1 Amsterdam Business School
Shareholders influence on CEO payment
through the non-binding vote at the shareholder meeting
Name: Joram Wielinga
Student number: 10439951 Master thesis
August 2, 2014
MSc Accountancy & Control, specialization Control
Faculty of Economics and Business, University of Amsterdam First supervisor: Prof. dr. L.R.T. van der Goot
2 Table of contents
1. Abstract ... 3
2. Introduction: Proposals submitted by shareholders ... 4
Payment Policy and Shareholder Characteristics ... 5
3. Background ... 6
Influence of institutional shareholders upon corporate governance ... 6
Influence of institutional shareholders upon CEO payment... 8
Influence on corporate governance policies through shareholder meetings ... 10
What does this paper add to the existing literature ... 11
4. Research question:... 12
Hypotheses ... 12
5. Data and Methods ... 14
The RM proposal database: variables to be used ... 14
6. Results: Descriptive statistics ... 19
Correlation between the variables ... 20
Testing the hypotheses ... 23
Regression analyses ... 31 Conclusions ... 36 7. Discussion ... 36 8. Appendix A ... 38 9. Appendix B ... 39 10.Literature ... 42
3 Abstract
This paper seeks to explore the use of the non-binding vote upon influencing the CEO payment by shareholders. Shareholders of different types are called upon by society to use their influence to curtail excessive CEO payment. Under rule 14a-8 of the Security and Exchange Commission shareholders of a publicly traded company in the United States of America are permitted to submit a proposal to put on the vote at a shareholder meeting and hence put pressure upon a company. This paper explores whether some shareholder groups are more inclined to use this instrument of the non-binding vote to influence CEO payment than other groups. As shareholders have different backgrounds being individuals or an institutional fund, one might expect that the background of a proposer does have an influence upon whether a proposal gains a majority or not. This paper finds that these assumptions do not hold for the non-binding vote. There is no significant difference between the amount of proposals related to CEO payment among the different categories of shareholders, there is no significant difference between the outcome of the votes for the different proposers, and CEO payment related proposals do not gain significant more support. This paper concludes that there is no significant effect on the acceptance of the non–binding vote to influence CEO payment, regardless of whom of the different shareholder groups is bringing the resolution to the vote. The effect of what the influence is from the occurrence of the non-binding vote is out of scope.
4 Introduction: Proposals submitted by shareholders
CEO payment has been a returning element on shareholder meetings. Especially institutional shareholders, like investment funds and pension funds, are called upon through society to monitor excessive CEO payments and to curtail these.
Institutional shareholders are also called upon by political and social movements. In the Netherlands for example, one of the labor unions (FNV) has issued a pamphlet to call upon the institutional shareholders1, and even political organizations ask institutional investors “to take responsibility”2. Shareholders seem to see this role as their responsibility as well. Examples are mentioned in the daily newspapers regularly3.
Shareholder can take this responsibility very easily. Under rule 14a-8 of the Security and Exchange Commission shareholders of a publicly traded company in the United States of America are permitted to submit a proposal to put on the vote at a shareholder meeting. Proposals submitted under this rule are non-binding to the management of the entity, but they do form a possible way of influencing an entity’s governance structure to provide shareholders with a means to inform corporate boards about issues that are important to them. The rule permits qualified shareholders, who own 1% of a company’s securities or shares with a market value greater than $2000, to propose a shareholders' vote on any topic within shareholders' purview concerning the corporation with limited exceptions (Thomas and Cotter, 2007). The Dodd-Frank Act (2010) has strengthened the focus on this non-binding vote by requiring a non-binding vote on executive compensation at publicly traded companies at least every 3 years (Levit and Malenko (2011); Dodd Frank Wall Street Reform and Consumer Protection Act, Subtitle E, Section 9514).
However, evidence on the effectiveness of these proposals in relation to CEO compensation is limited. Armstrong, Gow and Larcker (2013) found no real relationship between lower voting support or rejection of compensation plans and future lower compensation plans. They concluded that regulatory efforts to strengthen the shareholder voting right may have
1 As published on http://www.fnvbondgenoten.nl/site/dossiers/loon/downloadblokken/546234, March 2014.
2As published on
http://www.elsevier.nl/Politiek/nieuws/2013/6/SP-leider-Roemer-stijging-topinkomens-is-beschamend-1285292W/ , (March 2014).
3 For example:
Wilco Dekker (2014); Aanpak bonussen Heineken wekt woede beleggers; In: daily Dutch newspaper De Volkskrant; 25 april 2014, p.20.
Gerben van der Marel (2014); Pensioenreuzen strijden tegen topbeloningen; In: daily Dutch newspaper Het Financieel Dagblad, 22 may 2014, p. 16.
4 As published on http://www.gpo.gov/fdsys/pkg/PLAW-111publ203/html/PLAW-111publ203.htm (march
5 limited effect on firms’ compensation policies. Levit and Malenko (2011) also concluded that in general nonbinding votes fails to transfer shareholder upon management actions when manager and shareholder interest are not aligned. However, they do conclude that the presence of an activist shareholder can in fact enhance the advisory role of non-binding voting in case of a conflict of interest between management and shareholders.
Whether proposals related to CEO compensation have an influence upon the level of CEO compensation remains therefore unclear. The instrument is more often used and voters make use of it in a different way as a result of developments in change in the governance of companies (Morgan, Poulsen and Wolf, 2006), but the real live effects of these proposals upon governance policies remains disputable.
Payment Policy and Shareholder Characteristics
Institutional shareholders have different ways to influence payment policies of the companies they own. Studies have shown that there is a link between the size of the shareholder interest and the level of influence of payment policies in which a bigger institutional shareholder means less excessive CEO compensation (David, Kochhar and Levitas, 1998; Hartzell and Starks, 2003; Khan, Dharwad, Brandes, 2005). The same authors argue that it is actually in the interest of large institutional shareholders to keep a close look on the compensation policies of companies.
Institutional shareholders differ from ordinary shareholders in the way they are organized and can pull leverage on a company. Depending on the number of shares, institutional shareholders can exercise a level of control which includes appointing members of the management team. They can set agenda’s for shareholder meetings and can hold boards accountable. They can change management teams by forcing CEO’s to leave (Parrino, Sias and Stark, 2003).
But these influences are only for those who are big enough, and for those who take enough interest. The non-binding vote is an instrument which does not oblige management to pursue the actions proposed, but they can influence decision makers. They provide a social pressure tool against unwilling management (Renneboog and Szilagyi, 2011)
In this paper the focus will be on the influence of institutional shareholders through non-binding proposals at shareholder meetings. This paper will examine the proposed resolutions to see if they have been successful and compare these to the proposals made by individuals. Due to the availability of data, this paper will solely focus upon companies in the United States of America.
6 The paper has the following chapters: In the chapter Background, existing research about non-binding voting on shareholder meeting and their effect upon governance is explored. The chapter Research question and hypotheses will present the research question and the hypotheses and plot them to the literature. The chapter Research methodology will tell which data is used and how the research is done. The findings will be discussed in the chapter Results and the paper will and with a discussion about the research as it is conducted for this paper.
