The influence of CEO Compensation on CSR
Performance and the Moderating Effects of CEO Tenure
and Board of Directors’ Diversity.
MSc. Business Administration: Strategy
Master Thesis
Dimitri Coucoudis
10246967
Supervisor: Dr. Daniel Waeger
1 Statement of originality
This document is written by Dimitri Coucoudis, who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2 Abstract
This study examines the relationship between CEO compensation (short-term and long-term) on the one hand, and Corporate Social Responsibility (CSR) performance, as measured by KLD scores, on the other. In addition to this relationship, the moderating effect of the variables CEO tenure and board of directors’ diversity are tested. The final sample consists of 449 firms from the S&P 500 index over the period of 2014 and 2015. Agency and stakeholder theory are used to formulate the hypotheses. This study finds that short-term CEO compensation does not significantly influence CSR performance and that long-term CEO compensation does positively affect CSR performance. Also, evidence is found for a negative moderating effect of CEO tenure on the relationship between long-term CEO compensation and CSR performance. These results contribute to the limited scientific knowledge of simultaneous effects of executive compensation and individual executive characteristics on CSR performance.
3 Acknowledgements
I would like to thank my supervisor, Dr. Daniel Waeger, for his guidance
throughout the thesis process. His suggestions often resulted in eye-opening and valuable insights while always providing enough space for personal ideas and preferences.
4 Table of contents I. Introduction ... 6 2. Literature review ... 7 2.1 CEO compensation ...8 2.1.1. Base salary ...8 2.1.2. Annual bonus ...9 2.1.3. Stock options ...10 2.2. CSR Performance ...11
2.3. CEO tenure and board of directors’ diversity ...12
2.3.1. CEO tenure ...13
2.3.2. Board of directors’ diversity ...14
2.4. Prior literature on relationship ... 15
2.5. CEO power for CSR performance ... 16
2.6. Conclusion and research questions ... 18
3. Conceptual model ... 19
4. Research design ... 28
4.1. The Model ...28
4.1.1. Sample and data collection ...29
4.2. Measures ...31
4.2.1. Measurement of dependent variable – CSR Performance ...31
4.2.2. Measurement of independent variable – CEO compensation ...32
4.2.3. Measurement of control variables ...33
4.2.4. Measurement of moderator variables ...34
4.3. Statistical analysis ...35
5. Discussion ... 41
5.1 Academic relevance ...42
5.2 Managerial implications ...44
5.3 Limitations and suggestions for future research ...44
6.0 Conclusion ... 45
References ... 47
5 List of tables
Table Title Page
1 Measurement of CSR performance 52
2 Short- and long-term CEO compensation 53
3 Control variables 53
6 I. Introduction
The impact of modern economic activities on the quality of human and social life has led to greater societal awareness of CSR, which is defined as situations where the firm goes beyond compliance and engages in ‘‘actions that appear to further some social good, beyond the interest of the firm and that which is required by law’’ (McWilliams and Siegel, 2001; McWilliams et al., 2006). CSR performance is therefore becoming an increasingly important business outcome for researchers, investors and society as a whole (Deckop et al., 2006). Firms are under increasing pressure to enhance CSR performance in part because of emerging standards related to CSR performance such as the Caux Roundtable Principles for Business, the United Nations Global Compact and the Organization for Economic Cooperation and Development Guidelines for Multinational enterprises (Waddock, 2003) and in part because of the proliferation of independent evaluations and rankings such as the Kinder, Lydenberg, Domini (KLD) that make social performance more transparent (Deckop et al., 2006).
The nature of a firm’s corporate governance is used to persuade executives to emphasize particular firm goals and objectives (Mahoney & Thorne, 2006) and an important component of corporate governance is the CEO’s compensation structure (Jensen & Meckling, 1976). CEO’s compensation structure is a visible and potentially important mechanism through which owners and the board of directors can direct managerial attention to specific objectives having implications for CSR performance (McGuire et al., 2003). CEO support is essential for CSR outcomes and CEOs have extensive power to influence CSR performance (Fabrizi et al., 2014; Kochan, 2002; Orlitzky & Swanson, 2002; Rangan et al., 2015). This leads to the question whether CEO compensation is properly structured to provide incentives for improving CSR performance, for which various studies have provided answers (Merriman & Sen, 2012). These prior studies demonstrate that CEO compensation affects CSR outcomes but almost no literature examines executive characteristics and compensation in tandem. In other words, how individual
7 executive characteristics interact with executive compensation schemes in shaping CSR performance currently represents an area of research that has only received scant attention thus far (Hambrick, 2007). Furthermore, amongst others, Margolis and Walsh (2003) have commented on methodological and conceptual problems in CSR performance research, which include the omission of potential moderators. This study adds to these voids by examining two moderator variables that are expected to affect the strength of the relation between CEO compensation and CSR performance (Baron & Davidson, 1986), namely CEO tenure and gender diversity in the board of directors.
Research in the Upper echelons vein indicates that executive characteristics influence strategic outcomes (Hambrick et al., 1984). An example of such a characteristics is CEO tenure. Research shows that long-tenured CEOs tend to grow ‘‘stale in the saddle,’’ ceasing to make adaptive changes (Hambrick, 2007) and therefore CEO tenure can moderate the relationship between CEO compensation and CSR performance. Also in light of upper echelons theory, the board that supervises the CEO, the board of directors, has the power to influence CSR outcomes through the control it exerts on the CEO. If the board of directors has a high percentage of female directors, this characteristic is able to moderate the relationship between CEO compensation and CSR performance, as female directors tend to value CSR more.
The next section provides a review of the literature which leads to the research questions of this study. The third section describes the conceptual model and hypothesis. The fourth section describes the research design and statistical analysis. The fifth section provides the discussion and provides avenues for future research. The sixth section provides the conclusion. 2. Literature review
In this literature review, a review of the existing literature on the topic is given. In the first section, the variable CEO compensation and its concepts of short-term and long-term CEO compensation are explained. The second section describes the emergence and definition of CSR
8 performance. In the third section, the influence of CEO tenure and board of directors’ diversity on the main relationship is explained. The fourth section lists prior literature on the relationship between CEO compensation and CSR performance. The fifth section explains CEO power for CSR performance. Finally, a preliminary conclusion is given and the research questions are provided.
2.1 CEO compensation
The modern history of executive compensation began in the early 1980s and paralleled the emergence of agency theory (Jensen and Meckling, 1976), of which a description will be given later in this study. The compensation paid to CEOs of listed firms rose dramatically during the 1980s and 1990s, stimulating much debate on the determinants of executive pay. Therefore, there has been an extreme increase in academic research on executive compensation (Murphy, 1999).
