The influence of blockholder ownership on
CEO compensation in the Netherlands
University of Groningen, Faculty of Economics and Business University of Uppsala, Faculty of Social Sciences
Master Program International Financial Management MSc International Business and Management
MSc Business and Economics
Keywords: CEO compensation, corporate governance, ownership, blockholders JEL classification: G32, J33
Arjan J.J. Reinders, s1538160, A.J.J.Reinders@student.rug.nl January 2010
2
Abstract
This research looks at the influence of blockholders on CEO compensation. The sample consists of 100 Dutch firms, ranging from large cap (AEX), to midcap (AMX), small cap (AScX) and local firms. For each firm data was gathered over the years 2006-2008, resulting in 300 firm years of data. I take two perspectives regarding compensation, namely the level of total compensation, and the ratio of variable to total compensation. Concerning blockholders I made a detailed analysis in relation to the involvement of the blockholders in the firm (are they in the board of management, board of directors, or outsiders), the corporate governance system at the blockholders’ home country (Anglo-Saxon or Germanic), and whether the blockholders are institutional or not.
Several control variables are included to increase the accurateness of the model, such as sales, return on equity, board size, board size squared, leverage ratio, foundation dummy, CEO tenure, and new CEO appointment. All of the control variables were significant in at least one of the models, and many in most, or even all, of the models ran. Especially sales, return on equity, board size, board size squared, and new CEO appointment turned out have significant influences.
The model shows that several blockholders indeed have a significant influence on the compensation of CEOs. Blockholders in the board of directors have negative influence on the level of compensation, as do Germanic blockholders. Germanic blockholders also have a negative influence on the proportion of variable to total compensation. The other blockholder variables did not show any significant relations, although this changed when the year 2008 was excluded. A year when the worst financial crisis since the 1930’s scared investors and led to declining profits of companies, 2008 can be called exceptional and might therefore change the results. In general the results were similar, although both institutional and non-institutional blockholders turned out to have a significant negative influence on the proportion of variable to total compensation.
3 Table of Contents 1. Introduction 4 2. Literature Review 5 2.1 Executive Compensation 5 2.2 Shareholder Monitoring 8
2.3 The Influence of Blockholders 9
2.4 Institutional Ownership 10
2.5 Country of Origin Influence 12
3. Problem Indication and Research Question 12
4. Contribution of This Paper 13
5. Research Hypotheses 14
6. Research Methodology 17
7. Data 21
7.1 Data sources 21
7.2 Measuring the components of CEO compensation 22
7.3 Sample characteristics 23
7.4 Blockholders 26
8. Analysis and Results 29
9. Robustness Check 32
10. Conclusion 34
11. Limitations and Recommendations 36
References 38
Appendix 1: List of Companies Included in Dataset 43
Appendix 2: List of Blockholders with their Ownership Percentages 44
4
1. Introduction
Tirole (2001) defines corporate governance as "the design of institutions that induce or force management to internalize welfare of stakeholders." Executive compensation is one of the issues corporate governance deals with.
During the last decades executive compensation has received a significant amount of attention, not only from the media and general public, but also from academic theorists. One of the questions that keeps popping up is whether corporate governance systems have influence on the determination of executive compensation, and if so, what this influence is. As can be concluded from the definition of Tirole (2001), corporate governance systems aim at reducing agency conflicts between management and shareholders. Executive compensation is one of the mechanisms used to decrease the agency conflict. Other factors of influence are ownership and board structure.
In this paper I aim to take a closer look at the relation between executive compensation and ownership structure. Especially, I study the influence of large shareholders, so called blockholders, on compensation. Whereas several researchers incorporated the presence of blockholders in their research as control variables, or analyzed the effect of blockholders on a single dimension of executive compensation - the absolute level - only few took into account the influence of different types of blockholders on composition and level of executive compensation. This paper aspires to expand the current body of literature by analyzing these relations in the Netherlands. The way I do this is by performing regressions on the level of compensation of CEOs and the proportion of variable pay (cash bonus, options, and shares granted) to total pay.
5
2. Literature Review
2.1 Executive compensation
During the last decades executive compensation increased strongly (Economist 2000). The main theory dealing with executive compensation is the principal agency model and an extension of it: the managerial power view. The principal agency theory starts from the separation of ownership and control and tries to explain the characteristics of executive compensation by focusing on bridging the gap that this separation created. When ownership and management are separated, this might give substantial power to managers (Berle and Means, 1932). Managers have their own incentives, and those might not be the same incentives as investors have. The costs associated with the divergent management-shareholder objectives are called agency costs and they consist of two important sources. The first source involves the costs that are inherently associated with using an agent. Managers might use company resources for their own benefit. An example of this can be a private jet: it could be more cost-effective if management flies commercial, however, this causes more inconvenience for the executives. Another example of using organizational resources for the CEO’s personal benefit can be the acquisition of another company. If the acquiring company pays a too high price, the shareholders of this company will suffer. The CEO, however, might gain the prestige that comes with managing a larger firm (empire building).
The second source involves the costs of techniques in order to mitigate problems that can arise when using an agent. The divergence in objectives between executives, or agents, and stakeholders, make shareholders want to conciliate these incentives again. One way is by producing financial statements, so that investors have access to financial information about the company. Another, very important way to conciliate is by creating an efficient compensation package, called optimal contracting approach (Bebchuk and Fried, 2003). The agency theory focuses on how to bridge the incentive gap between owners and agents (for example: Fama, 1980; Fama and Jensen, 1983). Jensen and Murphy (1990a) suggest that in order to give managers the correct incentive to maximize firm value, compensation should be equity based, rather than cash based (see also Himmelberg, Hubbard and Palia, 1999).
6
agency problem, but also partly a cause of the agency problem, meaning that certain parts of observed compensation agreements actually increase the agency problem rather than decreasing it. In line with this statement, they claim that their extension of the theory can better explain certain puzzling characteristics of compensation than the optimal contracting approach (the view that pay arrangements are a remedy to the agency problem). For example why exercise prices of options are set at grant-date market prices and not indexed to general market movements. A CEO can get a windfall gain when the market sentiment is positive, and all shares go up. When share prices decline, the options are regularly reset, in order to maintain the income of the CEO. Another example of managerial power is the right of executives to exercise their options early and diversify their portfolio. This way the increased incentives from the options become undone again. Hence, the managerial power approach does not only view executive compensation as an instrument for solving the agency problem, but partly views the compensation structure as a cause of the agency problem.
CEOs are, as other individuals, seen as risk-averse people. This means that they want to structure their compensation in such a way that they bear the least personal risk (Harris and Raviv, 1979). The consequence of managers heavily influencing their own pay is reflected in the compensation mix, which reflects a CEOs preference for maximizing cash payments, and at the same time minimizing the long-term component of pay (Beatty and Zajac, 1994; Zajac and Westphal, 1994). A number of researchers have argued that compensation schemes seem to reflect managerial rent-seeking instead of providing efficient incentives (e.g. Yermack, 1997; Bertrand and Mullainathan, 2001). Bebchuck and Fried argue in their book Pay without Performance (2004) that directors cannot be expected to engage in arm’s-length bargaining with managers over their compensation, given the power that executives hold. For example, chief executives can often influence the appointment of a director. Also, a director in general works more closely with management than with shareholders, and challenging a director is a very difficult and costly process for shareholders. Finally, CEOs are often directors in other companies, meaning that two CEOs can be directors in each other’s board. Concluding, due to the interaction and interdependence between CEOs and directors, directors are not likely to bargain at arm’s-length.