Background
Ownership of companies has been changing over time. Since the 1980s the ownership structure is shifting from individual (managerial) ownership to more institutionalized ownership. More and more companies see a big part of their shares bought by institutional owners such as pension funds and (private) investment funds (Davis and Stout, 1992). These institutional owners tend to have a different view upon the internal corporate control mechanism compared with the managerial owners, which lead to a more fixed control upon (among others) CEO payment (Walsh and Seward, 1990).
This chapter will investigate existing literature in order to present an overview of current research findings on the influence of shareholders on CEO payment. This chapter will start with presenting some background literature about the influence of institutional shareholders upon corporate governance in general, and then focus upon literature about the influence upon CEO payment specifically. The chapter will end with exploring what has been researched upon the tool of the non-binding vote as mend under rule 14a-8 of the Security and Exchange Commission in relation to CEO compensation, and how this research will add to this specific research field.
Influence of institutional shareholders upon corporate governance
As stated by Davis and Stout (1992), institutional shareholder ship has been growing over the years. Several types of ownership can be distinguished. Ownership can be from inside the company (a director-owner, or inside shareholder) or from outside (an institutional owner, an outside shareholder). Research show that this type of ownership influences the corporate governance policies of the specific corporation. Hambrick and Finkelstein (1995), for example, made this distinction in ownership and examined the annual percentage change in CEO pay and related them to company size. They split their firms into two ownership configurations: the externally controlled firm en the management controlled firm, which has no single major shareholder. They found that in externally controlled firms the annual percentage was lower than in management controlled firms. Their explanation was that in externally controlled firms, owners
7
have a “strict utilitarian” view upon CEO compensation, where in management-controlled firms CEO’s seek legitimization for their pay (Hambrick and Finkelstein, 1995). They added to this that in an internally
controlled firms boundaries for CEO compensation were much less clear.
Institutional ownership tend to lead corporate governance into a specific direction: Hartzell and Starks (2003) found that institutional ownership concentration is positively related to the pay-for-performance sensitivity of executive compensation and negatively related to the level of compensation, even after controlling for firm size, industry, investment opportunities, and performance. Their results suggest that the institutions serve a monitoring role in mitigating the agency problem between shareholders and managers. Additionally, they find that clientele effects exist among institutions for firms with certain compensation structures, suggesting that institutions also influence compensation structures through their preferences. These finding matches with Jensen and Meckling (1976) who found this relation also in their research on the ownership structure of the corporation, including how equity ownership by managers aligns managers interests with those of owners. Conelly, Tihanyi , Certo and Hitt (2010) showed using the agency theory that the more dedicated an institutional shareholder, the more influence they exert over strategic and tactical competitive actions of the company.
Next to the alignment of interest of management decisions to the institutional shareholder there is also the alignment of internal controls towards the interest of the institutional shareholder. According to Gillan and Starks (2003), stakeholders have an important influence upon the internal control mechanisms of the corporate governance within companies (Gillan and Starks, 2003). These corporate governance policies can change when there is a new (partial) owner which has been shown in research investigating the effects of a takeover5 from an institutional shareholder upon corporate governance policies such as CEO payment. Takeovers take place when there is an opportunity caused by a slack in internal controls. This opportunity means that there is a possible benefit for those willing to take the actions needed to repair or intensify the internal controls. An outside shareholder can obtain this revenue by changing existing management as this management is accountable for the slack in the internal controls (Hadlock and Lumer, 1997). One can argue this is a great example of the influence institutional shareholders can have on the corporate governance of an organization.
5 A takeover is the case when an entity acquires the majority of shares that gains influence in the company. This
influence is already possible when small amounts of shares have been obtained. In terms of international accounting standards an entity has significant influence when the entity has control over 20% of the outstanding shares or more5. However, shareholders can also set the agenda during shareholder meetings. In that case an entity does not
8 However, Huson, Parrino, and Stark (2001) found indications that changes in the intensity of the takeover market are not associated with changes in the sensitivity of CEO turnover to firm performance. The theoretical implication of this is that a takeover does not necessarily have implications upon the position of a CEO. Huson , Parrino and Starks (2001) findings are in fact contrary to the theory that a more active takeover market strengthens internal control mechanisms.
Parrino, Sias and Starks (2003) studied whether institutional investors actually change their interest in a company when dissatisfied with a firm management. They found that the aggregate institutional ownership and the number of institutional investors decline in the year prior to forced CEO turnover. Further, Parrino, Sias and Starks (2003) found that when the amount of institutional ownership becomes smaller (hence: there are more smaller investors), the likelihood of a forced CEO turnover is bigger, and the change that an outsider is appointed as CEO is bigger as well.
The balance between investors who are able to resist management choice and those who do not proofs to be an important element in predicting the influence of stockholders upon CEO compensation. It is basically about the level of internal and external control. Internal control refers to organizationally based mechanisms of corporate control. External, market-based control mechanisms refers to influence from outside. Both can help to align the diverse interests of managers and shareholders (Walsh and Seward, 1990).
Influence of institutional shareholders upon CEO payment
Now how does the influence of institutional shareholders affect CEO compensation? CEO compensation has been found to be strongly related to performance but research has shown that compensation levels are also related to other elements ranging from stakeholders influence, ownership, composition and size of the board to the CEO’s social background. It is difficult to say which relation is strongest and under which condition these relations prevail. Barkema and Gomez-Mejia (1998) made an extensive overview of the literature in order to construct a general framework for research on executive pay. They argue that compensation packages are determined by a much broader range of variables then just performance. This observation is shared in other research: Some studies focus on those who appoint the new CEO and connect their research to the composition of the board (Daily., Johnson, Ellstrand , & Dalton, 1998; Conyon and Peck, 1998); other relate determinants of compensation packages to performance driven variables (Ciscell and Caroll, 1980). Hence, CEO compensation is driven by more factors than just performance.
9 Does that mean that there is no evidence of a relation between CEO payment and the existence of institutional shareholders? Managers and institutional shareholders are both perceived as self-serving entities who try to maximize their own interest in terms of power and revenue. In reaction to this clash of interests organizations have developed formal mechanisms in monitoring and reward structures. These are perceived to be necessary to align the incentives of top managers with the interests of shareholders (Jensen and Meckling, 1976; Barkema and Gomez-Maija, 1998). Another example is given by David, Kochhar, and Levitas (1998), who have examined institutional investors behavior using agency theory. They found that the presence of institutional block holders was associated with lower levels of CEO compensation and higher proportions of long-term incentives in total compensation (David, Kochhar, and Levitas, 1998). Other research indicated that this relationship was only found for companies with investors who did not depend for their business on the firm they invested in, such as banks and insurance companies. Brickley et al. (1988) suggested that CEOs exercised power over these investors because they could withhold business from them. He argued that pressure-resistant institutions are more likely to oppose CEOs than pressure-sensitive institutions when, for instance, voting on anti-takeover amendments (Brickley et al., 1988). All this shows that a clear relationship is difficult to show, and most likely not to be an excluding relationship.