Most CEO compensation packages contain four basic components: a base salary, an annual bonus, stock options and long-term incentive plans (Murphy, 1999). Research in the Upper echelons vein indicates that executive compensation affects company strategic outcomes (Hambrick, 2007). Research in both psychology and finance has specified important behavioral consequences of the compensation structure (Gerhart & Milkovich, 1990) and managerial incentives may encourage strategies or implementation methods having positive or negative implications for social performance (McGuire et al., 2003). The following three paragraphs will describe the first three components of CEO compensation. Furthermore, the short-term or long-term incentives that theory associates with these components are set out and discussed.
2.1.1. Base salary
Base salary is a contractual agreement between the CEO and the firm that is set out by the board of directors and is generally paid in cash. Base salary is typically determined through competitive benchmarking based on general industry salary surveys and is supplemented by
9 details from market peers (Murphy, 1999). Base salary is considered to be short-term compensation, since base salary is a fixed amount of cash and mostly based on past-year performance (Gerhart & Milkovich, 1990). However, Jensen and Murphy (1990) find that the link between base salary and corporate performance is fairly low.
Theory suggests that designing compensation schemes on a strategy of regular increments to base salary, compared to other performance based measurements to base salary, reduces the chance to achieve individual, group or organizational goals (Gerhart & Milkovich, 1990). This supports the view that base salary does generally not provide much incentives in achieving goals, or at least that there are other compensation structures that are more effective in doing so. Critics might claim that high salary levels themselves are indicative of poor social performance (McGuire et al., 2003). Deckop et al. (2006) argue that CEOs might have a particular disincentive to engage in CSR when their pay focuses on short-term incentives such as base salary. CSR performance strengths involve, amongst others, taking positive action in the areas of community relations, human rights and environmental strengths. Attention to these may not only have a direct negative effect on the firm’s short-term financial performance; it may even represent an opportunity costs for the CEO as it represents resources not spend on maximizing short-term performance (Margolis & Walsh, 2003).
2.1.2. Annual bonus
The second component of short-term compensation is the annual bonus and is paid out in cash, just like base salary. An annual bonus normally consist out of a minimum bonus which is paid at a threshold performance. There is usually a cap on this bonus and the range between the threshold and cap is labeled as the ‘‘incentive zone’’ (Murphy, 2000). The annual bonus is based on the performance of the previous year and therefore provides short-term incentives (Healy, 1985). It rewards executives for achieving short-term goals rather than for building the firm’s long-term potential (McGuire et al., 2003). For example, high levels of short-term bonus
10 payments may lead executives to focus on the more immediate ‘‘bottom line’’, rather than building long-term stakeholder relationships which can enhance CSR performance.
The payout for maximizing short-term earnings is more salient when this offers substantial rewards based on short-term performance or a lack of disincentive to focus on the short-term (Deckop et al., 2006). Research indicates that executives even manipulate earnings to maximize the present value of their annual bonus (Healy, 1985; Holthausen et al., 1994). Executives manipulate earnings downwards when they are at the upper bound of their annual bonus contracts in order to increase the present value of payments from this bonus. This will lead executives to withhold effort when the cap of the bonus is reached and shows that executives are willing to engage in potential future value destroying activities to secure current annual bonus. 2.1.3. Stock options
The last component of executive compensation discussed in this study, is executive stock options or restricted option grants. Executive stock options give the executive the right to buy a share of stock at a pre-specified exercise price for a pre-specified term (Hall & Murphy, 2000). Setting the exercise price, like setting the incentive cap on the annual bonus, defines the standard against which performance is measured. Executive stock options provide clear value-increasing incentives for CEOs (Murphy, 1990) and are usually exercisable over the stock price over a three-to five year period and are therefore considered to offer long-term incentives (Murphy, 1999). Long-term compensation started to make a noticeable impact on median pay from the 1960s and became a greater share of compensation over time, reaching more than 35% of total pay in 2005. This is mostly due to a growing popularity of stock options (Frydman & Saks, 2010). Executive stock options have emerged as the single largest component of executive compensation, after base salary (Hall & Murphy, 2000). The idea of an executive stock option is that it provides the CEO with the incentive to make decisions that will results in an increase in the future value of his or her holdings (Mahoney & Thorne, 2005). This reduces pressure to
11 maximize short-term earnings and provides a longer time frame wherein CSR performance is more likely to thrive (Deckop et al., 2006).
2.2. CSR Performance
Firms are under increasing pressure to increase CSR performance in part because of emerging standards related to social performance (Deckop et al., 2006). Despite the growing amount of highly visible CSR principles, standards and codes, reporting on these issues is inconsistent throughout countries (Holder-Webb, 2009). In general, CSR disclosures are more heavily and consistently provided outside the U.S. This might be due to a lack of regulation standards for U.S. firms (Holder-Webb, 2009). An explanation for this lack of regulation for U.S. firms is dependent on the power of the state, which is generally lower in the U.S. than in Europe (Matten & Moon, 2008). Furthermore, shareholders are relatively dispersed among public listed U.S. firms, which limits the power of certain stakeholder groups. Also, great prominence has typically been given to market self-organization in the U.S., which touches upon a significant number of CSR issues, such as liabilities for production and products (Matten & Moon, 2008). This self-organization reduces the consistency of CSR reporting among countries and among U.S. firms. To overcome these consistency problems, CSR performance is measured according to independent evaluations and rankings that make social performance more transparent and consistent (Waddock, 2003). This study therefore uses the independent database MSCI ESG KLD, which was created by Kinder, Lydenberg and Domini (KLD) in 1991. Hereafter, this study will simply refer to this database as the KLD database.
The KLD database is the largest multidimensional and publicly available CSR database and is extensively used in research on CSR performance (Harrison & Freeman, 1999). The KLD rating scheme has been tested for construct validity against other measures and has been found to be one of the best measures for CSR performance to date (Hilman & Klein, 2001). A downside of using a single rating agency such as KLD is that it might result in source bias (Manner, 2010).
12 Through company surveys and secondary sources such as the business press and academic journals, KLD obtains information on firms (Agle et al., 1999). In the KLD database, firms are rated on their impacts on seven dimensions: the community, corporate governance, diversity, employee relations, environment, human rights and product quality. Prior studies that use the KLD database to find CSR performance are now described.
McGuire et al. (2003) determine CSR performance by finding strengths and weaknesses along four dimensions: employee relations, community, product quality and environment. Then for each firm, strengths and weaknesses are aggregated along the four dimensions for a measure of total strengths and total weaknesses. Deckop et al. (2006) measure CSR performance along six dimensions: employee relations, product quality, community, environment, human rights and diversity. A net score per dimension is calculated by subtracting total concerns from total strengths and the net scores are summed from each dimension for a grand total. Callan and Thomas (2011) measure CSR performance along all seven dimensions in the KLD database. The authors construct a five-point scale where minus two indicates that there are two or more concerns and two indicates two or more strengths. An average is then calculated to determine CSR performance. Cai et al. (2011) construct a CSR composite index consisting out of strengths and weaknesses along five dimensions: community, environment, diversity, employee relations and product quality. These dimensions are added up together with the maximum possible number of concerns per dimensions. This is then divided by the total maximum possible number of strengths and concerns at the same year to construct the CSR composite index.