7
markets during the 1990’s and large additional compensation -. The reason why equity compensation was an additional reward is that the growth in equity-based compensation was not accompanied by a substitution effect, i.e. a reduction in non-equity compensation. The consequence was that executive compensation increased beyond what can be explained by changes in firm size and performance.
Even after extensive research there is still no consensus about the determinants of CEO pay, although some determinants are widely observed. Company size is one of those determinants. Already in 1959 Baumol observed the influence of size on executive compensation. Some later studies confirmed the relationship (e.g. Ciscel and Carrol, 1980; Conyon and Murphy, 2000), rationalizing that managing larger firms leads to greater demands from CEOs, that larger firms have a greater ability to pay, or that more hierarchical layers exist within the firm, leading to higher compensation at every following hierarchical level (Fox, 1983). Nevertheless, Jensen and Murphy (1990b) found that the relationship between size and compensation has not been persistent across firms, as for some CEOs performance is (nearly) the only determinator of compensation. As an example they mention David H. Murdock of Castle & Cooke, Inc., who is nearly completely dependent on performance of the firm for his compensation. A more recent example of a CEO whose compensation depends on firm performance is Steve Jobs. In 2006 Mr. Jobs earned a fixed salary of $1. However, in the same year he was awarded $646 million in stock compensation, dwarfing his fixed salary. Where Jensen and Murphy argue that the relationship between size and compensation has not been persistent, Gabaix and Landier (2008) state that size explains many of the patterns in CEO pay. Size can explain patterns across firms, over time, and between countries. Moreover, Gabaix and Landier find that the increase of CEO pay between 1980 and 2003 can be fully attributed to the increase in market capitalization of large companies during that same period.
Economic performance of the firm has an even less pervasive relationship with executive compensation, results varying from nil to strongly positive relations. Murphy (1985) and Finkelstein and Hambrick (1989) are two examples of studies that documented that CEOs are in general rewarded for corporate performance. Murphy noted that this is mainly the case through the shareholdings of the chief executive in the firm, but Finkelstein and Hambrick find that cash bonuses vary depending on performance. Hall and Liebman (1998) find that CEO compensation is highly related to the performance of the firm. In their sample an increase in firm performance from a median stock price performance to a 70th percentile performance increases CEO pay by nearly 50 percent. Consequently, corporate performance is often included as a determining variable. Other determinants are less often included, but examples are shareholder return, CEO tenure, and board size.
8
remuneration paid to executives. They analyzed about 500 firm-years over the period 1998-2001, a period before the Dutch law forced companies to publicize the compensation of top executives. They do not find a positive pay-performance relationship, which they argue with the managerial power theory: managers are able to influence their own compensation package, as there is little corporate governance pressure. On 30 December 2004, however, a new corporate governance was implemented: Code Tabaksblat. This new code aimed to increase the power and protection of shareholders, to empower the supervisory board, and to make corporate governance more transparent. If the Code Tabaksblat is successful, I would expect a positive pay-performance relationship to exist for the years after 2004. As my study covers the years 2006-2008, it is interesting to see whether this relationship indeed exists now.
2.2 shareholder monitoring
In theory all stakeholders of the company should be able to monitor the agents and are consequently well informed. Nevertheless, in reality firm ownership is often dispersed, which can make monitoring for small shareholders unprofitable, as monitoring involves costs. A clear example of such a cost is the cost involved with attending the annual shareholder meeting, as one will have to invest time and money in order to attend. The costs involved with monitoring a firm can lead to a free-rider problem, where every shareholder expects others to monitor, and hopes to benefit from this himself. If other shareholders monitor management, and take actions that benefit the shareholders, shareholders who do not monitor – and consequently do not have to invest the time and money to do this – will share in the benefits, but save the costs. With no large shareholders present, a situation might be created where managers are not being closely monitored at all, giving them considerable freedom. Bebchuk (2005) shows that shareholder’s power in companies with dispersed ownership is insufficient to ensure that value-increasing arrangements that management disfavors are adopted – for example the sale of an unprofitable subsidiary. Managers might disfavor this as managing a larger firm increases their prestige and reduces their risk of being dependent on a single business unit -. Apart from the problems associated with dispersed ownership, the power granted to shareholders is only a veto power, meaning that they can veto a decision, but lack the power to initiate a resolution. In companies with dispersed ownership, it is difficult to get a majority of the shareholders to challenge an initiative or a director, leading to a lack of power for shareholders (Bebchuk, 2005).
9
stocks easily. When they sell their position, the large quantities of shares being sold will lead to a decline in the share price, and consequently to a decline in the value of their holdings (Lane, Cannella, and Lubatkin, 1998). Hence, one can expect that corporate governance and thereby executive compensation might differ in the presence of blockholders. Previous research has indicated that this is indeed the case: CEOs of owner-controlled firms - firms with a blockholder present - have been shown to have lower levels of compensation than CEOs of manager-controlled firms - firms without a blockholder – (see for example Dyl, 1988; Hambrick and Finkelstein, 1995).
Jensen and Meckling (1976) state that executive compensation, ownership structure, and board composition are influencing each other. The issue of blockholder ownership is also highly relevant when looking at the dispersion of ownership: with a 10% share cutoff level, only 50% of US firms are widely held (La Porta et al., 1999). In the rest of the world this is even less: In European countries (for example: Spain 0%, Norway 10%, United Kingdom 10%,), Asian countries (e.g. Hong Kong 0%, Singapore 10%, Japan 20%), and countries in Oceania (New Zealand 0% and Australia 10%), widely hold firms are actually scarce (La Porta et al., 1999). If the presence of blockholders can be seen as a solution to overcome the agency problem, this could possibly explain why widely held firms are uncommon in many countries.
2.3 The influence of blockholders
There are several papers that find that blockholders can have an influential role on executive compensation level and structure. Both the incentive to maximize profits and control are thereby given. Bethel and Liebkind (1993) found that large blockholders are able to put pressure on management and influence its strategies. They state that blockholders can exercise a disciplinary effect on management. An illustration of blockholders pushing management is the restructuring of companies through downsizing and reduction in diversification in the 1980s. Kang and Shivdasani (1995) and Kaplan and Minton (1994) found an increasing turnover rate of managers after bad performance when blockholders are present. They conclude that those findings are evidence that large blockholders increase the monitoring of executives.
Sanders and Carpenter (1996) analyzed the relation of internationalization of a company and executive compensation, thereby incorporating blockholders defined as a 5 percent ownership of outstanding equity. The authors observe that the level of compensation of CEOs decreases when blockholders are present, on a 5 percent significance level.
10
the effect that blockholders have on the compensation structure, Chourou et al. (2008) analyze the economic factors of CEO stock option compensation in Canada over the years 2001-2004 by employing a Tobit model. They find a positive relation between the ratio of annual CEO stock option payments to cash compensation and the presence of blockholders, indicating that option payments are more often used to align incentives in firms with blockholders than in firms without blockholders.