However, there have been signs of some relationship and the direction of this relationship. Khan, Dharwadkar and Brandes (2005) did a limited research at how the size of institutional ownership affect executive compensation. Using the agency theory approach, they investigated how institutional ownership concentration and dispersion affect levels of CEO compensation, pay mix, and stock option pay sensitivity. They found that the largest owner's concentration is associated with lower levels of compensation, as well as with higher ratios of salary to total compensation and lower ratios of options to total compensation. However, the number of blockholders does not predict any aspects of CEO compensation. They showed that when institutional ownership is high within a company, it is more likely that the compensation packages are behavior-based as a percentage of the total compensation. They found ownership concentration to be negatively related to compensation level. According to Khan et all. institutional ownership is also about size: the more an individual institutional owner owns, the more influence (Khan, Dharwadkar and Brandes, 2005).
To add to all this, all this research does not take into account the personal ties that are sometimes important for the formation of CEO compensation policies. Butler and Gurun (2012) for example found that having the same educational backgrounds matter for the attitude from
10 institutional shareholders towards the CEO compensation. According to their research, the “old-school” network is a not-to-be neglected actor in the scrutiny of CEO compensation policies.
Influence on corporate governance policies through shareholder meetings
So what does this mean for the influence of institutional shareholders? Next to influence through ownership, acquiring ownership or being acquainted with each other? Since 1942, when the Security and Exchange Commission adopted rule 14a-8to and made it possible for shareholders to make any proposal towards the general shareholder meetings. Since then, shareholders have a non-binding but nevertheless direct way to bring matters forward to the board of a company (Pound, 1991). These proposals are a pretty clean and straightforward tool to show the management what shareholders want. As it does not have to linked to the current governance policies but also can be about future decisions, they can help with shaping the corporate governance policies in general. As they are non-binding, they function as a sort of advisory tool from shareholder towards management and as monitoring tool for management to inform them about the attitudes and opinions of their shareholders.
According to Thomas and Cotter (2007) this instrument has had an impact on corporate governance becoming more visible in the first decade of the 21th century. They found some big changes from earlier periods: more proposals receiving majority shareholder support in their sample from 2002-2004 relative to earlier studies, and this support has translated into directors implementing more of the actions called for by shareholders.
This influence however is conditional: Gillan and Starks (2000) found that proposals done in shareholder meetings by institutional owners gain substantially more support than proposals sponsored by individuals. And according to Levit and Malenko (2011), non-binding voting generally fails to convey shareholder views in a situation of nonalignment of interest. Next to the alignment of interests, it is found that in general shareholder pressure (e.g., the voting outcome and the influence of the proponent) and the type of proposals are the main determinants of the implementation decision for management (Ertimur, Ferri and Stubben, 2010). But the chance on reputational damage has also an impact, Ertimur Ferri and Stubben (2010) and Thomas and Cotter (2007) found that proposals that win a majority vote are more likely to be implemented.
Renneboog and Szilagyi (2011) found that shareholder proposals tend to be targeted at firms that both underperform and have generally poor governance structures. Proposal submissions are more likely to be made against firms that use antitakeover provisions to entrench management, have ineffective boards, and or CEOs who do not have an incentive to maximize
11 shareholder value, regardless of the proposal objectives. They also found that sponsor type is not a factor for success. They conclude that there is no systematic agenda-seeking by certain activist groups such as union pension funds.
What does this paper add to the existing literature
It is the last finding of Renneboog and Szilagyi (2011) that this paper seeks to explore more thoroughly. As stated in the beginning with some recent newspaper articles, society is asking for a more ethical attitude of big shareholders towards compensation policies. This paper will explore the question whether this is happening or not. To do so, proposals will be examined and classified, just as their proposers. It will be examined whether a relationship is visible between the type of proposal and the type of proposer. Special interest will be taken into those proposals that are related towards CEO compensation.
This thesis will add information to the research about the role of institutional shareholders in ethical issues through the use of the instrument of the non-binding vote. The outcomes are relevant scientifically and socially, as they add new information to the discussion within society about the roles and instruments of influence of institutional owners towards curtailing excessive CEO payments.
12 Research question:
This paper focuses upon the influence that institutional shareholders have through proposals at shareholder meetings. The main research question is whether institutional shareholders are more likely to succeed in influencing CEO compensation through proposals at a shareholder meeting in comparison to individual shareholders. To do so this paper will look at the relation between the likelihood that a proposal is accepted and the background of the proposer being an individual or an institution.
To investigate this the following must be researched:
- What is the amount of proposals made by individuals and institutions and how much are accepted with a yes-vote at the shareholder meeting?
- When looked more closely upon the proposals, what amount of proposals of the different proposers is accepted and is aimed to influence CEO compensation?
Hypotheses
Based upon the literature available the following hypotheses are formulated:
H1: Institutional owners are more likely to submit a proposal aimed to influence CEO compensation at a shareholder meeting than individual owners.
Literature showed that institutional owners have an incentive to align internal controls (Conelly, Tihanyi, Certo and Hitt , 2010). Hence it is likely that institutional owners are more often inclined to use the instrument of a non-binding vote. On the other hand one can argue that a dedicated institutional owner is more likely to try exert their interest in a company through internal controlling mechanisms (Gillan and Starks, 2003). A non-binding vote is not an internal control, but it comes from outside and is accessible to anyone with 1 % of the outstanding shares or more than $2000 worth in outstanding shares. Testing this hypothesis will therefore help in understanding who is using this instrument for the cause of influencing CEO compensation. To research this hypothesis the type of proposer (institutional/individual) will be determined in the data and the proposal will be determined as relevant to CEO compensation. Next the relation between the two will be examined.
13
H2: A proposal submitted by an institutional owner is more likely to be accepted than a proposal submitted by an individual.
This hypothesis will test if and to which extent the success of a proposal is related to the background of the proposer. An answer will be given by researching the likelihood of acceptance for the different proposers. Herewith this research will also validate the results of Renneboog and Szilagyi (2011) who found that sponsor type is not a factor for the success of a proposal.
H3: Proposals which refer to CEO compensation are more likely to obtain a majority vote than proposals with another subject.
By determining the proposals that refer to CEO compensation and relate them to the success of proposals about other subjects, it is possible to make a statement about the use of the instrument of the non-binding vote in relation to the discussion about efforts of influencing the CEO compensation.
14 Data and Methods
The possibility to test the hypotheses is limited by the availability of data. The University of Amsterdam gives its students access to a range of different databases. From the available databases the following is useful to test the hypotheses:
- RiskMetrics – Shareholder proposals (further referred to as RM proposal database) This data set is available from the Wharton Research Data Services (WRDS) – website (http://wrds-web.wharton.upenn.edu/wrds/), after registering on the website with an active email address and approval from the WRDS contacts at the University of Amsterdam.
The database contains information about American companies. The variables needed to test the hypotheses are to be related to the same company and the same year. The years on which this paper will conduct its research are largely the result of the availability of data and changes to the dataset overtime and the limited time to process this data. At the time of writing this paper, RM proposal database has data available from 1/1/1997 up to 12/31/2012. The time period researched in this paper will be 1/1/2007 up to 12/31/2012. The database sample used to test the hypotheses differ from each other as the hypotheses narrow down from all proposed resolutions to resolutions related to CEO payment that were brought to the vote and cast a majority vote.
