistic) (t-statistic) Anglo-American 0,1342*** 0,1298*** -board member (4,56) (4,48) -Anglo-American -0,1309*** - -0,1284*** exchange listing (-3,56) - (-3,48) Control variables Annual stock 0,0231 0,0184 0,0214 return (0,81) (0,79) (0,80) Return on equity 0,0396 0,0365 0,0384 (0,88) (0,85) (0,84) Tobin_Q 0,0049 0,0052 0,0047 (1,20) (1,18) (1,16) Log Sales 0,0235** 0,0266** 0,0233*** (2,40) (2,53) (2,63) Log Employees -0,0255 -0,0231 -0,0145 (-1,95) (-1,94) (-1,68) Board size 0,0074 0,0051 0,0056 (1,25) (1,18) (1,33) Blockholder -0,0067 -0,0059 -0,0052 (-0,26) (-0,18) (-0,35) Age 0,0001 0,0002 0,0002 (0,34) (0,47) (0,64) R-squared 0,19 0,27 0,27 N 243 242 242
*p < 0.10 (two-tailed), **p < 0.05 (two-tailed) and ***p < 0.01 (two-tailed) Performance pay
performance is linked to CEO compensation (e.g. Jensen and Murphy, 1990). The coefficient of the variable Return on Equity is not significant, so the negative sign is no evidence against the theory by Jensen and Murphy. Perhaps, this result can be interpreted as evidence that Return on equity is a poor (accounting) measure of firm performance, but the insignificant coefficient can also be caused by a lack of data. Tobin’s Q is included in the model as measure for a performance/value effect. In addition, Jensen and Murphy (2004) suggest that the remuneration in stocks and options could also serve as opportunity for the CEO to exploit optimism in the market. This theory should result in a highly significant Tobin’s Q’
coefficient for the remuneration components shares and options. The results do not provide any evidence for this theory, Tobin’s Q is not a significant factor when the ratio of shares and options divided by total remuneration (a.k.a ‘performance pay’) is the dependent variable. The variables Sales and Employees show that firm size is a very important factor for CEO compensation. This finding is in line with previous literature. The size effect was already noticed by Beamol (1967), and has remained essential in cross-section research on CEO remuneration.
The variable Board size is significant positive related to CEO compensation. This is according to results in previous literature that show a positive relation. The main theory is that a larger board size proxies for worse corporate governance, because the productivity of the board decrease when the size of the board increase, and that should result in higher CEO
compensation (Jensen, 1993) after controlling for other factors. Board size cannot serve as another measure of the size effect, because the correlation table (table IX in the section Data) shows that board size is not significantly correlated to the ‘size measures’ Sales and
Employees. The other proxy for the standard of corporate governance, the variable blockholder, shows an expected negative influence on CEO compensation.
6. Conclusion
The aim of this thesis was to study the effect of Anglo-American financial markets on CEO compensation in Dutch firms. Based on the framework by Oxelheim and Randøy and own extensions, I used the four hypotheses from the section literature to find out whether there were Anglo-American effects on CEO compensation for listed Dutch firms in the period 2002-2006.
Using the database of the Dutch Investors Association, I was able to perform a statistical analysis of the framework developed by Oxelheim and Randøy for Dutch firms. Although it was not possible to separate the four channels of Anglo-American influence, (1) institutional contagion or spill over effects (2) demand conditions (3) supply conditions (4) a risk
premium (see introduction), the actual existence of an Anglo-American influence could be measured.
In a univariate test, Dutch firms were split into two groups based on any Anglo-American influence, and the group with Anglo-American influence paid their CEO significantly more. Both measures, an Anglo-American board member dummy and an Anglo-American exchange listing dummy, were able to obtain to this result.
After controlling for factors related to firm performance, firm size and corporate governance, the aggregate Anglo-American influence on CEO compensation of Dutch firms was still significant and positive. However, while the measure Anglo-American board member still reported a positive Anglo-American influence, the measure Anglo-American exchange listing reported a negative effect on CEO compensation. This result passed several robustness
aggregate Anglo-American influence on the relative amount of performance related pay is positive.
This study provides some additional support for the rationale developed by Oxelheim and Randøy, but also raises some questions about the appropriateness of the measures of the Anglo-American influence. The variable Anglo-American board member passed the test and proves to be a good measure of the American influence, but the variable Anglo-American exchange listing failed for this test with Dutch companies. Further research has to explain whether the remarkable results of the variable Anglo-American exchange listing are an exception or frequent.
As with any study, this study has its limitations. Due to limited data availability, the sample size is rather small, measured in number of (listed) firms and/or number of years. And the number of firms with actual Anglo-American influence is, therefore, quite limited.
Furthermore, The Netherlands is just a small country in Continental Europe, so this result can not be generalised for a total corporate governance system.