Based on the principal agency theory, Khan et al. (2002) researched the topic of blockholders by looking at the relation between institutional ownership and CEO compensation and the ratio of options to total compensation. They conclude that large blockholder concentration reduces the level of CEO compensation and at the same time leads to a higher proportion of fixed salary to overall compensation, which contradicts to the findings of Chourou et al. (2008). Even though Khan et al. (2002) find that the presence of blockholders influences compensation, they do not find a relation between the number of blockholders and CEO compensation.
2.4 Institutional ownership
Institutional ownership concentration, which is a form of blockholder ownership, has been the topic of extensive research. The reason is that institutional investors became increasingly powerful during the previous decades. Their aggregate ownership of U.S. equity increased from 16% in 1965 to 57% in 1994 (Useem, 1996), and 76.4% by year-end 2007 (IIR, 2008). In the United Kingdom institutional investors owned around 80% of equity as of December 2003 (ONS 2004). Institutional investors are a heterogeneous group of organizations, and include public and private pension funds, banks, mutual funds, and insurance companies.
11
much attention to monitoring a single individual firm, and hence rely more on pay-for-performance as an alternative to monitoring.
Ozkan (2007) examines the influence of ownership and board structure of companies on the level of CEO compensation for a sample of 414 large UK companies. He finds that both institutional ownership and blockholder ownership have a negative impact on CEO compensation. Those results are consistent with the theory that blockholders and institutional shareholders are actively monitoring the company in which they invest and can exert pressure on management. This way they can prevent management from expropriating wealth from shareholders in the form of excess pay. The research of Ozkan also shows that CEO compensation is lower when the director’s ownership is higher. Directors are also able to exert pressure on management, and when their ownership stake increases, so does their interest to actually apply this pressure and prevent the expropriation of shareholders.
Furthermore, Kyonghee (2005) finds that not only the presence of blockholders on their own affects CEO compensation, but also the blockholders’ investment horizon in years. She observes a negative relation between investment horizon and level of CEO compensation and payment in stock options. Wallace (1994) also finds a negative correlation between the presence of insiders and outside blockholders and the proportion of equity based compensation.
The results for the relation between blockholders with a seat on the board of management and CEO compensation are mixed. For example, Holderness and Sheehan (1988) find that managers who are a majority shareholder in their own company receive higher salaries than other executives. This can be interpreted as managers who use corporate resources to benefit themselves, by rewarding themselves higher compensation. Allen (1981), on the other hand, finds that the level of CEO compensation decreases when the CEO owns more equity in the firm. Yet another research, performed by Cheung et al. (2005) in Hong Kong, suggests that CEOs with considerable equity ownership use dividends as a way to supplement their cash compensation, and are consequently less concerned with the level of their cash compensation.
12
compensation, and the composition of the board of directors. As a final point, Mehran found that the more outside directors there are present in the firm, the higher the percentage of executive compensation that is equity-based.
2.5 Country of origin influence
Anglo-American influences are also seen as an influence on both the composition and level of pay (Oxelheim and Randøy, 2005). An Anglo-American firm in general pays a higher salary, where the variable pay is more important. One of the reasons for this is that the Anglo-American corporate governance system is less tolerant to poor performance (Lucier et al., 2004). Therefore a rational CEO will know a priori that he/she may be more severely penalized for bad performance and demand a higher premium for the risk of dismissal. The results of Oxelheim and Randøy might also be relevant for the background of blockholders. Anglo-Saxon blockholders bring their own values with them and are likely to be more willing to pay a higher salary, with the variable component taking a more central place.
3. Problem Indication and Research Questions
I will research the influence of different kind of blockholders on executive compensation in the Netherlands. First of all, I test the influence of three kinds of blockholders, namely blockholder in the Board of Management, blockholders on the Board of Directors, and outsider blockholders. I distinguish between those types of blockholders since I expect them to have a different influence on executive compensation. If compensation really is an alternative to active monitoring, one would for example expect that outsider blockholders pay a higher percentage of variable pay than blockholders that are in the supervisory board, or are even part of management. This is confirmed by a previous study of Tosi and Gomez-Mejia (1989), who found that management-controlled firms (firms with blockholders in the board of management) have less at-risk CEO pay, meaning that the variable component of compensation was relatively low.
13
system). As shown by the study of Oxelheim and Randøy (2005) Anglo-Saxon influences often lead to higher compensation levels and a larger variable part than a Germanic background.
Thirdly, I will divide the blockholders in institutional and non-institutional blockholders and see whether these types of blockholder have a different influence on executive compensation. It is interesting to study the case of institutional blockholders separately, as they are distinctively different from other blockholders. Institutional blockholders, being pension funds, banks, mutual funds, and insurance companies, are likely to act differently, as they often have significant holdings in multiple firms. The reason for this is that institutional investors often manage large sums of money. Mutual funds regularly have several billion under management, and pension funds tens of billions (The Dutch pension fund ABP had after the second quarter of 2009 181 billion euro under management; ABP, 2009). Consequently, these institutional investors will not be able to spend as much time and effort in monitoring a single firm as a non-institutional blockholder, who often has only one or a few significant holdings. Non-institutional blockholders are usually individuals, founders of the company, and rich private investors. As they manage their own wealth, their total amount of investments is typically a great deal lower than the investments of institutional investors.
Concerning, compensation, I look at two measures. The first one is the level of total compensation, which includes fixed salary, annual cash bonus, stock awards, option awards, pension payments, and other forms of compensation (company car, company apartment, expense allowances, etc.). The second measure is the proportion of variable to total compensation. I define variable compensation as the sum of annual bonus, stock awards, and option awards, as those compensation components are more or less dependent on performance.
Lastly, it is necessary to define a blockholder. I take a cutoff rate of 10% of the voting rights. To begin with, block holdings over 5% are publicly available due to the Wet Melding Zeggenschap of the Netherlands. The reason I use a cut-off rate of 10% instead of 5% is that a blockholder who controls 10% of the voting rights has a considerable influence on firm policy (La Porta et al., 1999), whereas the influence of a blockholder with 5% of the votes is more doubtful.
4. Contribution of this paper
14
blockholders are regularly included as a control variable in models, studies that put blockholders central are scarce, especially in the Netherlands. One of the reasons for this is that the law stating that ownership stakes above 5% have to be public, the so called Wet Melding Zeggenschap, was effected in 2006. Information about blockholders before 2006 is therefore not known. As one would need several years of firm data in order to be able to perform a sound analysis, there has been little opportunity to perform such an analysis yet: the annual reports over the year 2008 are available since May/June 2009.
In a more international perspective, there has been no research that specifies blockholders as detailed as I do. The distinction institutional/non-institutional is quite often made, as is Board of Management/Board of Directors. However, there has been no research that looks at the origin of blockholders and none of the papers existent cover all the distinctions. A research usually includes only one distinction of blockholders, and this paper is the first that covers a whole range of blockholder characteristics.