The RM proposal database: variables to be used
The RM proposal database consists of proposals that came to a vote as well as those that did not (e.g., because they were withdrawn by the proponent or allowed to be omitted from the proxy by the SEC). Variables include the lead filer of the proposal, the meeting date, and outcome; some fields are not complete every year for every company, but the vast majority of meeting dates are present6.
The classification used in the RM proposal database has to be moderated and examined to some extent. Not all presented variables are reliable; for example the RM proposal database has labelled the proposals under the variable Resolution Type as SRI (social responsibility issues; proposals related to the relation of the company towards society) and GOV (Governance related; proposals related to the governance within the company). These labels are a “loose qualification” according to the WRDS-website7 and therefore not unambiguous as some of the proposals could be labelled under both labels. Another uncertain variable is the variable Passed which says whether
6 Taken from the introduction at the WRDS website as available at http://wrds-web.wharton.upenn.edu/wrds/ds/riskmetrics/proposals/index/index.cfm (june 2014).
15 the proposal is accepted or not. Of the 5904 proposals in the database 482 are marked as passed but for the years 2007 and 2010 there is no data available at all, although in these years proposals did get 50% or more of the vote. This classification is therefore ignored.
The RM proposal database also does not have a consistent way of labelling sponsors of shareholder proposals. The variable Sponsor Type consists of 17 different labels which overlap each other more or less.
To take into account these inconsistencies, the data from the RM proposal database is restructured by constructing some new variables and some dummy variables which are presented here.
Classification of the ownership
For the purpose of this study the classification as used by Renneboog and Sziliagyi (2011) will be used. This means that all sponsor types will be classified as either union pension funds; public pension
funds; investment funds; social responsible/religious (SRI) or as individuals. Renneboog and Sziliagyi (2011)
also distinguish “coordinated investors” as a sponsor type, but as the RM proposal database does not offer this distinction, and therefore this classification is not used. The RM proposal database does have a classification of others or empty. For this paper these will be renamed as other/unknown. In appendix A an overview can be found of the different names given in the RM proposal database and the classification used for the purpose of this paper and in table 1 an overview is given of the amount of proposals done during the year, classified by type of proposer.
Overview of types of sponsor
Proposer 2007 2008 2009 2010 2011 2012 Total Individual 364 341 322 328 236 127 1718 Other 67 109 90 115 73 233 687 Public Pension fund 117 122 152 123 93 157 764 SRI 237 297 255 259 187 127 1362 Union 375 278 298 185 166 71 1373 Total 1160 1147 1117 1010 755 715 5904
Table 1: An overview of the number of proposed resolutions done by the different sponsors as diverted from the RM proposal database.
Classification of whether the proposal is related to CEO payment or not
To classify the resolutions made by the proposers, all 5904 resolutions had to be classified as related to CEO compensation or not. To qualify them as being related to compensation policies
16 the following key words were used to classify the proposals: CEO, pay, compensation, executive and
performance. Each word has been searched for separately and each proposal which contained them
qualified as being a proposal which in some way intend to influence compensation policies within the corporation the resolution as aimed at. Appendix B gives an overview of all the resolutions which contained these keywords as stated in the RM proposal database. A new variable stating whether the proposal is related to CEO payment was added to the database with two values: Yes (1) or No (2). The results over the years are presented in table 2.
CEO pay related proposals per sponsor type per year
2007 2008 2009 2010 2011 2012 Total
Individual sponsor 364 341 322 328 236 127 1718
CEO pay related 95 63 54 43 18 3 276 Not CEO pay related 269 278 268 285 218 124 1442
Pension fund 117 122 152 123 93 157 764
CEO pay related 15 17 30 15 2 2 81
Not CEO pay related 102 105 122 108 91 155 683
Union fund 375 278 298 185 166 71 1373
CEO pay related 110 97 76 47 23 6 359 Not CEO pay related 265 181 222 138 143 65 1014
SRI 237 297 255 259 187 127 1362
CEO pay related 20 26 48 52 1 - 147
Not CEO pay related 217 271 207 207 186 127 1215
Other 67 109 90 115 73 233 687
CEO pay related 12 16 16 15 8 13 80
Not CEO pay related 55 93 74 100 65 220 607
Total 1160 1147 1117 1010 755 715 5904
CEO pay related 252 219 224 172 52 24 943 Not CEO pay related 908 928 893 838 703 691 4961
Table 2: Overview of all proposals in the sample 2007-2012, qualified as related to CEO payment. For an overview of the proposals themselves, see appendix B.
Overview of the variables used
To test the hypotheses the following variables are needed: Independent variable:
Year – Fiscal year
CUSIP – alphanumeric code which identifies a North American financial security for the purposes of facilitating clearing and settlement of trades
17 Construct for independent variable
Sponsor Type 2 – Construct made on the basis of sponsor types as presented in the RM proposal database. See for detail appendix A.
Dummy variable Individual Sponsor – individual sponsor (as from sponsor type 2) has value 1, all other 0.
Dummy variable Public Sponsor - Public sponsor (as from sponsor type 2) has value 1, all other 0.
Dummy variable Union Sponsor – Union sponsor (as from sponsor type 2) has value 1, all other 0.
Dummy variable SRI Sponsor – SRI sponsor (as from sponsor type 2) has value 1, all other 0.
Dummy variable Other Sponsor – Other sponsor (as from sponsor type 2) has value 1, all other 0.
Dummy variable year 2007 – dummy used to identify the fiscal year as presented in the RM proposal database. Year 2007 has value 1; all other years have value 0.
Dummy variable year 2008– dummy used to identify the fiscal year as presented in the RM proposal database. Year 2008 has value 1; all other years have value 0.
Dummy variable year 2009– dummy used to identify the fiscal year as presented in the RM proposal database. Year 2009 has value 1; all other years have value 0.
Dummy variable year 2010– dummy used to identify the fiscal year as presented in the RM proposal database. Year 2010 has value 1; all other years have value 0.
Dummy variable year 2011– dummy used to identify the fiscal year as presented in the RM proposal database. Year 2011 has value 1; all other years have value 0.
Dummy variable year 2012– dummy used to identify the fiscal year as presented in the RM proposal database. Year 2012 has value 1; all other years have value 0.
Dependent variable
Votes For - % of votes cast that is in favor of the proposal; presented as a percentage in the RM proposal database.
18 Construct for dependent variable
CEO pay related proposal yes or no – Proposal is related (value 1) or not related (value 0) to CEO payment as one or more of the following key words were used in the proposal:
CEO, pay, compensation, executive and performance as stated in the variable resolution in the RM
proposal database. For a complete overview of the proposals qualified as CEO pay related see appendix B.
To test the hypotheses the data will be analyzed and tested. First the correlation between the variables will be tested. Then the relationship between sponsor type and proposal will be analyzed through a t test. The dependence of the likelihood of acceptance on the sponsor type will also be tested through a t-test, as well as the likelihood of the acceptance of proposals which are related to CEO compensation will be tested through a t test.
A regression analysis will be conducted on three models in which the influence of the different variables is measured at the percentage of the votes in favor.