2.3. CEO tenure and board of directors’ diversity
This section describes two moderator variables and their importance for the main relationship between CEO compensation and CSR performance. In the conceptual model, these variables and the intuition behind their expected effects on the main relationship are explained in more depth
13 and hypotheses are derived. In the following paragraph, CEO tenure is discussed. In the second paragraph, board of directors’ diversity is discussed.
2.3.1. CEO tenure
Hambrick and Mason (1984) synthesis of the ‘‘Upper echelons perspective’’ describes strategic choices and performance levels that are partially predicted by observable managerial background characteristics. This perspective argues that CEOs act on the basis of their psychological orientations, values, cognitions and beliefs (Hambrick et al., 1993). This logic contends that complex decisions are largely the outcomes of behavioral factors rather than attempts to achieve economic optimization. This may provide some benefits by offering greater power to predict organizational outcomes such as CSR performance. In a recap of this article, Hambrick et al. (2007) conclude that almost no literature examines executive characteristics and compensation in tandem, or their interactive effects in shaping company outcomes. The authors provide an example for future research by considering the effects of executive characteristics and compensation.
Examples behind the intuition of differences in CEO behavior towards CSR are now provided, based on CEO tenure. Consider the effect on CSR performance of a long-term incentive plan for a short-tenured CEO and consider the effect on CSR performance of the same long-term incentive plan for a long-tenured CEO. Short-tenured CEOs have a longer time horizon and, therefore, long-term compensation will impact them more strongly in their choices. Therefore, the relationship between long-term compensation and CSR performance will be strengthened for short-tenured CEOs. Conversely, because long-tenured CEOs have a shorter time orientation, the relationship between short-term compensation and CSR performance will be stronger for firms with such CEOs compared to firms with short-tenured CEOs.
Another reason for these differences in CEO behavior arises because short-tenured CEOs have a lower entrenchment with the status quo than long-tenured CEOs (Shen & Canella, 2002).
14 In light of upper echelon’s perspective (Hambrick & Mason, 1984), the observable managerial characteristic CEO tenure can predict how ‘‘entrenched’’ a CEO is to the status quo. The degree of entrenchment then moderates the relationship between short- and long-term CEO compensation and CSR performance.
2.3.2. Board of directors’ diversity
Board of directors’ diversity is defined as the percentage of female directors in the board of directors. The one-tier board of directors of publicly listed U.S. firms makes decisions on shareholders’ behalf as a fiduciary. The hiring and firing of senior management, option policies, executive compensation and setting broad goals are among the duties of the board (Eisenberg et al., 1998).
Hambrick (2007) introduces the idea that a focus on the demographic characteristics of other top employees will yield a stronger explanation of organizational outcomes than the sole focus on the individual CEO. CSR performance can be defined as an organizational outcome. The use of demographic data, although better available and accessible, leaves us at a loss to the real psychological and social processes that are driving executive behavior. The board of directors has the power to use base salary, annual bonus and stock options as means to provide the CEO with rewards for superior performance and penalties for poor performance (Jensen & Murphy, 1990). Therefore, demographic characteristics that explain the orientation of the board of directors can help to yield a stronger explanation of organizational outcomes that contribute to the explanation of the relationship between CEO compensation and CSR performance.
Upper echelon theory (Hambrick & Mason, 1984) describes comparable demographic board characteristics that can determine strategic outcomes and CSR performance. A demographic characteristic of the board of directors is its diversity, as measured by the percentage of female directors. General research to date has provided insight in the difference between male and female behaviour. These insights partially explain the general different
15 attitudes towards CSR for men and women, where women tend to care more for the future and for CSR (Adams & Funk, 2011).
Therefore the gender composition of the board of directors is important to fully explain CEO’s behaviour to engage in activities that can increase CSR performance. The board of directors’ diversity has a time-related component to it, just like CEO Compensation and CSR performance, as gender partly determines future discounting. The effect of CEO compensation on CSR performance will be different depending on the board of directors’ gender diversity, which will be explained in the conceptual model of this study.
2.4. Prior literature on relationship
There have been some studies regarding the effects of CEO compensation components on CSR performance. This paragraph provides a chronological order of this literature.
Stanwick and Stanwick (2001) examine the relationship between CEO base salary and stock gains, and the firm’s reputation based on the firm’s commitment to the community and the environment, using the Fortune Reputation Index. Their sample of 372 firm observations of Fortune 500 firms shows a strong relationship between CEO compensation, firm environmental reputation, firm size and financial performance. In particular, they find a strong negative correlation between CEO compensation and firm environmental reputation, which supports the view that CEO compensation reduces CSR performance of the firm. McGuire et al. (2003) examine the relationship between CEO incentives and CSR performance for 374 S&P firms and find that high levels of base salary and long-term incentives have a positive association with social performance as measured by the KLD database. Mahoney and Thorne (2005) examine the relationship between long-term compensation and CSR for a small sample of 393 firm-year observations of publicly traded Canadian firms and find a positive relationship between long-term compensation and attention to CSR, as measured by ratings of the Canadian Social Investment Database (CSID).
16 Deckop et al. (2006) examine whether CEO compensation is properly structured to provide incentives towards improving CSR performance, as measured by the KLD database. Using a sample of 313 S&P 500 firms, they find that a short-term CEO pay focus is negatively related to CSR performance whereas a long-term focus is positively related to CSR performance. In a similar study as Deckop et al. (2006), Mcguire et al. (2003) find no significant relationship between short-term incentives and CSR performance; this might be due to model specifications or the different inclusion of dimensions of CSR performance (Deckop et al., 2006). Berrone and Gomez-Mejia (2009) examine various relationships between CEO compensation and CSR performance. Using data on 469 U.S. firms and 2088 firm-year observations, they find that long-term compensation increases pollution prevention success, as measured by environmental governance information from proxy statements reported to the Securities and Exchange Commission (SEC).
Fabrizi et al. (2013) study monetary incentives and non-monetary incentives in relationship with CSR using a sample of 597 US firms and 2520 firm-year observations and find that both monetary and non-monetary incentives have an effect on CSR decisions. Specifically, monetary incentives designed to align CEO’s and shareholders’ interests have a negative effect on CSR and non-monetary incentives have a positive effect on CSR. Here, monetary incentives are equity incentives such as stock options and restricted option grants and non-monetary incentives are related to career concern variables.
2.5. CEO power for CSR performance
Upper echelons theory (Hambrick & Mason, 1984) states: ‘‘Organizational outcomes, both strategies and effectiveness, are viewed as reflections of the values and cognitive bases of powerful actors in the organization.’’ Therefore it is expected that CEOs are important for CSR performance. This section will describe studies that examined this topic.