5. Research Hypotheses
As mentioned before, I study the influence of three kinds of blockholders (blockholders in the Board of Management, blockholders in the Board of Directors, and outsider blockholders), from two origins (Anglo-Saxon corporate governance system and Germanic corporate governance system), and with two different characteristics (institutional and non-institutional) on two measures of CEO compensation (level and composition). This leads to the following 7 by 2 matrix with hypotheses (table 1):
TABLE 1 Hypotheses
Hypotheses. Plus or minus is the expected deviation from base case scenario: no blockholders present
15
I argue my hypotheses with the agency and managerial power theory. When a blockholder is a member of the board of management he could use his power as a blockholder to extract wealth from other shareholders by setting high levels of salary for management. Hence, I expect this blockholder to have a positive influence on the level of compensation. For the proportion of variable to total pay the effect is opposite. A manager prefers a fixed, secure income over an uncertain income. Therefore, when the manager is also a blockholder, he will use his influence to increase the proportion of fixed pay, and hence, lower the proportion of variable pay. This leads to hypotheses 1A and 1B:
Hypothesis 1A: A blockholder who holds a seat on the board of management has a positive influence on the level of compensation.
Hypothesis 1B: A blockholder who holds a seat on the board of management has a negative influence on the proportion of variable to total compensation.
Blockholders who have a seat in the board of directors are in an excellent position to control their stake in the firm. They can easily and effectively monitor management and take actions. They also have the power to block compensation increases for CEOs. Therefore, I expect that blockholders with a seat on the board of directors have a negative influence on the level of compensation. Agency theory states that variable compensation is used as an alternative to monitoring management. By awarding a variable pay, the manager is given the incentive to act in the interest of the shareholders. Even though a blockholder with a seat on the board of directors is in a position to actively monitor management, it is likely that blockholders who take the effort of monitoring do so because they try to control their investment optimally, increasing the possibility that they employ both actively monitoring management, and make sure that the interests of managers are aligned with their own interests through variable compensation. Hence, I hypothesize that blockholders who are also director have a positive influence on the proportion of variable to total compensation.
Hypothesis 2A: A blockholder who holds a seat on the board of directors has a negative influence on the level of compensation
Hypothesis 2B: A blockholder who holds a seat on the board of directors has a positive influence on the proportion of variable to total compensation.
16
compensation, and as they would like to align the interest of managers with their own interests, I expect outsider blockholders to have a positive influence on the proportion of variable to total compensation.
Hypothesis 3A: Outsider blockholders have a negative influence on the level of CEO compensation Hypothesis 3B: Outsider blockholders have a positive influence on the proportion of variable to total compensation.
Concerning the origin of blockholders, I expect that the blockholders bring their ideas about compensation from their own corporate governance system with them. This implies that companies with an Anglo-Saxon blockholder will pay a higher remuneration, and that the variable component is relatively large as well (see for example Oxelheim and Randøy, 2005), as this is the case in Anglo-Saxon countries, compared to the Netherlands. For Germanic blockholders, who are used to more conservative compensation packages, I expect a lower total compensation (Oxelheim and Randøy, 2005). Even though Germanic blockholders pay lower levels of compensation, they still want to align the interests of managers with their own interests, and will therefore probably include a larger variable component in the compensation package compared to a scenario with no blockholders. Summed up this leads to hypotheses 4 and 5:
Hypothesis 4A: Anglo-Saxon blockholders have a positive influence on the level of CEO compensation Hypothesis 4B: Anglo-Saxon blockholders have a positive influence on the proportion of variable to total compensation
Hypothesis 5A: Germanic blockholders have a negative influence on the level of CEO compensation Hypothesis 5B: Germanic blockholders have a positive influence on the proportion of variable to total compensation.
17
Furthermore, those non-institutional investors, like all blockholders, hold the power to influence the compensation level, and are therefore more likely to block pay raises. I therefore also expect them to have a negative influence on the level of compensation. Institutional investors will most likely also have a negative influence on the level of compensation. The reason why I expect this is that institutional investors have responsibilities to their investors, and would like to both prevent rent extraction by management and avoid the harmful publicity that comes with high executive pay. They are therefore likely to prevent excessive compensation schemes.
Hypothesis 6A: Institutional blockholders have a negative influence on the level of CEO compensation Hypothesis 6B: Institutional blockholders have a positive influence on the proportion of variable to total compensation
Hypothesis 7A: Non-institutional blockholders have a negative influence on the level of CEO compensation
Hypothesis 7B: Non-institutional blockholders have a negative influence on the proportion of variable to total compensation
6. Research Methodology
In order to test whether blockholder ownership has an influence of the level and composition of CEO compensation, I test two regressions: one for the level of payment and one for the structure. I run three models: the first with three variables for BOM blockholders, BOD blockholders, and outsider blockholders; the second with two variables for Anglo-Saxon blockholders and for Germanic blockholders; and the last with two variables for Institutional blockholders and Non-institutional blockholders. The values I use for those variables are the ownership percentages that these investors have.
18
annual report over the year 2008. The same line of reasoning is used for the years 2006 and 2007. Another variable that is often included in comparable analyses is board size, as board size is expected to have an influence in determining the effectiveness of monitoring management. Jensen (1993), for example, argues that corporate governance is affected by board size, independent of other variables. This argument stems from the productivity loss associated with larger work groups: as work groups increase in size, they become less effective because of communication, coordination, and process problems. Hence, I also include this variable. I also include board size squared, as the relationship between board size and compensation might not be linear. Even though larger boards of directors might become less effective, this is likely to only happen after a certain size; in other words, supervisory boards might initially become more effective when size increases, but after a certain threshold, due to the problems mentioned before, they become less effective again. Squaring board size measures this non-linear relationship. My final company characteristics variable is leverage, defined as liabilities/assets. In a firm with a high leverage, the CEO has considerable less freedom to act, as he or she is forced to work efficiently in order to make payments to the creditors of the company (Georgantzas and Ritchie-Dunham, 2002). Also, debt holders are likely to closely monitor managerial activities thereby reducing the payment of excess compensation (Kabir and Duffhues, 2008). In addition, managing a highly leveraged firm requires more skills, as mistakes can have more severe consequences. Leverage increases the risk of the firm, which in turn makes the payment of higher compensation necessary.
A blockholder control variable is a dummy for foundations created to guarantee the continuity of the company. Those foundations, often called Stichting Administratiekantoor, Stichting Continuiteit, or Stichting preferente aandelen in Dutch, are created for protection reasons. They have votes in the company (usually 100%, meaning that they can block every decision from the regular shareholders), but no economic rights to the dividends. The foundations are usually not taking part in decision making processes, except in stressful situations, such as a hostile takeover. As they foundations theoretically do have a vote, and therefore might influence compensation of CEOs, I included them as a dummy, called foundations dummy.
19
his forerunner, a drop of 43%. Table 2 summarizes my model with the dependent, independent, and control variables.