19 Results: Descriptive statistics
The data selection between 2007 up to 2012 from the RM proposal database presents 5904 observations in which a proposal is given. These observations are spread out over different amount of companies each year (2007: 1140 companies and 1160 proposals; 2008: 916 companies and 1147 proposals; 2009: 939 companies and 1117 proposals; 2010: 1001 companies and 1010 proposals; 2011: 755 companies and proposals; 2012: 715 companies and proposals). On average there is 1,07 proposal per company per year in the data selection with a declining trend in the amount of proposals over the years in this sample.
Descriptive Statistics
N Minimum Maximum Mean Std. Deviation
Data Year 5904 2007 2012 2009,20 1,65
Vote For Percentage 3329 0 99,10 34,33 23,30
CEO pay related resolution yes
(1) or no (0) 5904 0 1 0,16 0,37 Sponsor type 2 5904 1 5 2,75 1,39 Vote 50 % 5904 0 1 0,12 0,33 Dummy_Individual 5904 0 1 0,29 0,45 Dummy institutions (2,3,4) 5904 0 1 0,59 0,49 Dummy_Pension 5904 0 1 0,13 0,34 Dummy_Union 5904 0 1 0,23 0,42 Dummy_SRI 5904 0 1 0,23 0,42 Dummy_Other 5904 0 1 0,12 0,32 Dummy year 2007 5904 0 1 0,20 0,40 Dummy year 2008 5904 0 1 0,19 0,40 Dummy year 2009 5904 0 1 0,19 0,40 Dummy year 2010 5904 0 1 0,17 0,38 Dummy year 2011 5904 0 1 0,13 0,33 Dummy year 2012 5904 0 1 0,12 0,33 Valid N (listwise) 3329
Table 3. Descriptive of variables used. CEO pay related resolution yes (1) or no (0) has a value 1 for all proposals that are related to CEO payment and a value 0 for all other resolutions (see appendix A for a total overview) . Vote For Percentage contains the vote in favor of the proposal as a percentage of the total votes cast. Sponsor Type 2 identifies the sponsor type as presented in appendix B. Vote 50% is a dummy variable with values 0 for proposals that do not have a Vote For percentage equal to or above 50%, and those with 50% or more value 1. Dummy Individual value 1 means Proposer is an Individual, value 0 means Other than individual or institutional. Dummy institutions (2,3,4) has for every Pension, Union or SRI sponsor value one, all other have value 0. Dummy Pension value 1 means Proposer is an Pension fund, value 0 means all Other proposers. Dummy Union value 1 means
Proposer is an Union fund, value 0 means all Other proposers. Dummy SRI value 1 means Proposer is an Social, Religious Interest fund, value 0 means all Other proposers. Dummy Other value 1 means Proposer is an Other, unspecified kind of institutional fund, value 0 means defined sponsor type. The dummies for the years are dummies for all the values given in Data
Year (2007 up to 2012).
Most proposed resolutions do not relate to CEO payment as can be derived from the mean of the CEO pay related resolution variable. With a value of 0 for non-related variables and a value of 1
20 for CEO payment related variables, the mean of 0,16 as presented in table 4 indicates that the value of this variables is closer to 0 than to 1 in this population. Non-binding votes in general do not acquire a majority of the votes in this population as is shown by the description of the variable Vote For Percentage. Of the 5904 proposals made, 3329 (56% of the total proposals) are actually brought to the vote and these had a mean of 34% in favor of the resolution. In this population in general the non-binding proposals do not acquire a vote in favor of 50% or more. None of the proposals got a full 100% of the vote. Variables Vote 50% and Vote CEO proposal
50% are constructs of the Vote For Percentage. Vote 50% has a value 0 if the Vote in favor of the
proposal is less than 50% and 1 if equal or above 50% in favor. Of these with value 1 all those which are related to CEO payment have a value 1 for variable Vote CEO proposal 50% and all other have a value of 0. The means show that the amount of proposals with a vote in favor equal or above 50% is about 12% of the total amount of proposals and of those related to CEO proposals and a majority of the votes cast is about 1 %.
For the dummies of the group of sponsors (Individual, Pension, Union, SRI or Other), the mean represents the total percentage of the amount of proposals made by the group as defined by the dummy compared to the total amount of proposals made. So 29% of the proposals made were entered by individual shareholders when compared to the different institutional groups; 13% by Pension funds; 23% by Unions; 23 by SRI funds; 12% by undefined funds.
Correlation between the variables
In SPSS the correlation between the variables is tested. The correlation gives an idea of the relationship between the variables and indicates a possible influence amongst each other. Correlation is weak when close to 0 and strongest in situations where the statistical correlation coefficient is close to 1 or -1. For the variables used this coefficient is presented in the Pearson Correlation coefficient in table 4.
21 Table 4 Correlations Du mm y_ In div id ua l Du mm y_ Pe ns io n Du mm y_ Un io n Du mm y_ SRI Du mm y_ O the r Du mm y ye ar 20 07 Du mm y ye ar 20 08 Du mm y ye ar 20 09 Du mm y ye ar 20 10 Du mm y ye ar 20 11 Du mm y ye ar 20 12 C EO p ay re la ted re so lutio n ye s (1) or no (0) V ote 50 % V ote Fo r Pe rc en ta ge
Dummy_Individual Pearson Correlation 1 -0,247** -0,353** -0,351** -0,232** 0,025 0,007 -0,003 0,034** 0,018 -0,093** 0,002 0,130** 0,119**
Sig. (2-tailed) 0,000 0,000 0,000 0,000 0,056 0,600 0,824 0,009 0,162 0,000 0,901 0,000 0,000
N 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 3329
Dummy_Pension Pearson Correlation 1 -0,212** -0,211** -0,140** -0,042** -0,034** 0,010 -0,010 -0,007 0,100** -0,057** 0,024 0,090**
Sig. (2-tailed) 0,000 0,000 0,000 0,001 0,010 0,461 0,428 0,585 0,000 0,000 0,069 0,000
N 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 3329
Dummy_Union Pearson Correlation 1 -0,301** -0,200** 0,106** 0,011 0,039** -0,053** -0,011 -0,117** 0,153** -0,038** 0,063**
Sig. (2-tailed) 0,000 0,000 0,000 0,381 0,003 0,000 0,377 0,000 0,000 0,003 0,000
N 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 3329
Dummy_SRI Pearson Correlation 1 -0,199** -0,031* 0,033* -0,003 0,028* 0,015 -0,047** -0,077** -0,179** -0,348**
Sig. (2-tailed) 0,000 0,017 0,011 0,833 0,033 0,235 0,000 0,000 0,000 0,000
N 5904 5904 5904 5904 5904 5904 5904 5904 5904 5904 3329
Dummy_Other Pearson Correlation 1 -0,090** -0,033* -0,054** -0,004 -0,023 0,243** -0,043** 0,077** 0,095**
Sig. (2-tailed) 0,000 0,012 0,000 0,786 0,071 0,000 0,001 0,000 0,000
N 5904 5904 5904 5904 5904 5904 5904 5904 5904 3329
Dummy year 2007 Pearson Correlation 1 -0,243** -0,239** -0,225** -0,189** -0,184** 0,078** -0,026* -0,068**
Sig. (2-tailed) 0,000 0,000 0,000 0,000 0,000 0,000 0,048 0,000
N 5904 5904 5904 5904 5904 5904 5904 5904 3329
Dummy year 2008 Pearson Correlation 1 -0,237** -0,223** -0,188** -0,182** 0,042** -0,028* -0,038*
Sig. (2-tailed) 0,000 0,000 0,000 0,000 0,001 0,034 0,029
N 5904 5904 5904 5904 5904 5904 5904 3329
Dummy year 2009 Pearson Correlation 1 -0,219** -0,185** -0,179** 0,054** 0,046** 0,061**
Sig. (2-tailed) 0,000 0,000 0,000 0,000 0,000 0,000
22
Dummy year 2010 Pearson Correlation 1 -0,174** -0,169** 0,013 -0,006 0,005
Sig. (2-tailed) 0,000 0,000 0,314 0,657 0,767
N 5904 5904 5904 5904 5904 3329
Dummy year 2011 Pearson Correlation 1 -0,142** -0,095** -0,008 0,015
Sig. (2-tailed) 0,000 0,000 0,527 0,387
N 5904 5904 5904 5904 3329
Dummy year 2012 Pearson Correlation 1 -0,128** 0,025 0,030
Sig. (2-tailed) 0,000 0,059 0,080
N 5904 5904 5904 3329
CEO pay related resolution yes (1) or no (0)
Pearson Correlation 1 -0,064** -0,015
Sig. (2-tailed) 0,000 0,378
N 5904 5904 3329
Vote 50 % Pearson Correlation 1 0,778**
Sig. (2-tailed) 0,000
N 5904 3329
Vote For Percentage Pearson Correlation 1
Sig. (2-tailed)