17 Studies on the boundaries of CEO’s power in relation to CSR performance like Godos-Díez (2011) try to explain the importance of CEOs to CSR practices. Using survey data from 149 CEOs in Spain, they find that CEOs closer to the stewardship model are more inclined to implement CSR practices, relative to CEOs that enhance the agency model. Stewardship theory is based on the idea that there is a moral incentive for managers to do the right thing, without regards to how such decisions affect firms’ financial performance (Donaldson & Davis, 1991). Agency theory, in its simplistic form, states that CSR is indicative of self-serving behaviour on the part of managers and therefore reduces shareholder wealth (Friedman, 1970).
In addressing the roots of some early century corporate scandals, Kochan (2002) explains that CEOs possess significant decision-making power and have the ability to change the firm’s impact to the community. Orlitzky and Swanson (2002) provide similar conclusions when examining the proposition that executives’ receptivity to social values is key for aligning CSR behaviour. Executives are increasingly under pressure to respond to various stakeholder expectations as explained by stakeholder theory, which will be defined later in this study. The authors furthermore recognize country specific and cultural differences across national boundaries as factors that potentially influence CEO values and hence CSR decision-making. Deckop et al. (2006) acknowledge the view by Kochan (2002) that the role of the CEO has been scrutinized by business scandals in the early century, especially in the context of the stakeholder view. The authors further elaborate on Orlitzky and Swanson (2002) and find that CEOs are recognized to have extensive decision making power and have the ability to effectively affect CSR performance of their firms.
Waldman et al. (2006) empirically test for the role of the CEO in determining the extent to which firms engage in CSR. Using survey-based data from 56 U.S. and Canadian firms, they find that CEO intellectual stimulation is significantly associated with the propensity to invest in CSR. Therefore this CEO attribute significantly determines a part of CEO’s power on CSR
18 performance in this study. Fabrizi et al. (2014) empirically confirm the prominent role of the CEO in relation to CSR decisions, based on 2520 firm-year observations for 597 firms, while also recognizing the complexity of non-monetary factors affecting CSR. Rangan et al. (2015) conduct a survey on 142 executives who followed Harvard’s Business School CSR executive program and consistently conclude that coordinated support for CSR at executive management is critical for CSR performance.
Based on these studies, CEO power seems to be extensive enough to significantly influence CSR performance. CEO power is affected by various forms of uncertainty and CEO characteristics, therefore in addition to the moderator variable CEO tenure, a control variable for CEO power will be included in the regression. The next section will provide a short conclusion based on the literature review and provides two research questions.
2.6. Conclusion and research questions
This study provides insights in the relationship between CEO compensation structure and CSR performance, as measured by various dimensions of the KLD database. CEO compensation consists out of short-term and long-term compensation. Short-term compensation is generally associated with weak CSR performance whereas long-term compensation is associated with strong CSR performance. Data on CSR performance is obtained from the KLD database. CEOs matter for CSR performance; they have sufficient power to influence CSR outcomes and are considered to be an essential force when trying to increase CSR performance. This leads to the following research question: What is the relationship between CEO compensation structure and CSR performance?
This study extends this research question and prior research by adding two moderator variables to the main relationship. The board of directors’ diversity and CEO tenure are expected moderator variables, for which an extensive rationale is provided in the next section. To my knowledge, there have been no prior studies that examined these moderator variables in addition
19 to the first research question. This leads to the second research question: How do CEO tenure and board of directors’ diversity moderate the relationship between short- and long-term CEO compensation and CSR performance?
3. Conceptual model
In this section, a total of six hypotheses are developed in regard to the research questions stated in the preceding paragraph. First, the expected relation between short-term and long-term CEO compensation and CSR performance is explained using agency theory and stakeholder theory, as these are important theories regarding the causal impact of executive compensation on CSR performance (Cai et al., 2011). Second, the expected moderating effects of CEO tenure and board of directors’ diversity on this relation are discussed.
Agency theory can provide arguments for the incentives of short- and long-term CEO compensation and its effects on CSR performance. In its original formulation by Jensen and Meckling (1976), agency theory is defined as a contract under which one or more principals engages the agent to perform some service on their behalf which involves delegating some decision making authority to the agent. An agency problem can appear because of risk sharing that arises when the principal and agent have different attitudes towards risk (Eisenhardt, 1989). The problem here is that the principal and the agent might exert utility from different actions because of their different risk preferences. Under the assumption that both the principal and the agent are utility maximizers, there is good reason to believe that the agent will not always act in the best interests of the principal (Jensen & Meckling, 1976). The principal can limit divergences from his interest by establishing appropriate incentives for the agent. Applying agency theory to this study, the shareholder represents the principal and the CEO represents the agent. In the situation where shareholders have complete information regarding CEO’s activities and the firm’s investment opportunities, a contract could be designed specifying and enforcing CEO’s action to be taken in each state of the world (Jensen & Murphy, 1990). However, CEO’s actions
20 are not perfectly observable to shareholders as shareholders have information and time constraints that limit perfect observability. Therefore, agency theory predicts that compensation policy will be designed to give the CEO incentives to select and implement actions that increase shareholder wealth (Jensen & Murphy, 1990). Agency theory assumes that CEOs serve shareholders’ interests, who primarily value financial performance (Eisenhardt, 1989; Jensen & Murphy, 1990). Stakeholder theory is used to explain that, in addition to shareholders, CEOs serve various other stakeholders who value different performance measures such as CSR performance. In addition to agency theory, stakeholder theory will be used to enlarge the standard principal-agent paradigm (Donaldson & Preston, 1995).
The classic definition of a stakeholder was formulated by Freeman (1984) as ‘‘A stakeholder in an organization is by definition any group or individual who can affect or is affected by the achievement of the organization’s objectives.’’ In contrast to the single shareholder in agency theory, stakeholder theory argues that managers bear a fiduciary duty to a range of different stakeholders. The interest of all stakeholders merits consideration for its own sake and not merely because of its ability to further the interests of some other group, such as the shareholders (Donaldson & Preston, 1995). Stakeholders include, but are not limited to, employees, customers, suppliers, communities and the general public (Hill & Jones, 1992). These stakeholders have different degrees of salience to CEO’s management depending on the legitimacy of their expectations, the urgency of their claims and power regarding the firm (Mitchel et al., 1997). The instrumental part of stakeholder theory establishes a framework for examining the connections between the practice of stakeholder management and the achievement of various corporate performance goals. Firms that practice stakeholder management and thus are socially responsibly firms, will be relatively successful in conventional performance terms such as profitability and growth (Donaldson & Preston, 1995). This outcome of stakeholder management or stakeholder theory is therefore aligned with serving shareholders’ primary
21 interest, as described in agency theory. In this study, stakeholder theory will supplement agency theory in formulating hypothesis for the expected effects of short- and long-term CEO compensation on CSR performance and for the effects of the two discussed moderator variables. Although there are various ways to classify CEO compensation, the most common one is by short-term and long-term compensation (Murphy, 2012). As previously mentioned, base salary and annual bonus are considered to be short-term compensation as these are generally based on past-year performance (Gerhart & Milkovich, 2010). Socially responsible investments are general those with a longer time horizon (Mahapatra, 1984), which take more time to generate returns that contribute to financial performance or ‘‘The short-term performance pressures companies face rule out indiscriminate investments in social value creation.’’ (Porter and Kramer, 2006). In addition, investments in CSR normally do not have easy calculable returns a priori. Because base salary is based on retrospective accounting measures, executives compensated solely based on salary and annual bonus are not focused on factors that influence CSR performance (Mahoney & Thorne, 2006). Furthermore, assuming that CEO’s total compensation is partially determined by firm risk, firms that actively undertake CSR activities face a lower firm risk due to a smaller degree of conflict of interest between stakeholders (Cai et al., 2011). Short-term CEO compensation will then be lower as part of a lower total CEO compensation.