TABLE 2
Model and Variables
Dependent variables Independent variables Control variables
Y1 = Level of CEO compensation Involvement:
X1 = BOM blockholder percentage X2 = BOD blockholder percentage X3 = Outsider blockholder percentage
Firm Characteristics: X4 = Firm size (sales) X5 = Profitability (ROE) X6 = Board size
X7 = Board size squared X8 = Leverage
Y2 = Proportion of variable pay to total pay of CEO
Origin:
X1 = Anglo-Saxon blockholder percentage
X2 = Germanic blockholder percentage
Blockholder control variables: X9 = Foundations dummy Institutionalization: X1 = Institutional blockholder percentage X2 = Non-institutional blockholder percentage CEO characteristics: X10 = CEO tenure X11 = New CEO
As the dataset involves both time series and cross-sectional elements it is a panel of data. Panel data might lead to several problems when using an Ordinary Least Squares method, which can be solved using fixed effect or random effect approaches. To see whether I need to use one of these approaches I perform several tests, using Eviews. First of all, my data is a balanced panel, meaning that it has the same number of time-series observations for each cross-sectional unit. To determine whether fixed effects are necessary or not I run a redundant fixed effects test. The redundant fixed effects test has a null hypothesis that the simple, restrictive model is appropriate. This test indicates a probability for the F-test of 0.5073 and 0.4907 for the Chi-square test concerning the level of compensation, meaning that I cannot reject the null hypothesis and that I can employ a pooled sample. The probabilities for the variable to total compensation regression are 0.2655 for the F-test and 0.2488 for the Chi-square test, also indicating that a normal pooled sample can be used.
20
of those regressions is run separately for the involvement of blockholders, the origin of blockholders, and the type of blockholders, meaning that a total of six regressions are run.
The regressions run are as follows:
Y1 = αX1 + βX2 + γX3 + δX4 +εX5 + ζX6 + ηX7 + θX8 + ιX9 + κX10 + λX11 And
Y2 = αX1 + βX2 + γX3 + δX4 +εX5 + ζX6 + ηX7 + θX8 + ιX9 + κX10 + λX11
The control variables are all included in every regression. I do not run a regression with all the variables by which I define blockholders (involvement, origin, type) included at once, as this will lead to a statistical problem known as multicollinearity (see table 3). Institutional blockholders show a high correlation with outsiders and Anglo-Saxon blockholders, whereas non-institutional blockholders show a high correlation with Board of Directors and Germanic blockholders.
TABLE 3
Correlations blockholder variables
BoM BoD Outsider Germanic
Anglo-Saxon Institutional Non-institutional BoM 1.00 .209 -.191 .376 .256 -.002 .517 BoD 1.00 -.202 .395 .362 .049 .559 Outsider 1.00 .403 .208 .586 .109 Germanic 1.00 -.143 .331 .685 Anglo-Saxon 1.00 .212 .265 Institutional 1.00 -.247 Non-institutional 1.00
The base-case scenario (the situation where there is no BoM/BoD/Outsider, Germanic/Anglo-Saxon, Institutional/Non-institutional blockholder) is a firm with no blockholders at all. Hence, each regression shows the change that is occurring relatively to the setting where a company has no blockholders. For example, if the BoD variable as a beta coefficient of -€1000, this means that the CEO of a company with a blockholder in the board of directors earns a salary that is €1000 lower per percentage ownership of the blockholder than his colleague at a firm with no blockholders, ceteris paribus. If this blockholder owns 25% of the shares in the firm, the CEO will earn €25,000 less than the CEO of a firm that is exactly equal, but does not have any blockholders.
21
7. Data
7.1 Data Sources
For my research I look at public companies in the Netherlands. I collect data from three different sources: I use the firms’ annual reports for details on CEO compensation and CEO tenure, as well as information on revenues, profits, and the value of equity. Profits divided by equity is called the Return on Equity (ROE), one of the control variables I use.
The monitor of ‘Het Financieele Dagblad’ I use for information on blockholders. The WMZ-monitor is a database with information of all shareholders with a stake of more than five percent in a public company. According to the Wet Melding Zeggenschap (WMZ), an investor who owns more than five percent of the shares of a public company has to make this information publicly available. This information is collected in the WMZ-monitor. In order to further identify the blockholders I use annual reports and company websites, both of the firms I investigate, as well as reports and website of the blockholders (when present).
TABLE 4 Data sources
Variable Source
Y1 = level of CEO compensation Annual Reports
Y2 = proportion of variable pay to total pay of CEO’s Annual Reports
X1a = BOM blockholder percentage WMZ-monitor / Annual Reports
X2a = BOD blockholder percentage WMZ-monitor / Annual Reports
X3a = Outsider blockholder percentage WMZ-monitor / Annual Reports
X1b = Anglo-Saxon blockholder percentage WMZ-monitor / Company Website
X2b = Germanic blockholder percentage WMZ-monitor / Company Website
X1c = Institutional blockholder percentage WMZ-monitor / Company Website
X2c = Non-institutional blockholder percentage WMZ-monitor / Company Website
X4 = firm size (annual revenues) Annual Reports
X5 = profitability (ROE) Annual Reports
X6 = board size Annual Reports
X7 = board size squared Annual Reports
X8 = leverage Annual Reports
X9 = foundations dummy WMZ-monitor
X10 = CEO tenure Annual Reports
X11 = new CEO Annual Reports
22
denominated on the Dutch stock exchange I made a simple random selection using a simple random sample generator (W.H. Freeman and Company), where all the local firms had the same change ending in the sample I used for this study. The reason I choose 32 firms is because I aimed at a total sample of 100 firms. I collect the data over the years 2006 to 2008, since those are the most recent years of which the annual reports are available. This means that the total sample includes 300 firm years. When a CEO was replaced during 2008, I take the information about the CEO who was in place for the largest part of the year. When the CEO was replaced exactly on the first of July, I take the CEO who was in charge at the end of 2008. The pooled research design is such that a CEO may be counted more than once if he or she remained in office for two years or more. Nevertheless, the sample includes 123 different executives.
7.2 Measuring the components of CEO compensation
Following the literature, I define variable compensation as the compensation components for which the CEO has to achieve certain performance targets, which is the sum of annual bonus, shares awarded, and options awarded in a specific year. For measuring the option value I use a tool from the Dutch Association of Investors (Vereniging van Effectenbezitters), which follows the Black and Scholes (1973) formula, which is as follows:
Value of stock option = P e−ln(1+d )TN(z) – X e−ln(1+r )TN (z- σ
T
) Where z =T
]
2
/
)
1
ln(
)
1
[ln(
)
/
ln(
2σ
σ
T
d
r
X
P
+
+
−
+
+
P stands for the grand-date share price, X is the exercise price, T the time remaining until expiration, d the dividend yield (set at 4%1), σ the stock price volatility (set at 40%2), r the risk/free discount rate (set at 3.7%3), and N() is the cumulative normal distribution function. Although the value of options is quoted in annual reports (legally mandatory since 2002, article 383c BW2), not every company uses the same valuation method. In order to obtain an objective valuation of the options granted, I base the value of the options granted on the Black and Scholes formula through a tool of The Dutch Association of Investors (Vereniging van Effectenbezitters) This tool is accessible through the website of the VEB.
1 Figure used by Vereniging van Effectenbezitters (Dutch interest group for shareholders) to calculate an objective value of
options
2 Figure used by Vereniging van Effectenbezitters (Dutch interest group for shareholders) to calculate an objective value of
options
3
23
An increasing number of companies are using long-term incentive plans (LTIPs). LTIPs become vested if certain performance targets are achieved. Depending on whether the CEO attains his/her targets the number of options granted can be doubled or taken away, or anywhere in between. Following current literature, I assume that the CEO achieves targets that give the right to the initially awarded number of shares or options, i.e. that 100% of the initially awarded options become vested. In reality the amount of options that become vested can vary from 0% to 200% of the initially awarded number (Vereniging van Effectenbezitters), depending on how the CEO performs relatively to his targets. Setting the vested part at 100% is an acceptable and realistic simplification of reality.