N 3329
**. Correlation is significant at the 0.01 level (2-tailed). *. Correlation is significant at the 0.05 level (2-tailed).
23 Testing the hypotheses
H1: Institutional owners are more likely to submit a proposal aimed to influence CEO compensation at a shareholder meeting than individual owners.
The first hypothesis tests if there is a relationship between type of ownership and the likelihood of submitting a proposal related to CEO compensation.
To test this hypothesis an independent samples t-test is performed to compare the amount of CEO pay related resolutions between individual proposers and institutional proposers. The institutional proposers are taken together as one group, consisting of pension funds, union funds and SRI funds. The proposals of the undefined group “Other” is not taken into account and those labelled as not related to CEO payment. These are defined as missing in this test. The individual shareholder proposed 1718 resolutions of which 276 were CEO pay related; the institutional shareholders proposed 3499 resolutions of which 587 were CEO pay related.
Individual Shareholder with CEO related proposal
Frequency Percent Valid Percent Cumulative Percent
Valid Proposal not submitted by individual 1442 24,4 83,9 83,9
Proposal submitted by individual 276 4,7 16,1 100,0
Total 1718 29,1 100,0
Missing 4186 70,9
Total 5904 100,0
Table 5. Frequencies for individual sponsor type. Values missing are all other sponsor types that are not taken into account in this overview.
Institutional Shareholder (Pension, Union, SRI) with CEO pay related proposal
Frequency Percent
Valid Percent
Cumulative Percent
Valid CEO Proposal not submitted by institution 2912 49,3 83,2 83,2
CEO Proposal submitted by institution (pension,
union,sri) 587 9,9 16,8 100,0
Total 3499 59,3 100,0
Missing 2405 40,7
Total 5904 100,0
Table 6. Frequencies for institutional sponsor type. Values missing are all other sponsor types that are not taken into account in this overview. Institutions are Pension funds, Unions and SRI funds, and identified as specified in appendix A.
There is a highly significant difference (p=0,000) in the score for proposing a CEO payment related resolution by individual shareholders (Mean: 0,16; Standard Deviation: 0,367) and Institutional shareholders (Mean: 0,17; Standard Deviation: 0,374).
24 One-Sample Statistics
N Mean Std. Deviation Std. Error Mean
Individual_CEO_proposal 1718 0,16 0,367 0,009
Institutional (2,3,4) with CEO pay proposal 3499 0,17 0,374 0,006
Table 7. Dummy variables for the groups of individuals and institutions in which a proposal related to CEO payment has a value 1. N is the total amount of proposals done by the sponsor type. The mean gives the percentage of proposals with a value 1 (CEO
pay related).
One-Sample Test
Test Value = 1 (proposal is CEO pay related)
T Df Sig. (2-tailed) Mean Difference 95% Confidence Interval of the Difference Lower Upper Individual_CEO_proposal -94,714 1717 0,000 -0,839 -0,86 -0,82
Institutional (2,3,4) with CEO pay
proposal -131,731 3498 0,000 -0,832 -0,84 -0,82
Table 8. Outcome independent sample t test between individuals and institutions.
These results suggest that there is significant difference between the proposition of CEO related resolutions by both the institutional owner and the individual owner. The hypotheses Institutional
owners are more likely to submit a proposal aimed to influence CEO compensation at a shareholder meetings than individual owners cannot be rejected, but this outcome does not imply a direction of whom has
more influence either. Either the individual as well as the institutional shareholder has an influence. The institutional shareholder group consists of Pension funds, Union funds and SRI funds. The results do not tell whether one of these funds is more different than the others.
The regression analyses will shed a light on this later on.
As an extra test an ANOVA analyses is performed with the sponsor type as independent variable and the CEO pay related resolution as dependent variable, not taking into account the distribution over the years. The independent variable Sponsor type consists of 5 groups (Pension funds, Unions and SRI funds as well as the group “Other” as presented in table 2) which are subject to the ANOVA analysis. Table 9a shows the descriptives of each type of Sponsor.
25 Table 9a. Descriptives ANOVA test
CEO pay related resolution yes (1) or no (0)
N Mean Std. Deviation Std. Error Minimum Maximum
1. Individual 1718 0,16 0,367 0,009 0 1 2. Pension 764 0,11 0,308 0,011 0 1 3. Union 1373 0,26 0,440 0,012 0 1 4. SRI 1362 0,11 0,310 0,008 0 1 5. Other 687 0,12 0,321 0,012 0 1 Total 5904 0,16 0,366 0,005 0 1
Descriptives of the values 1-5 for Sponsor type which are tested in the ANOVA test. N is the amount of observations of this value, mean is the percentage of these in the total population.
The ANOVA test also shows a significant effect of the sponsor type of the resolution on the subject of the resolution being related to CEO payment at p<0,05 level for the 5 sponsor types [(4, 5899)=40,854, p=0,000]. ANOVA Table Sum of Squares df Mean Square F Sig.
CEO pay related resolution yes (1) or no (0) * Sponsor type 2
Between Groups (Combined) 21,359 4 5,340 40,854 0,000
Within Groups 771,023 5899 0,131
Total 792,382 5903
Table 9b. ANOVA table performed on CEO pay related resolutions per sponsor type. There is no significance between the different sponsor groups. Shown is total sum of squares and degrees of freedom between and within the types of sponsor.