The expected negative effect of short-term incentives on CSR performance can be further worsened when the CEO fails to hit his or her short-term financial targets, which not only affects short-term CEO compensation but also the job security of the CEO (Cai et al., 2011). The CEO is then expected to invest even less in CSR projects as these will not provide the necessary short-term improvements in financial performance. Short-short-term compensation is expected to have a negative effect on CSR performance; its short-term incentives will not stimulate CSR related investments. This leads to the first hypothesis.
22 Hypothesis 1: There is a negative relationship between short-term CEO compensation and CSR performance.
As stated in the literature review, executive stock options are considered to be long-term compensation as these are generally based on three-to five year performance (Murphy, 1999). Stock options are contingent on market valuation and therefore consistent with a concern for CSR as executives must consider potential costs and benefits of environmental and social factors that may impact firm performance in the long-term (Mahapatra, 1984).
In addition to imposing risk on executives and aligned with agency theory, long-term compensation can be used to align executives’ self-interests with that of society, because socially responsible actions are generally those with a longer term time horizon (Mahapatra, 1984). For example, environmental abuses may go unnoticed and unpunished in the short-term but the probability of detection increases as the number of incidents over time increase (Mahoney & Thorne, 2006). Consequently, firms that focus CEO compensation on executive stock options are expected to take actions that are more socially responsible than firms with less focus on long-term compensation. A long-long-term compensation focus reduces pressure to maximize short-long-term earnings (Deckop et al., 2006). Furthermore, Flammer and Bansal (2017) find that long-term executive compensation leads to an increase in investments in R&D and stakeholder engagement, especially pertaining to employees and the natural environment. Employee relations and the natural environment are common indicators for CSR performance, as described in part 2.2. Long-term compensation provides the long-term orientation and timeframe in which CSR performance is more likely to increase. This leads to the second hypothesis.
23 Hypothesis 2: There is a positive relationship between long-term CEO compensation and CSR performance.
As stated in part 2.3.1 of the literature review, CEO tenure is an observable managerial characteristic that is expected to have a moderating effect on the relationship between short- and long-term CEO compensation and CSR performance. Hambrick and Mason (1984) describe various psychological factors that are of central significance to their Upper echelons theory. However, these phenomena are rarely studied or measured directly in empirical studies of CEOs (Hambrick et al., 1993). Rather, psychological orientations are typically imputed from more readily observable characteristics, such as CEO tenure.
Long-tenured CEOs have a stronger psychological commitment to the status quo (Hambrick & Mason, 1984), which is defined as a belief in the enduring correctness of current organizational strategies and profiles (Hambrick et al., 1993). Some CEOs seem to become psychologically attached to ‘‘what is’’, which derives both from CEO’s preferences as well as from knowledge.
On the one hand, a CEO can be committed to the status quo because one beliefs that the firms ought to continue just as it is, as one values the current state of affairs and would incur a loss if there were changes. As long-tenured CEOs spend more time in the firm, they become convinced of the correctness of the organization’s ways (Hambrick et al., 1993). This occurs both through socialization and self-selection. These CEOs have both psychologically and tangibly invested a great deal in the status quo and have generally struggled for years to achieve their position. They often have far more to lose than to gain from organizational changes and are thus committed to the status quo (Hambrick et al., 1993). Long-tenured CEOs also have much more power to resist external pressures (Meyer, 1995), this increases the homogenizing of the organization and increases CEO’s autonomy and influence (Miller, 1991). It also makes it easier
24 for them to resist external pressures for changes in strategy and structure, which can deteriorate the link between the organizational and environmental fit. Long-tenured CEOs tend to have a weaker fit between the firm’s strategy and environmental demands (Shen & Cannella, 2002).
On the other hand, a CEO could be committed to the status quo because it is all he or she knows, unaware of other options. CEOs carry as part of their cognitive and emotional givens the experiences they have had during their careers. Long-tenured CEOs within the same firm can be assumed to have relatively limited perspectives (Hambrick & Mason, 1984). Long tenure creates a common, internally-shared perspective and long-tenured CEOs may be less able to grasp new ideas and learn new behaviors (Hambrick & Mason, 1984). Over time, CEO’s beliefs evolve unified, internally reinforcing configurations among elements of strategy that are close to the status quo. Eventually these perceptions of strategy are narrowed and the need for reorganization goes unnoticed (Shen & Cannella, 2002). Commitment to the status quo is as a reason why long-tenured CEOs tend to lead to strategic persistence (Hambrick et al., 1993), while a change in strategy is often necessary to increase CSR performance.
In light of stakeholder theory, current stakeholders will demand more environmental actions that lead to CSR performance. Short-tenured CEOs engage significantly more in CSR than long-tenured CEOs (Lewis et al., 2014). This may be due to the fact that the new CEO need to gain legitimacy towards a broad group of stakeholders and has a longer time horizon (Fabrizi et al., 2014). Their longer time horizon enables them to adhere less to the short-term incentives of salary and annual bonus. Therefore the negative relationship between short-term compensation and CSR performance will be less strong for firms with short-tenured CEOs. This leads to the following hypothesis.
25 Hypothesis 3: CEO tenure positively moderates the relation between short-term CEO compensation and CSR performance, such that this relationship is stronger for CEOs with longer tenure.
Short-tenured CEOs have a weaker entrenchment with the status quo and are expected to be more sensitive to stakeholders’ CSR pressures. Following stakeholder theory, short-tenured CEOs are more likely to adapt their firms to evolving demands of a varied set of stakeholders. This ensures that firms continue to fit well into their external environment (Lewis et al., 2014). This should increase CSR performance as current stakeholders demand more CSR performance. Conversely, because short-tenured CEOs tend to have a longer time horizon, long-term compensation will have a stronger impact on them than on long-tenured CEOs. Therefore, the positive relationship between long-term compensation and CSR performance will be strengthened for short-tenured CEOs compared to firms with long-tenured CEOs. This leads to the following hypothesis.