When calculating the value of options and shares in a certain year I only take into account the options and shares granted as part of compensation during that specific year, due to data and time constraints. A CEO often owns options that were awarded at several moments in the past, with different expiration dates in the future. Calculating the value of those options is a rather extensive task. Furthermore, as I investigate the influence of blockholders on compensation awarded to a CEO, it is less relevant to include the value of options in previous years than the value of the options awarded in the years I research. The reason for this is that this research focuses on the influence of blockholders on the compensation awards to CEOs. Option holdings from previous years, although relevant from the perspective of the CEO, are not part of the current compensation, and therefore not being influenced by blockholders. I am aware of the fact that since Hall and Liebman (1998) it is more common in the literature to include all the option holdings of CEOs, and not just the ones awarded in the years under investigation. However, in the Hall and Liebman study option holdings of CEOs are included in the dataset to prove that CEO wealth has a strong relationship with firm performance. The relationship that I want to investigate is whether large investors have an influence on CEO compensation. It therefore makes more sense to focus on the compensation that is being awarded in certain years, as this compensation can be influenced by those blockholders.
7.3 Sample Characteristics
24
small cap and local chief executives. The reason why CEOs of local companies receive such a
significant lower amount of options is that option awards become far less common the smaller
the company is. In the AEX option payments were awarded in 42% of the firm years, while
this happened only in 15% of the local firm years.
TABLE 5
Descriptive characteristics on CEO compensation components, 2006-2008 averages
Base Cash Share Option Pension Other Total Index salary bonus awards awards payment compensation remuneration
all € 460,606 € 372,091 € 280,008 € 115,191 € 131,527 € 54,769 € 1,414,010 AEX € 850,052 € 1,020,886 € 885,523 € 300,863 € 278,387 € 146,738 € 3,482,450 AMX € 507,384 € 350,029 € 240,268 € 78,305 € 178,454 € 28,815 € 1,383,254 AScX € 310,118 € 132,790 € 22,987 € 103,819 € 56,827 € 11,333 € 637,873 Local € 233,658 € 58,394 € 23,106 € 9,913 € 36,675 € 32,722 € 394,126 TABLE 6
CEO compensation per year, averages for 100 Dutch firms
year base salary cash bonus share awards option awards pension payment other compen-sation total remune-ration variable compen-sation var/total 2006 € 454,443 € 366,589 € 260,579 € 111,475 € 120,601 € 47,333 € 1,361,019 € 738,643 0.362605 2007 € 459,266 € 445,791 € 316,591 € 123,481 € 123,260 € 78,719 € 1,546,320 € 885,862 0.379996 2008 € 468,110 € 303,893 € 262,855 € 110,618 € 150,720 € 38,495 € 1,334,691 € 677,366 0.313428
25
FIGURE 1
Variable to Total compensation, 300 firm years for 2006-2008
The red column shows the average variable compensation of the CEOs in the dataset: 35.2%
For the average CEO 35.2% of his compensation is variable. As can be seen in figure 1 there
is however quite some difference between CEOs, ranging from zero till 100 percent. What's
more, there is no clear grouping around a certain percentage, meaning that there are
widespread differences between firms in their variable compensation policy.
Table 7 presents descriptive statistics on the control variables. The average size (sales) of the
firm is €7129 million, but as can be seen there is a wide spread, with 1 firm year having sales
of only €0.01 million, while another firm year has sales of €313,248 million. This wide
deviation between firms is visible for nearly every variable. One of the firms in the sample
made a loss of 6 million Euros, while having an equity value of only €0.6 million, leading to a
ROE of -10. Also board size differs significantly, as Arcelor Mittal has a board with 16
people, while Oranjewoud had a year without any supervisory board at all. Another
remarkable dissimilarity between firms concerns their leverage ratio, ranging from nearly
completely equity financed (NV Bever Holding in 2006 with 0.04 leverage), to nearly
completely financed with liabilities (ING Group in 2008 with a leverage ratio of 0.987).
0 5 10 15 20 25 30 35 40 45 50 N umb er o f C EO "s
Percentage Variable/Total compensation
26
TABLE 7
Descriptive statistics for firm characteristics, ownership, and CEO characteristics for the years 2006-2008
Mean Median Minimum Maximum S.D.
Sales (€ millions) 7129 679 0.01 313248 28221 Profitability (ROE) -0.001 0.154 -10 0.841 0.776 Board Size 5.35 4 0 16 2.781 Leverage (liabilities/assets) 0.595 0.591 0.04 0.987 0.187 Number of blockholders 1.32 1 0 5 1.167 % blockholder ownership 29.821 24.545 0 92.39 27.901 Foundations dummy 0.443 0 0 1 0.498 CEO tenure (in years) 5.596 4.2 0.3 37.1 5.355 New CEO dummy 0.13 0 0 1 0.337
The average CEO has held his position for 5.6 years, but also here we can observe differences
between the several indices. AEX CEOs have had their job on average for 4.5 years, AMX
CEOs for 5.5 years, while AScX CEOs have occupied their seat at the head of the board for
6.7 years. CEOs from local firms, finally, have held their position for 5.8 years on average.
The CEO longest occupying his place has been in position for 37.1 years.
7.4 Blockholders
27
made within the company one has to take into the account the voting rights, and not the ownership of capital (hence 50.005% in this case instead of 25.99%).
Certain block holdings are not explicitly included in my data. Those are the blockholders that exist in order to protect the company in case of a hostile takeover and to guarantee the company can exist as a going concern. Examples of those blockholders are special trusts, called Stichting Administratiekantoor, Stichting Preferente Aandelen, and Stichting Beheer van prioriteitsaandelen. The preferential shares owned by these trusts are protection preferential shares, and are used only to guard the company from a takeover. Priority shares are often not publicly issued, but kept by the company and placed in a special trust. The trust funds are usually managed by the board of management, the board of directors, or a combination of those together with outsiders. The owners of priority shares can vote for the appointment of a new member of the board of management or board of directors, and vote in other important decisions the board wants to take. Since these blockholders are created to guarantee the continuity of the company, they are not investing in the company. The shares are created by the company and placed in a trust. As long as the continuity of the company is not threatened, those trusts are usually not active, and the shares are not used to vote. However, as those voting rights may be used, they are important for the company. For example, management could use the voting rights of the trusts to influence their own compensation package, by voting in favor of a pay raise, even when the other blockholders are not supportive. According their regulations (in Dutch: Statuten) the foundations are allowed to vote on compensation issues, even though in reality this is rarely happening. Nevertheless, I still expect the trusts to have an influence, even if this influence is not directly: the presence of trust funds can scare away activist shareholders or takeover parties, thereby indirectly influencing the position and compensation of management. As I expect the presence of trusts to be more important than their actual voting patterns, I decided to include these foundations as a dummy variable.