When looking at the strength of the effect of the variances for the resolution type being CEO pay related or not, Eta squared can be used. An ETA of above 0,04 is interpreted as a significant relationship while above 0,36 it is strong. The calculated Eta squared is <0,04 and emphasized the findings above that there is no relationship between the different Sponsor types and whether the resolution is CEO pay related or not.
Measures of Association
Eta Eta Squared
CEO pay related resolution yes
(1) or no (0) * Sponsor type 2 0,164 0,027
Table 10. Eta squired calculated for resolution type and sponsor type
Hypothesis 1 Institutional owners are more likely to submit a proposal aimed to influence CEO compensation
26 ANOVA test both groups are found to be significant. In the regression analysis an attempt will be made to construct an order in which type of sponsor is most likely to submit a CEO pay related proposition.
H2: A proposal submitted by an institutional owner is more likely to be accepted than a proposal submitted by an individual.
The first hypothesis in this research is about proposing the resolution; this second hypothesis is about the acceptance of the proposed resolution. The subject of the resolution is not taken into account. The RM proposal database has the attribute Vote For percentage which is the percentage of voters that voted for the proposal. Of the 5904 proposals 3329 were filled in and 2575 had missing values. Of the 3329 filled, 738 had a majority of vote (50% or more) which is 22 % of the total proposals of which a vote is available.
Filed Vote For per sponsor type and minimum 50% of the vote
2007 2008 2009 2010 2011 2012 Row total Individual sponsor 364 341 322 328 236 127 1718 Less than 50% of the votes 174 135 142 131 94 55 731 Of which at least 50% in favor 57 61 78 72 43 19 330 Not brought to the ballot 133 145 102 125 99 53 657
Pension fund 117 122 152 123 93 157 764
Less than 50% of the votes 46 43 62 46 29 52 278 Of which at least 50% in favor 17 10 16 13 17 38 111 Not brought to the ballot 54 69 74 64 47 67 375
Union Fund 375 278 298 185 166 71 1373
Less than 50% of the votes 149 90 106 96 79 23 543 Of which at least 50% in favor 40 27 35 17 17 4 140 Not brought to the ballot 186 161 157 72 70 44 690
SRI 237 297 255 259 187 127 1362
Less than 50% of the votes 125 156 116 128 90 80 695 Of which at least 50% in favor 4 3 10 4 2 23 Not brought to the ballot 108 138 129 127 95 47 644
Other 67 109 90 115 73 233 687
Less than 50% of the votes 34 33 41 51 37 148 344 Of which at least 50% in favor 7 21 36 16 10 44 134 Not brought to the ballot 26 55 13 48 26 41 209
Total 1160 1147 1117 1010 755 715 5904
Less than 50% of the votes 528 457 467 452 329 358 2591 Of which at least 50% in favor 125 122 175 122 89 105 738 Not brought to the ballot 507 568 475 436 337 252 2575
Table 11. Overview of filled in Vote percentages in RM proposal database and Votes of at least 50% in favor of the proposal per Sponsor type and per year.
27 During the period 2007-2012 the amount of proposals brought to a vote declines, from 653 in 2007 to 463 in 2012 where the amount of proposals with a 50% of the vote or more fluctuates from 125 resolutions in 2007 to 105 in 2012 with an up- and downfall in 2009 and 2011. The fact that resolutions who did not get to the vote, or those resolutions in which the variable Vote For did not present a value are ignored and calculated as a case without 50% of the vote.
An independent t test was conducted to compare the amount of proposals with a 50% or more of the vote brought to the vote by individual shareholders and by institutional shareholders. For both groups the dummies were used. There was no significant difference in the score for the institutional sponsor of the proposal having a 50% of the vote (mean=0,17; Standard Deviation=0,378), conditions t=0,569, p=0,569 (equal variances assumed) or t=0,581, p=0,561 (equal variances not assumed). Individual shareholders do have a significant difference in their score for having 50% or more of the vote (mean=0,05; Standard Deviation=0,227), conditions t=5,755 (equal variances assumed) or t=6,951 (equal variances not assumed), both p=0,000.
Group Statistics
Dummy
VoteForProposal50% N Mean Std. Deviation Std. Error Mean
Dummy Individual with CEO pay related proposal
Vote in favor under 50% 731 0,19 0,391 0,014
Vote in favor of proposal
equal or above 50% 330 0,05 0,227 0,013
Dummy Institutional (2,3,4) with CEO pay proposal
Vote in favor under 50% 1516 0,19 0,389 0,010
Vote in favor of proposal
equal or above 50% 274 0,17 0,378 0,023
Table 12. Statistics of the two dummy variables Individual sponsors with a CEO pay related proposal and Institutional sponsors with CEO pay related proposals that made it to the vote. Values are given for the amount of proposals cast to the ballot and
whether they had <50% or 50% and above of the votes cast.
These results suggest that there is a difference in the significance of the occurrence of getting a majority vote between proposals brought forward by individual sponsors next to individual sponsors. This is the opposite of what is expected in hypothesis 2.
28
t-test for Equality of Means
t df Sig. (2-tailed) Mean Difference Std. Error Difference Dummy Individual with CEO pay
related proposal
Equal variances
assumed 5,755 1059 0,000 0,133 0,023
Equal variances not
assumed 6,951 993,990 0,000 0,133 0,019
Dummy Institutional (2,3,4) with CEO pay proposal
Equal variances
assumed 0,569 1788 0,569 0,014 0,025
Equal variances not
assumed 0,581 385,334 0,561 0,014 0,025
Table 13. Results independent T test for comparing the individual and institutional sponsor types as sponsors of CEO related proposals on the results of the voting on these proposals (majority of the votes or not).
29
H3: Proposals which refer to CEO compensation are more likely to obtain a majority vote than proposals with another subject
To test the third hypotheses a comparison is made between proposals with a majority vote related to CEO payment and proposals with a majority vote not related to CEO payment. Two constructs have to be used. First the vote in favor of the proposal of 50% or more has to be identified: to do so the Vote For a dummy is created in which value 0 is equal to an outcome of the vote in favor of the proposal of less than 50% and value 1 is 50% or more. The proposed resolutions are divided into CEO pay related (value 1) or not CEO pay related (value 0 ). All CEO pay related resolutions are (as quoted in the RM proposal database) presented in appendix B of this paper. In table 8 an overview is given of the total amount of proposals per year with a majority vote of >50% of the votes cast and of proposals which are CEO pay related or not.
Vote For equal to or above 50% and CEO pay related
2007 2008 2009 2010 2011 2012 Total
Individual sponsor 57 61 78 72 43 19 330
Vote >50% and CEO pay related 7 3 3 5 - - 18 Vote >50% not CEO pay related 50 58 75 67 43 19 312
Pension 17 10 16 13 17 38 111
Vote >50% CEO pay related 4 2 6 2 - - 14 Vote >50% not CEO pay related 13 8 10 11 17 38 97
Union 40 27 35 17 17 4 140
Vote >50% CEO pay related 5 5 6 2 - - 18 Vote >50% not CEO pay related 35 22 29 15 17 4 122
SRI 4 3 10 4 2 0 23
Vote >50% CEO pay related 1 2 9 3 - - 15 Vote >50% not CEO pay related 3 1 1 1 2 - 8
Other 7 21 36 16 10 44 134
Vote >50% CEO pay related 1 4 2 - - 7
Vote >50% not CEO pay related 7 20 32 14 10 44 127
Total 125 122 175 122 89 105 738
Vote >50% CEO pay related 17 13 28 14 - - 72 Vote >50% not CEO pay related 108 109 147 108 89 105 666
Table 14.Total amount of proposals per year with a majority vote of >50% of the votes cast, diverted into proposals which are CEO pay related and those who are not.