Hypothesis 4: CEO tenure negatively moderates the relationship between long-term CEO compensation and CSR performance, such that this relationship is weaker for CEOs with longer tenure.
Most CEO positions of the firms in the dataset are dominated by men. The firms in this study contain only 23 female CEOs out of 449 firms, which is 5.12%. Although CEO’s gender could be a moderator variable for the relationship between short- and long-term CEO compensation and CSR performance, this small sample size does not allow for significant empirical research. One particularly relevant information system for monitoring executive behaviours is the board of directors (Eisenhardt, 1989). Therefore the diversity of the body that supervises the CEO, i.e. the
26 board of directors’ diversity is examined as a demographic characteristic that can be a moderator variable for the relationship between CEO compensation and CSR performance, as stated in part 2.3.2. Differences between male and female behaviour are at the foundation of this effect. These behavioural differences stem from evolved human nature (Griskevicius et al., 2012).
From an evolutionary perspective, men should discount the future more steeply than women (Griskevicius et al., 2012). In the vast majority of mammals, including humans, the male counterparts have lower obligatory investments in offspring (Griskevicius et al., 2012). Theories of parental investment and sexual selection posit that the sex in a species with lower obligations for their offspring will be more risk taking, aggressive and have a shorter life expectancy than women (Daly & Wilson, 1988). This implies that female directors should be more long-term oriented than male directors.
Another implication of this gender difference is that men are more willing to deplete environmental resources and engage in wasteful consumption for financial gain. This is backed up by various studies that find that men are more willing to conspicuously waste the environment for financial gains. Furthermore, female directors significantly value universalism and benevolence more than male directors (Adams & Funk, 2011), which implies that female directors are more stakeholder orientated (Van Zijl et al., 2011), and therefore tends to enhance CSR performance. Following the literature, a board of directors that contains more female members will then be more long-term oriented and more focused on its role as a principal to stakeholders’ demand for CSR performance (Hillman & Dalziel, 2003).
In conclusion, research has provided arguments for why men are more short-term orientated than women and why women are less willing to deplete environmental resources. Since men are more short-term orientated than women, male directors should care more for short-term compensation than long-term compensation so that the negative relation between
27 short-term compensation and CSR performance is weakened when more women are part of the board of directors. This leads to the following hypothesis.
Hypothesis 5: Board of directors’ diversity negatively moderates the relationship between short-term compensation and CSR performance, such that this relationship is weaker for firms with higher percentages of female directors.
From an agency theory perspective, boards can be used as monitoring devices for shareholder interests (Fama & Jensen, 1983), whereas stakeholder theory asserts that boards should balance the conflicting claims of multiple stakeholders (Eisenhardt, 1989). An ambitious theoretical integration of agency theory and stakeholder theory led to the creation of the stakeholder-agency theory, which constitutes a generalized theory of agency (Hill & Jones, 1992). This theory entails that the CEO is the manager of all stakeholders of the firm, instead of just the shareholders. If the board of directors would supervise the CEO more according to the stakeholder theory or stakeholder-agency theory model, CSR performance would increase as environmental and social issues receive greater attention than under the agency theory model.
Since women are more stakeholder and long-term oriented than man, boards with more female directors are likely to have a long-term orientation under which CSR performance thrives. Therefore, if the body that supervises the CEO, i.e. the board of directors, contains more female directors, the relationship between long-term compensation and CSR performance should be strengthened. This leads to the following hypothesis.
Hypothesis 6: Board of directors’ diversity positively moderates the relationship between long-term CEO compensation and CSR performance, such that this relationship is stronger for companies with higher percentages of female directors.
28
Short-term CEO compensation
Long-term CEO compensation
CSR Performance H3: CEO Tenure +
H5: Board of directors’ diversity -
H4: CEO Tenure -
H6: Board of directors’ diversity +
H1
H2
-
+ An overview of the expected relations can be found in figure 1.
Figure 1: Conceptual model
4. Research design
In the following section, a description of the empirical model and sample and data collection is given. Next, the different variables used in the statistical analysis are explained. Finally, the statistical tests used are explained and the results of the study are presented.
4.1. The Model
Based on the literature review, the empirical model used in this study is presented below.
CSRPerformancet =
α
0 +β1ST.CEOcompt + β2LT.CEOcompt + β3ROAt + β4TAt + β5BoardIndt29 Whereby:
- CSRperformancet is a proxy for net CSR performance in year 2015;
- STCEOcomptis the independent variables short-term CEO compensation in thousands of dollars for the year 2014;
- LTCEOcomptis the independent variable long-term CEO compensation in thousands of dollars for the year 2014;
- ROAtis the control variable Return on Assets as a proxy for financial performance in the year 2014;
- TAtis the control variable Total Assets in thousands of dollars as a proxy for firm size in the year 2014;
- BoardIndt is the control variable percentage of independent/outside directors as a proxy for board independence in the year 2014;
- CEOownershiptis the control variable percentage of shares held by the CEO as a proxy for CEO power in the year 2014;
- CEOTenure * Short-termCompensation is the interaction variable Z1 that acts as a moderator variable for CEO tenure.
- CEOTenure * Long-termCompensation is the interaction variable Z2 that acts as a moderator variable for CEO tenure.
- BODGender * Short-termCompensation is the interaction variable Z3 that acts as a moderator variable for board of directors gender.
- BODGender * Long-termCompensation is the interaction variable Z4 that acts as a moderator variable for board of directors gender.
4.1.1. Sample and data collection
The research questions as stated in paragraph 2.6. and the empirical model described above, provide a justification for a quantitative research design. This study focuses on S&P 500 firms and examines how short- and long-term CEO compensation relate to CSR performance. To test the hypotheses and provide answers for the research questions, data is collected from Wharton Research Data Services (WRDS) and its constituents for the years 2014 and 2015. Data on the independent, control and moderator variables is collected for 2014 and data on the dependent
30 variable is collected for 2015, assuming a one-year lag on CEO compensation and CSR performance. The choice of the appropriate lag is an interesting challenge, as no studies specifically address this particular matter (Coombs and Gilley, 2005). Although it is uncertain to exactly how long it takes compensation to affect CSR, the one-year lag seems appropriate as most studies work with a one-year lag. The year 2015 is chosen for CSR performance as this is the most recent year where information on CSR performance is available through KLD.