If further research proves that the results of this study are not an exception, then this study can be used as first evidence that the Anglo-American effects on CEO compensation in other countries are not only positive but also negative. In the introduction, it was explained that internationalization should lead to a sort of catch-up in CEO compensation in Continental Europe, which is currently lower than CEO compensation in Anglo-American countries. After controlling for other factors, this difference was partly caused by differences in corporate governance. The results of this study are not totally contradictory to this theory, but they indicate that it is not the whole story. Apparently, Anglo-American elements of corporate governance can also temper CEO compensation in non-Anglo American countries.
7. Appendix
Appendix 1
Table XVII: Data sources
Variable Main source of data
Remuneration VEB
Anglo-American exchange listing Annual reports Anglo-American board member Annual reports
Annual stock return Datastream
Tobin's Q Amadeus
Sales Amadeus
ROE Amadeus
Board size Annual reports
Blockholder Amadeus
Industry Amadeus
Employees Amadeus
Appendix 2
Table XVIII: Definition of variables
Variable Description
Remuneration
Salary The fixed part of the cash remuneration Bonus The variable part of the cash remuneration
Pension Company’s contribution to the pension plan of the CEO Stock The value of shares granted to the CEO
Options The value of options granted to the CEO
Other remuneration Remuneration not assigned to other categories, the so called "perks"
Independent variables
Anglo-American A dummy for an Anglo-American director board member
Anglo-American A dummy for an Anglo-American exchange listing of the firm exchange listing
Control variables
Annual stock The difference between the stock price on 31 December
return of the year and the opening stock price on 1 January of the year, divided by the stock price on 1 January
Return on equity The return on shareholder's funds is used as a proxy for the ROE Tobin_Q The market capitalisation per share divided by the book value per share Sales All revenues from the firm's core activities per year
Employees The number of employees of the firm Board size The number of board of directors
Blockholder A dummy that is rewarded a one when a shareholder of the company owns more than 20% of the shares
Appendix 3
Table XVIV
Anglo-American influences on CEO remuneration components (lagged control variables)
Model: Ln (compensation) = a + b1*Anglo-American board memberi,t-1 + b2*Annual Stock Returni,t-1 + b3*Return on Equityi,t-1 + b4*Tobin_Qi,t-1 + b5*Ln (Sales)i,t-1 + b6*Ln (Employees)i,t-1 + b7*Board sizei,t-1 + b8*Blockholderi,t-1 + b9*Agei,t-1
The dependent variables are not directly put into the Panel-OLS regression, but by taking the natural logarithm of the variables. The regressions concerning the different parts of the remuneration are performed without the observations with zeros on that specific remuneration component. The independent variables are lagging behind one year.
Salary Bonus Pension Shares Options Other Variable Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) Anglo-American 0,1227*** 0,1798* 0,0422 0,4295 0,2808 -0,0984 board member (3,33) (1,70) (0,37) (1,43) (1,34) (-0,23) Anglo-American - - - -exchange listing - - - -Control variables Annual stock 0,0484* 0,2724*** 0,0145 0,4941 0,3750* -0,1412 return (1,72) (3,33) (0,18) (1,65) (1,90) (-0,61) Return on equity -0,0201 -0,1687* 0,2024* -0,0874 -0,1411 0,0611 (-1,00) (-1,68) (1,91) (-0,23) (-0,55) (0,43) Tobin_Q 0,0006 0,0562*** -0,0052 0,0632* 0,2011*** 0,0856 (0,35) (3,39) (-0,94) (1,96) (4,31) (1,34) Log Sales 0,0963*** 0,2102*** 0,0616** 0,1032 0,1525** -0,1095 (10,33) (6,92) (1,99) (1,28) (2,14) (-1,46) Log Employees 0,0698*** -0,0057 0,1835*** 0,1170 -0,0019 0,1743* (5,95) (-0,16) (5,06) (0,90) (-0,02) (1,89) Board size 0,0789*** 0,2004*** 0,0694** 0,0307 0,1024** 0,1530** (9,45) (8,58) (2,57) (0,61) (2,33) (2,29) Blockholder -0,0706** -0,2052** -0,2004** -0,4059 -0,3611* -0,0014 (-2,38) (-2,39) (-2,14) (-1,50) (-1,89) (-0,01) Age -0,0004** -0,0002 0,0004 0,0009 0,0041*** 0,0005 (-2,31) (-0,42) (0,71) (0,49) (2,89) (0,25) R-squared 0,78 0,57 0,42 0,31 0,41 0,21 N 491 402 389 100 196 123
Table XX
Anglo-American influences on CEO remuneration components (lagged control variables)