A table 14 shows, as of 2011 there are no resolutions proposed which are related to CEO payment, and 2009 is a year with a peak in the number of proposals related to CEO payment. There are two possible explanations for this: either the instrument of the non-binding vote is no longer used because under 2009 Dodd Franck Act companies are obliged to put CEO
30 compensation to the vote at least every three years. Only 72 proposals (12,5 %) related to CEO payment obtained 50% or more of the votes cast.
An independent samples t test is conducted to compare the group of CEO pay related proposals with a majority vote and the group of non CEO related proposals with a majority vote.
There was a significant difference in the scores between CEO pay related resolutions with (mean=0,13; standard deviation 0,341) and those who are not CEO pay related (mean=0,24; standard deviation=0,426) with a majority vote (t=5,401 (variances assumed equal) or 6,279 (variances assumed unequal); p=0,000).
Group Statistics
CEO pay related
resolution yes (1) or no (0) N Mean Std. Deviation Std. Error Mean
Dummy VoteForProposal50% Not CEO related proposal 2790 0,24 0,426 0,008
CEO pay related proposal 539 0,13 0,341 0,015
Table 15. Statistics on the dummy variable for proposals with a majority vote for whether the subject of the proposal is CEO pay related or not.
These results suggest that there is a significant difference between the proposals who obtained a majority vote as a result of their subject. Therefore hypothesis 3 cannot be rejected.
Independent sample t test
t-test for Equality of Means
t df Sig. (2-tailed) Mean Difference Std. Error Difference Dummy VoteForProposal50%
Equal variances assumed 5,401 3327 0,000 0,105 0,019
Equal variances not
assumed 6,279 897,383 0,000 0,105 0,017
Table 16. Results on the independent T test on the dummy variable for proposals with a majority vote for whether the subject of the proposal is CEO pay related or not.
31 Regression analyses
Sponsors do bring the subject of CEO payment to the vote, but the analyses cannot predict how the different factors influence the amount of CEO pay related proposals, or their success in the vote.
To explore this aspect a regression analyses is conducted in which dummies for the different groups are used, as well as dummies for the different years. By performing a correlation analysis and a regression analysis the association of the different variables with each other can be explored. A linear regression analyses helps to interpret the significances that were shown in the test and to see which variable has most influence.
The following models are explored: Model 1
Model 1 analyses the association between the percentage of the votes in favor of proposals and who is casting the vote.
Model 2
Model 2 analyses the association between the percentage of the votes in favor of CEO payment related proposals dependent on who is casting the vote.
Model 3
Model 3 analyses the association between the percentage of the votes in favor of CEO payment related proposals dependent on who is casting the vote and the year in which the votes were cast. ( )
32 Results of the regression analyses with two dependent variables: Percentage of Votes in
Favor of Proposals (model 1) and Percentage of Votes in Favor of CEO Pay related Proposals (models 2) during the years 2008-2012 (model 3)
OLS regressions – results
Model 1. Model 2. Model 3.
% of votes in favor of proposals
% of votes in favor of CEO pay related proposals % of votes in favor of CEO pay related proposals during 2008-2012 B t B t B t Constant 39,712 39,752 33,155 15,768 33,636 13,445 Dummy_Individual -1,338 -1,112 -6,015 -2,458** -8,259 -3,241*** Dummy_Pension 0,403 0,270 10,045 3,185*** 6,061 1,923* Dummy_Union -2,484 -1,907* 1,753 0,735 -0,131 -0,053 Dummy_SRI -20,821 -16,149*** 3,788 1,428 -0,476 -0,175 Dummy year 2008 2,150 1,203 Dummy year 2009 7,337 3,999*** Dummy year 2010 3,824 1,901* Dummy year 2011 -10,371 -3,051** Dummy year 2012 -13,673 -3,284*** Adjusted R-Square 0,121 0,077 0,148 F-value 115,869*** 12,234*** 11,414*** N 3328 538 538
* = significant at a 0,10, level, ** = significant at a 0,05 level, *** = significant at a 0,01 level
Table 17. Results of the OLS regression analyses on model 1, 2 and 3. In the models the following independent variables are used: Dummy individual (sponsor type individual shareholder); dummy Pension (sponsor type of the
shareholder is Pension fund); dummy Union (sponsor type of the shareholder is Union fund); dummy SRI (sponsor type of the shareholder is SRI fund); Dummy year 2008 (record is applicable to year 2008); Dummy year 2009 (record is applicable to year 2009); Dummy year 2010 (record is applicable to year 2010); Dummy year 2011 (record is applicable to year 2011);Dummy year 2012 (record is applicable to year 2012). All are dummies and have
a value 1 if the dummy is applicable and a value 0 if not. B (unstandardized coefficient) and t value are given.
In table 17 all results of the regression analyses are presented. In the next section the results are discussed per model. The B value (unstandardized regression coefficient) is used instead of the Beta (standardized regression coefficient) because as they are all dummies, the variables are all measured in the same units.
33 Model 1
In model 1 the Percentage of the votes in favor of proposals is the dependent variable. The regression analysis show that the model has an adjusted R square of 0,122 which means that approximately 12% of the variation of the percentage of votes in favor of proposals can be explained through the independent variables (the dummies individual, Pension, Union and SRI).
There is a significant effect of the sponsor type on the percentage of votes in favor of the
proposition at the p<0.05. This is consistent with our findings for hypothesis 2. The dummies for Unions and SRI are negative, which can be interpreted that these variables do not add to a higher vote in favor, with the highest score for SRI. SRI is highly significant (p=0,000) and Union is marginal significant (p=0,057) in explaining the percentage of votes in favor of proposals. Based upon the findings in model 1 it could be expected that proposals done by SRI generate the lowest amount of when compared to the other sponsor types. The other sponsor types have no
significant scores and therefore are not to be expected to have significant influence.
Model 2
The dependent variable Percentage of votes in favor of CEO pay related proposals is dependent on the dummies of the sponsor types for individual, Pension, Union and SRI shareholders. Model 2 generates an adjusted R square of 0.077, which means that approximately 7.7% of the variation of the percentage of votes in favor of proposals related to CEO payment can be explained through the independent variables in model 2.
The outcome of the votes for the different proposals when proposed by the 4 different sponsor types (individual, Pension, union and SRI) gives a significant effect of the sponsor type on the percentage of votes in favor of the proposition at the p<0,05.
In model 2 the individual sponsor (p=0,014) and the pension sponsor (p=0,002) have a
significant influence upon the percentage of votes in favor of a CEO pay related proposal. The pension sponsor even has a highly significant influence (p<0,01). The individual sponsor has a negative B which suggests that when an individual proposes the resolution, the effect on the percentage of vote in favor is negative. The B value for the pension dummy is positive which