The S&P 500 firms over 2014 and 2015 are found using Bloomberg. This consists out of a sample of 500 firms for both years. The two years are then compared and firms not active in the index in both years are deleted, this leads to a preliminary sample of 479 firms. After finding the control variables ROA and TA, 474 firms remain. After finding the independent short- and long-term compensation, three firms were dropped from the sample for which no complete data on short- and long-term compensation was found, resulting in a sample of 471 firms. After finding the dependent variable CSR performance, seven firms for which no complete set of CSR performance along all five dimensions was found were dropped from the sample. Then six firms were dropped where no information on the control variable CEO independence or the moderator variable supervisory board gender was found. Finally, nine CEOs had a negative tenure, which indicates that they became CEO after December 31, 2014. The nine firms of these CEOs were deleted from the sample, leading to a final complete sample of 449 firms.
Aligned with prior studies, the normality of the data is tested by computing the values of skewness and kurtosis. Skewness is considered acceptable when the values lie in the interval of -0,8 and -0,8. Kurtosis is considered acceptable when the values are in the interval of -3 and 3. Variables that are located out of the acceptable interval are transformed by taking the logarithmic value. The transformation that provides the lowest values of skewness and kurtosis is then used, although kurtosis can still be outside the acceptable interval for some variables. Table 1 displays the initial descriptive statistics for the sample.
31 Table 1: Descriptive statistics for the sample
Variable N Mean Standard
deviation Min Max CSR performance 449 2.403 2.317 -2 12 Short-term compensation 449 1394.853 1186.653 .001 10523.75 Long-term compensation 449 7902.805 6708.803 0 90845.6 ROA 449 4.727 11.350 -118.607 32.691 Total Assets 449 67228.92 210757.4 1702.96 2351698 CEO ownership percentage 449 .643 2.394 0 22.997 Director independence percentage 449 .833 .094 .333 1 CEO tenure 449 2523.866 2302.399 1 19357
Board of directors’ diversity 449 .200 .092 0 .556
Notes: This table presents the descriptive statistics for the sample of this study. The non- logarithmic values are reported here, while in the regressions most variables are transformed to correct for skewness and/or kurtosis. Short- and long-term compensation and total assets are reported in thousands. CEO Tenure is reported in days.
4.2. Measures
In the following section, the different variables as stated in the model in 4.1. will be discussed. First, the dependent variable and then the two independent variables will be explained. Then, the four control variables and two moderator variables will be discussed.
4.2.1. Measurement of dependent variable – CSR Performance
This study uses CSR performance as dependent variable. CSR performance is calculated from data obtained from the KLD database. The KLD database provides information on a wide range of CSR strengths and weaknesses that are used to determine CSR performance. Strengths and weaknesses are available along the dimensions community, corporate governance, diversity, employee relations, environment, human rights and product safety features. Corporate governance is not used to determine CSR performance as there are two control variables related to corporate governance in the empirical model and KLD does not provide useful data for this
32 dimension. Diversity is not used to determine CSR performance, as KLD does not provide a large amount of relevant strengths and weaknesses for the time-period of this study. Also, in the variables that are available, there is an overlap with the moderator variable board of directors’ diversity in this study. The resulting final five dimensions, and its strengths and weaknesses along which CSR performance is measured, are summarized in appendix 1.
KLD assigns a one if a firm possesses a certain strength, a zeo if this firm does not have this strength and a NA if there is no information available to determine whether the firm possesses this strength. Similarly a one, zero or NA is assigned to indicate whether a firm has or does not have a specific weakness. CSR performance is then developed by calculating the sum of all strengths minus the sum of all weaknesses per firm. This is the most straightforward calculation and aligned with prior studies (Deckop et al., 2006; McGuire et al., 2003).
Because the initial data on CSR performance is skewed (skewness = .840) and (kurtosis= 3.574), it is transformed in to a logarithmic value using the formula: Y = LOG10 (K + 1 – X), where K is the highest value for CSR performance found in the data and X is CSR performance. The transformed new variable is significantly less skewed (skewness = -.057) and (kurtosis = 1.917) but loses 102 observations. Therefore the following formula in Stata is used to transform the CSR performance variable to zero skewness: Y = LN(± EXP – K), where EXP is CSR performance and K is again the highest value for CSR performance found in the data. Skewness of the new variable is then -.0005 and kurtosis is 2.772, which are both acceptable to assume a normal distribution and no observations are lost.
4.2.2. Measurement of independent variable – CEO compensation
This study has two independent variables, short-term and long-term CEO compensation. These two variables are obtained from the Execucomp database. The Execucomp database contains data on various aspects of short- and long-term CEO compensation.
33 Short-term compensation is defined as the sum of the value of salary and annual bonus (Deckop et al., 2006; Fabrizi et al., 2003 McGuire et al., 2003) and long-term compensation is defined as the sum of the value of option awards and stock awards (Berrone & Gomez-Mejia, 2009; Mahoney & Thorne, 2005; McGuire et al., 2003; Stanwick & Stanwick, 2001). For a specification of the exact Execucomp variables used to determine short- and long-term compensation, see appendix 2.
Short-term CEO compensation has an initial skewness of 4.604 and kurtosis of 28.023. After transformation in to its logarithmic value using the formula Y = LN(± EXP – K) in Stata, skewness is low ( -0.000) and kurtosis is fairly high (9.806). Skewness is then acceptable but kurtosis is still outside the acceptable interval. Long-term CEO compensation has an initial skewness of 5.591 and kurtosis of 58.615. After transformation in to its logarithmic value using the formula Y = LN(± EXP – K), skewness (0.000) and kurtosis (5.005) are further improved. Kurtosis for both independent variables is thus outside the acceptable interval, however skewness is highly acceptable for both.
4.2.3. Measurement of control variables
This study uses a comprehensive set of control variables, based on prior research. The first two are the most widely recognized determinants of CEO pay, firm size and financial performance (Berrone & Mejia, 2009). Prior studies tend to control for firm size (Berrone & Gomez-Mejia, 2009; Callan & Thomas, 2010; Callan & Thomas, 2014; Coombs & Gilley, 2005; Deckop et al., 2006; Fabrizi et al., 2013; McGuire et al., 2003; Stanwick & Stanwick, 2001) and financial performance (Benson & Davidson, 2010; Cai et al., 2011; Deckop et al., 2006; Fabrizi et al., 2013; Mahoney & Thorne, 2006). It is not surprising that CEO compensation increases with firm size and performance (Murphy, 1998) and both have been linked to CSR performance (Mahoney & Thorne, 2005). Firm size is measured as the amount of total assets and financial performance is measured as ROA, as shown in appendix 3.
34 The variable ROA is highly skewed and therefore transformed using the formula: Y = LOG10 (K + 1 – X), after which (skewness = -.747) and (kurtosis = 3.117) are fairly low but 40 values are dropped from the sample. Therefore the formula: Y = LN(± EXP – K) is used, which provide a skewness of 0.000 but a higher kurtosis of 14.443. The variable TA is highly skewed (7.800) and kurtosis (72.273) and therefore transformed to its logarithmic value using the formula: Y = LOG10 (K + 1 – X), skewness is then .722 and kurtosis is 3.661.