Model: Ln (compensation) = a + b1*Anglo-American exchange listingi,t-1 + b2*Annual Stock Returni,t-1 + b3*Return on Equityi,t-1 + b4*Tobin_Qi,t-1 + b5*Ln (Sales)i,t-1 + b6*Ln (Employees)i,t-1 + b7*Board sizei,t-1 + b8*Blockholderi,t-1 + b9*Agei,t-1
The dependent variables are not directly put into the Panel-OLS regression, but by taking the natural logarithm of the variables. The regressions concerning the different parts of the remuneration are performed without the observations with zeros on that specific remuneration component. The independent variables are lagging behind one year
Salary Bonus Pension Shares Options Other Variable Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
(t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic) (t-statistic)
Anglo-American - - - -board member - - - -Anglo-American -0,0061 -0,1292 0,0708 -0,4424 -0,4496* -0,1651 exchange listing (-0,12) (-0,86) (0,47) (-1,21) (-1,72) (-0,42) Control variables Annual stock 0,0529* 0,2805*** 0,0171 0,6368** 0,4331** -0,1319 return (1,85) (3,42) (0,18) (2,14) (2,21) (-0,58) Return on equity -0,0224 -0,1753* 0,1980* -0,2822 -0,1830 0,0663 (-1,10) (-1,74) (1,86) (-0,73) (-0,72) (0,46) Tobin_Q 0,0009 0,0597*** -0,005 0,0684** 0,1811 0,0777 (0,54) (3,61) (-0,90) (2,15) (3,76) (1,19) Log Sales 0,0948*** 0,2129*** 0,0572* 0,1506 0,1802** -0,0985 (9,81) (6,92) (1,80) (1,63) (2,44) (-1,26) Log Employees 0,0749*** 0,0005 0,1896*** 0,1238 -0,0007 0,1604* (6,30) (0,01) (5,23) (0,93) (-0,07) (1,84) Board size 0,0884*** 0,2163*** 0,0703 0,0767 0,1288*** 0,1495** (11,02) (9,56) (2,69) (1,42) (3,07) (2,35) Blockholder -0,0859*** -0,2253*** -0,2022** -0,3937 -0,4658** 0,0112 (-2,89) (-2,63) (-2,18) (-1,43) (-2,48) (0,03) Age -0,0004** -0,0003 0,0004 0,0003 0,0034** 0,0002 (-2,28) (-0,57) (0,79) (0,16) (2,39) (0,09) R-squared 0,77 0,56 0,42 0,3 0,4 0,21 N 491 402 389 100 196 123
References:
Aguilera, R.V. and Jackson, J. (2003) ‘The Cross-National Diversity of Corporate
Governance: Dimensions and Determinants’, Academy of Management Review, vol. 28: 447-465
Agarwal, N.C. (1981) ‘Determinants of executive compensation’, Industrial Relations, vol. 20: 36–45
Baumol, W.L. (1967) ‘Business Behavior, Value and Growth’, Harcourt, Brace and World: New YorkMacmillan, 1959.
Bebchuk, L. and Fried J. (2003) ‘Executive Compensation as an Agency Problem’, Journal of Economic Perspectives , vol. 17: 71-92
Core, J.E., Holthausen, R.W. and Larcker, D.F. (1999) ‘Corporate governance, Chief Executive Officer compensation, and firm performance’, Journal of Financial Economics , vol. 51: 371–406
Conyon M.J.and Schwalbach J. (1997) ‘European differences in executive pay and corporate governance’, E-Books
Economist (1999) ’No man is an Island’ vol. 351 Issue 8118, Pay Survey: 17-18
Jensen, M. and Meckling W. (1976) ‘Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure’, Journal of Financial Economics , vol. 3: 305-360
Jensen, M.C. and Murphy, K.J. (1990) ‘Performance pay and top-management incentives’, Journal of Political Economy, vol. 98: 225–264.
Jensen, M.C. and Murphy, K.J. (2004) ‘Remuneration: Where We've Been, How We Got to Here, What are the Problems, and How to Fix Them’, ECGI, Finance Working Paper No. 44/2004
La Porta, R., Lopez-de-Silanes, F. and Shleifer, A. (1999) ‘Corporate ownership around the world’, Journal of Finance, vol. 54: 471–517.
Lucier, L., Schuyt, R. and Handa, J. (2004) ‘The Perils of ‘Good’ Governance’, Annual Studies on CEO Succession: London, Booz Allen Hamilton.
Ozkan N. (2006) ‘Do corporate governance mechanisms influence CEO compensation? An empirical investigation of UK companies’, Journal of Multinational Financial Management, vol. 17: 349-364
Oxelheim, L. and Randøy, T. (2003) ‘The impact of foreign board membership on firm value’, Journal of Banking and Finance, vol. 27: 2369–2392.
Oxelheim L. and Randøy T. (2006) ‘The Anglo-American Financial Influence on CEO Compensation in non-Anglo-American Firms’, Journal of International Business Studies, vol 36: 470-483.
Randøy, T. and Nielsen, J. (2002) ‘Company performance, corporate governance, and CEO compensation in Norway and Sweden’, Journal of Management and Governance, vol. 6: 57– 81.
Roberts, D. (1956) ‘A General Theory of Executive Compensation Based on Statistically Tested Propositions’, Quarterly Journal of Economics, LXX (May, 1956): 270-294.
Sanders, W.G. and Carpenter, M.A. (1996) ‘Internationalization and Firm Governance: The Roles of CEO compensation, Top Team Composition and Board Structure, Academy of Management Journal, vol. 41: 158-178