28
significant relationships less likely. With a 10% cut-off rate 37.5% of companies listed in the AEX have a blockholder. When firms become smaller, chances of a blockholder being present increase: In the AMX 69.6% of the firms has at least one blockholder, in the AScX 90.5%, and among local companies this is 90.6%. 73% of the companies in the sample have at least one blockholder. Of the blockholders with at least 10% of the shares average shareholdings are 22.59% per blockholder, or 29.82% per company (see table 8). The maximum blockholder ownership is 92.00% of the total shares of a company, and the maximum number of blockholders per company is 5. Table 8 also shows that Germanic blockholders control a much larger share of the Dutch stock market than Anglo-Saxon blockholders: Of all the shares outstanding are 24.46% owned by Germanic blockholders, while only 5.36% by Anglo-Saxon blockholders.
TABLE 8
Blockholder descriptives
Blockholder characteristic Minimum Maximum Mean Standard Deviation BoM 0 89% 6.81% 18.99% BoD 0 85% 11.30% 20.66% Outsider 0 90% 13.93% 20.32% Germanic 0 90% 24.46% 26.00% Anglo-Saxon 0 92% 5.36% 14.51% Institutional 0 90% 11.38% 18.16% Non-institutional 0 90% 18.44% 26.15% Any blockholder 0 1 0.73 0.445 Number of blockholders 0 5 1.32 1.167
Total blockholder ownership 0 92% 29.82% 27.90%
Foundation blockholder 0 1 0.44 0.498
Percentages in the table are the percentages of the shares that blockholders own.
TABLE 9
Number of blockholders – type, origin, and institutionalization
Type Origin Institutionalization
Management Board 20 Germanic 107 Institutional 66 Supervisory Board 36 Anglo-Saxon 25 Non-institutional 66 Outsiders 81 Total 137 132 132
29
Table 9 shows the number of blockholders for the 100 firms included in the dataset. The number of institutional and non-institutional blockholders are coincidently exactly equally divided. The origin of blockholders is not, as can be expected. Most blockholders originate in the Netherlands, which has a Germanic corporate governance system. Anglo-Saxon blockholders are scarcer. Table 6 also shows that the different roles of blockholders add up to more than 132. The reason for this is that some blockholders are involved with both the management board and the supervisory board. This dual involvement is apparent in 5 companies out the sample of 100. An example is ASMI, where the family Del Prado controls 15.32% of the voting rights. Chuck del Prado is CEO of ASMI, while his father, Arthur del Prado is honorary chairman of the board of directors. Another example is Alanheri, which is controlled by Recalcico Beheer (60.02% of the voting rights). Mr. Faas is fund manager of Recalcico Beheer and was a member of the supervisory board of Alanheri. After he became unsatisfied with the performance of management he took over the position of CEO himself, temporarily quitting his position as a supervisory board member. Finally, Arcelor Mittal is an example of a company where one person has a seat on both the management board and the supervisory board. Mr. Laskmi Mittal serves both as president of the board of directors, and as CEO of the company in which his family owns 43.6% of the shares.
8. Analysis
Table 10 shows the results of the regressions for the level of compensation. Model 1 includes the Board of Management, Board of Directors, and Outsider blockholders; model 2 includes the Germanic and Anglo-Saxon blockholders; and model 3 includes the institutional and non-institutional blockholders.
30
expected that the presence of foundations leads to higher compensation for the CEO, as they might scare away activist shareholders or companies that try to take over the firm: usually investors that have a negative influence on compensation. Yet, as it appears, compensation levels decrease. A possible explanation for this could be that the foundations provide additional security for the CEO, who has to pay for this security in the form of a lower salary. When the company underperforms changes are smaller that it will be the victim of a hostile takeover, sell out, or restructuring, as the foundation guarantees the continuity of the firm. Consequently, chances that the CEO will be fired due to these actions are smaller as well. A CEO with this kind of insurance might have to ‘pay’ for this and accept a lower salary.
TABLE 10
Results regression for the level of compensation (in Euros per year)
model 1 Sign. 2 Sign. 3 Sign.
R-squared 0.687 0.688 0.686 F-statistic 57.464 0.000 63.801 0.000 63.002 0.000 constant -1,933,533 0.000 -1,734,308 0.000 -1,763,904 0.000 BoM percentage 1,917 0.626 - - - - BoD percentage -5,729 0.073 - - - - Outsider percentage -389 0.913 - - - - Germanic percentage - - -5,287 0.056 - - Anglo-Saxon percentage - - 1,903 0.661 - - Institutional percentage - - - - -3,341 0.381 Non-institutional percentage - - - - -3,766 0.171 Sales millions 32.40 0.000 32.62 0.000 32.45 0.000 ROE percentage 264,699 0.001 259,992 0.002 254,479 0.003
Board size number 828,612 0.000 779,462 0.000 786,942 0.000
Board size squared number -36,670 0.000 -34,744 0.000 -34,729 0.000
Leverage percentage 470,629 0.190 499,706 0.159 469,676 0.187
Foundation dummy -204,651 0.114 -221,741 0.087 -213,601 0.102
CEO tenure years -9,958 0.468 -2,792 0.819 -3,463 0.778
New CEO dummy -465,072 0.016 -431,664 0.024 -457,372 0.017
31
The second significant variable is Germanic blockholder. When this type of blockholder is present, salaries decline with €5,287 per percentage of ownership. Anglo-Saxon blockholders seem to increase compensation, as expected, but with a significance level of 0.661 I cannot accept the hypothesis. The same issue arises with institutional and non-institutional investors, which both turn out to be insignificant.
Table 11 shows the same models, but for the proportion of variable to total compensation. Also here some of the same control variables turn out to be significant: return on equity, board size, board size squared, and the new CEO dummy. Some other control variables are significant in 1 of the models, for example sales, leverage, and CEO tenure. Return on equity is a logical explanatory variable: if a CEO delivers a high return in a certain year, his variable compensation increases in the next year. Board size and board size squared indicate that when the size of the supervisory board increases, initially the CEO receives a higher percentage of variable pay. However, after a certain threshold the proportion of variable pay to total pay decreases again. An explanation for this could be that when the size of boards increase initially, the boards gain power compared to the CEO as together they will have substantial knowledge about the firm, and they can still easily agree on decisions. When the size of the board increases even further, though, it is more difficult to reach consensus and the boards lose power relatively to the CEO. Furthermore, new CEOs receive substantial lower variable compensation than CEOs who hold their position for a longer time. As the CEO tenure also has a negative sign, this implies that CEOs use their power base to lower the variable component of their salary relatively to the fixed component, and even though new CEOs do not have this powerbase, they still receive a lower variable compensation. Hence, I can conclude that overall the variable compensation is decreasing in favor of fixed compensation: in model 1 newly appointed CEOs have an 8.4 percentage point lower variable to total compensation, and every year a CEO is in service another 0.4 percentage point disappears.
32
TABLE 11
Results regression of variable to total compensation
model 1 Sign. 2 Sign. 3 Sign.