Another set of common control variables accounts for governance structure. CEOs that have considerable power could determine their own compensation and have a direct influence on the level of CSR performance (Cai et al., 2011), it is therefore important to control for this variable. CEO ownership percentages serve as a proxy for CEO power (Berrone & Gomez-Mejia, 2009; Cai et al., 2011; Coombs & Gilley, 2005; Mahoney & Thorne, 2006) and is directly found in Execucomp, as shown in appendix 3. Furthermore, the proportion of independent directors is often used as a control variable as indicator of board independence (Berrone & Gomez-Mejia, 2009; Cai et al., 2011; Deckop et al., 2006; Fabrizi et al., 2014). Independent board members tend to be more likely to serve stakeholders’ CSR interests (Fabrizi et al., 2014) and are therefore expected to increase CSR performance. Board independence is computed by dividing the number of independent directors by the total amount of board members.
The variable CEO ownership is highly skewed (5.961) and is therefore transformed to its logarithmic value using the formula: Y = LOG10 (K + 1 – X), skewness is then .718 and kurtosis is 4.452. The variable director independence is skewed (-1.371) and therefore transformed using the formula: Y = LN(± EXP – K), skewness is then 0.000 and kurtosis is 4.733.
4.2.4. Measurement of moderator variables
This study tests for the effects of two moderator variables. Data for the moderator variable CEO tenure is obtained through Execucomp and data on the moderator variable board of directors’ diversity is obtained through ISS, as listed in appendix 4. CEO tenure is calculated by
35 subtracting the date as CEO from the date 31-12-2014. The data is transformed in Microsoft Excel to calculate the number of days as CEO as a measure of CEO tenure. CEO tenure has an initial skewness of 2.121 and is therefore transformed using the formula Y = LN(± EXP – K), which gives an acceptable skewness of 0 and kurtosis of 2.452.
The variable board of directors’ diversity is calculated by collecting the size of the board of directors of the corresponding company and the amount of female directors. The amount of female directors is then divided by the total board size to compute the moderator variable board of directors’ diversity. The moderator variable board of directors’ diversity has an initial skewness of .598 and kurtosis of 3.724 and is therefore transformed to its logarithmic value using the formula Y = LOG10 (K + 1 – X), which leads to an acceptable skewness of -.184 and kurtosis of 2.393.
4.3. Statistical analysis
To conduct the statistical analysis, the statistical software package Stata is used. Table 2 presents the correlation between the independent, dependent, control and moderator variables. The
variables are tested on multicollinearity by computing the correlations between the independent and the control variables. The general threshold for multicollinearity is considered to be 0.60 (Deckop et al., 2006). None of these correlations has a higher value than 0.60, as can be seen from table 2. To further test for other forms of multicollinearity, the variance inflation factor (VIF) is used. The VIF test provides a highest score of 1.64 and a mean score of 1.33. This is well below the conventional point of 10 (Neter et al., 1985). Therefore, no evidence for multicollinearity issues are found.
36 Table 2: Pearson correlation matrix
Variables 1 2 3 4 5 6 7 8 9 1 CSR performance(1) 1 2 Short-term comp.(1) 0.063 1 3 Long-term comp.(1) 0.201 ** 0.455 ** 1 4 ROA(1) -0.069 0.081 0.027 1 5 TA(1) 0.149 ** 0.358 ** 0.218 ** 0.274 ** 1 6 CEO ownership(1) -0.140 ** -0.147 ** -0.073 -0.049 * 0.249 ** 1 7 Director independence -0.073 -0.115 * -0.118 * -0.113 * -0.200 ** 0.154 ** 1 8 CEO tenure(1) -0.047 0.191 ** 0.169 ** -0.037 -0.025 0.549 ** 0.025 1 9 BOD diversity(1) 0.150 ** 0.031 0.070 -0.084 0.168 ** -0.088 -0.182 ** -0.071 1
Notes: This table reports the Pearson correlation coefficients between the variables for the 449 firm observations in 2014 and 2015. ** and * represents statistical significance at the 1 and 5% level respectively.
(1)
: The logarithmic values are used for these variables.
To test whether short-term compensation and long-term CEO compensation are related to CSR performance, two separate hierarchical regressions are performed. The first hierarchical regression, with short-term CEO compensation as independent variable, is performed in regression 1. First, the control variables are added to the model. Then short-term compensation is added to the model, which does not significantly increase R2. Table 3 shows the results of this regression. The influence of short-term CEO compensation on CSR performance is statistically not significant (B = -0.054, SE = .034, P = .083). Therefore, there is no support for hypothesis 1. The second hierarchical regression with long-term CEO compensation as independent variable, is performed in regression 2. Table 3 shows the results of this regression. Adding long-term CEO compensation to the model increases R2 with 0.029 to 0.077, which is significant at the 1% level. The influence of long-term CEO compensation on CSR performance is statistically significant (B = 0.118, SE = 0.032, P = 0.000). Therefore, hypothesis 2 is supported.
37 Table 3: Regression results
Variables Regression 1 (H1) Regression 2 (H2) Regression 3 ROA(1) -0.147* (0.060) -0.139* (0.059) -0.139* (0.059)* TA(1) 0.038** (0.014) 0.029* (0.013) 0.035* (0.014) CEO ownership(1) -0.022* (0.010) -0.022* (0.010) -0.022* (0.010) Director independence(1) -0.028 (0.032) -0.019 (0.032) -0.019 (0.032) Short-term CEO compensation(1) -0.054 (0.034) -0.031 (0.018) Long-term CEO compensation(1) 0.118** (0.032) 0.071** (0.017) Constant 1.875** 0.882* 1.895** R2 0.048 0.075 0.081 Adjusted R2 0.037 0.065 0.068 F-Statistic 4.39** 7.16** 6.44**
Notes: This table reports the robust regression results from the multivariate regressions for regression 1 and regression 2. Standardized variables are reported for both independent variables in regression 3.
** and * represents statistical significance at the 1 and 5% level respectively.
(1)
: The logarithmic values are used for these variables.
Because the regressions consist out of more than two variables, the moderator effect could be tested with a multiple F-test that represents the change in the regression by adding the interactions (Osborne, 2008). The first step in testing for the significance of the moderation effect is to enter all predicting variables and the second step is to enter the moderator variable and interaction term (Frazier et al., 2004). If an interaction is present, then the difference between the two R2 values will be significant, using the hierarchical F test. The interaction effect can also be calculated by first testing the effect of the moderator variable on the dependent variable and then testing the effect of the interaction variable on the dependent variable. The interaction variable consists out of the moderator variable and the independent variable. To test for the moderating effects as described in hypothesis 3 to 6, regression 3 in table 3 is extended by including the different moderator variables and their interactions with the independent variables. Before creating the interaction terms, the independent and moderator variables are standardized