R-squared 0.414 0.419 0.410 F-statistic 18.461 0.000 20.874 0.000 20.102 0.000 constant -.101 0.195 -.095 0.209 -.101 0.182 BoM percentage .001 0.260 - - - - BoD percentage .000 0.255 - - - - Outsider percentage .000 0.204 - - - - Germanic percentage - - -.001 0.046 - -
Anglo Saxon percentage - - .001 0.233 - -
Institutional percentage - - - - -.001 0.120
Non-institutional percentage - - - - .000 0.443
Sales millions 8.8E-7 0.101 9.1E-7 0.087 8.8E-7 0.101
ROE percentage .051 0.001 .052 0.001 .046 0.004
Board size number .154 0.000 .146 0.000 .149 0.000
Board size squared number -.008 0.000 -.007 0.000 -.007 0.000
Leverage percentage -.113 0.093 .090 0.172 -.099 0.138
Foundation dummy .036 0.131 .034 0.154 .039 0.107
CEO tenure years -.004 0.082 -.002 0.209 -.002 0.283
New CEO dummy -.084 0.018 -.070 0.049 -.079 0.027
The hypotheses in model 1, however, are not significant, meaning that I cannot say that the involvement of blockholders has an influence on the ratio of variable to total compensation. In the second model, both signs are as anticipated, but only the Germanic influence on variable compensation is significant: the presence of a Germanic blockholder decreases the ratio of variable to total compensation by 0.1 percentage points per percentage of ownership. The third model shows signs for institutional and non-institutional investors that are contrary to what I expected, but both are not significant. Nevertheless, with a p-value of 0.12 the institutional blockholder value is close to being significant. A possible explanation why institutional blockholders have a negative effect on variable compensation could be that I underestimated the ability of institutional investors to monitor management. Institutional investors are usually large parties with numerous employees. This large staff could be able to monitor management, making variable compensation less necessary.
9. Robustness Check
33
compensation do not really change. Two control variables (leverage and foundation dummy) change in one of the models. The only important alteration is the influence of blockholders in the board of directors. Whereas in the full dataset this variable is significant, it is not over the years 2006-2007, meaning that the hypothesis stating that blockholders with a seat on the board of directors have a negative influence on the level of compensation is not supported anymore.
TABLE 12
Level of compensation regressions (in Euros per year), over the years 2006-2007
Model 1 Sign. 2 Sign. 3 Sign.
R-squared 0.688 0.691 0.688 F-statistic 37.715 0.000 42.170 0.000 41.621 0.000 Constant -2,256,860 0.000 -1,938,369 0.000 -1,970,727 0.000 BoM percentage -248 0.960 - - - - BoD percentage -5,461 0.176 - - - - Outsider percentage 729 0.913 - - - - Germanic percentage - - -6,283 0.072 - - Anglo-Saxon percentage - - 1, 344 0.806 - - Institutional percentage - - - - -3,495 0.483 Non-institutional percentage - - - - -4,887 0.159 Sales millions 34.29 0.000 34.52 0.000 34.23 0.000 ROE percentage 393,210 0.023 377,558 0.027 383,650 0.036
Board size number 872,440 0.000 823,213 0.000 828,328 0.000
Board size squared number -38,500 0.000 -36,720 0.000 -36,489 0.000
Leverage percentage 752,116 0.097 688,793 0.126 669,539 0.139
Foundation dummy -146,692 0.377 -172,948 0.295 -173,128 0.301
CEO tenure years -15,128 0.377 -11,345 0.459 -12,345 0.422
New CEO dummy -536,196 0.041 -519,433 0.046 -538,828 0.040
34
TABLE 13
Variable to total compensation regressions, over the years 2006-2007
model 1 Sign. 2 Sign. 3 Sign.
R-squared 0.510 0.541 0.517 F-statistic 17.785 0.000 22.291 0.000 20.191 0.000 constant -.047 0.584 -.009 0.914 -.022 0.792 BoM percentage .000 0.645 - - - - BoD percentage .000 0.159 - - - - Outsider percentage .000 0.261 - - - - Germanic percentage - - -.002 0.001 - - Anglo-Saxon percentage - - .001 0.201 - - Institutional percentage - - - - -.002 0.048 Non-institutional percentage - - - - -.001 066 Sales millions 5.0E-7 0.454 5.7E-7 0.378 4.8E-7 0.463 ROE percentage .087 0.002 .088 0.001 ..77 0.008
Board size number .138 0.000 .132 0.000 .136 0.000
Board size squared number -.006 0.000 -.006 0.000 -.006 0.000
Leverage percentage -..065 0.367 -.070 0.318 -.075 0.296
Foundation dummy .036 0.131 .022 0.387 .026 0.324
CEO tenure years -.008 0.006 -.007 0.006 -.007 0.004
New CEO dummy -.083 0.049 -.068 0.092 -.082 0.048
The most important change in table 13 is the influence of institutional/non-institutional blockholders. In the original model both hypotheses could not be accepted, as the values were insignificant. In the robustness check, alternatively, both are significant. Non-institutional blockholders lower the proportion of variable to total compensation, as stated in hypothesis 7B. Contrary to hypothesis 6B, institutional investors also lower this ratio. As I mentioned before, a possible explanation could be that I underestimated the ability of institutional investors to monitor management, and that they are not in need to use variable compensation as an alternative for monitoring.
10. Conclusion
35
management, board of directors, or are they outsiders), the corporate governance system at the blockholders’ home country (Anglo-Saxon or Germanic), and whether the blockholder is an institutional blockholder or not. I define blockholders as shareholders who control at least 10% of the voting rights. To begin with, block holdings over 5% are publicly available due to the Wet Melding Zeggenschap of the Netherlands. The reason I use a cut-off rate of 10% instead of 5% is that a blockholder who controls 10% of the voting rights has a considerable influence on firm policy, whereas the influence of a blockholder with 5% of the votes is more doubtful.
The data shows that the issue of blockholders is a highly relevant one: in 73% of the companies in the dataset a blockholder is present and over all the firms there are on average 1.32 blockholders who control an average of 29.82% of the shares. Blockholders also have high incentives to monitor agents, as the benefits of monitoring the agents are believed to outweigh costs and blockholders can easier influence the companies' operations because they have more voting rights. Another reason why blockholders are more likely to actively monitor a firm is that their ownership stakes do not permit them to divest the stocks easily.
The sample I took consists of 100 Dutch firms, ranging from large cap (AEX), to midcap (AMX), small cap (AScX) and local firms. For each firm I gathered data over the years 2006-2008, resulting in 300 firm years of data. I take two perspectives regarding compensation, namely the level of total compensation, and the ratio of variable to total compensation. Variable compensation includes cash bonuses, share awards, and option awards.
I analyze the data using an ordinary least squares method, and run separate regressions for each of the blockholder models (one with BoM/BoD/Outsider blockholders, one with Germanic/Anglo-Saxon blockholders, and one with Institutional/Non-institutional blockholders), as well as for the level of compensation and the proportion of variable to total compensation. Several control variables are included to increase the accurateness of the model, such as sales, return on equity, board size, board size squared, leverage ratio, foundation dummy, CEO tenure, and new CEO appointment. All of the control variables were significant in at least one of the models, and many in most, or even all, of the models ran. Especially sales, return on equity, board size, board size squared, and new CEO appointment turned out have significant influences, proving the validation of using these variables to determine executive compensation.
