CSR-related Executive Compensation and the Cost of Equity Capital – An Empirical Analysis

(1)

Master’s Thesis A&C, Controlling.

CSR-related Executive Compensation and the Cost of

Equity Capital – An Empirical Analysis

University of Groningen, Faculty of Economics and Business

MSc Accountancy & Controlling (A&C) Master’s Thesis Controlling (EBM870B20)

Name: Luke Lowijs

Student number: S2976617

E-mail: L.Lowijs.1@student.rug.nl

Supervisor: Dr. R.C. Trapp

Co-assessor: Dr. C.A. Huijgen

Word count: 11.550

(2)

1

Abstract

There have been no conclusive answers regarding the effect of CSR on the financial performance and value of a firm. There are lines of reasoning for both positive and negative effects of CSR activities on firm performance and firm value. This study tries to expand this discussion by investigating the effects of CSR-related executive compensation on the Cost of Equity Capital. According to agency theory, executives will give explicit attention to CSR activities when measures based on CSR activities are included in their compensation scheme. The hypotheses tested whether CSR-related executive compensation influenced the Cost of Equity Capital either directly or through financial performance. CSR-related executive compensation has been categorized as either social or environmental for 119 firms by means of 2016 proxy statements which describe the executive compensation over 2015. The other financial and control variables have been collected for fiscal 2015, and fiscal 2016 as a lagged year and robustness check. The total sample sizes were 111 firms for the fiscal year of 2015 and 94 firms for the fiscal year of 2016. Firms were randomly selected from the Standard & Poor’s 500 index. This study shows that CSR-related executive compensation does not influence the Cost of Equity Capital directly. CSR-related executive compensation does however reduce the financial performance of the firm (measured through ROA). This was robust for the social dimension as this is significant for the fiscal year of 2015 and the fiscal year of 2016. The environmental dimension of CSR was only significant with a decline in ROA in fiscal 2015. Weak evidence suggests that the social dimension of CSR-related executive compensation reduces the financial performance and will subsequently increase the Cost of Equity Capital. This was not found for the environmental dimension of CSR-related executive compensation.

(3)

2

1. Introduction

Society has increasing interest in Corporate Social Responsibility (hereafter: CSR) (Klassen & McLaughing, 1996; El Ghoul et al., 2011). CSR is defined by the European Commission as an approach “whereby companies integrate social and environmental concerns on their

business operations and in their interaction with their stakeholders on a voluntary basis”

(COM, 2002; Neal, 2008, p. 464). Activities linked to CSR thus may consist of social and environmental activities, e.g. waste/pollution prevention and reduction (King & Lenox, 2002), reduction of chemical emission (Connors & Silva-Gao, 2008), (product) safety, sustainability, diversity and employee well-being (Hong et al., 2016).

Firms recognise this need of society and have put increasing emphasis on CSR activities and the disclosure of CSR activities (Dhaliwal et al., 2011). A reason for this is that members of society stress the importance of the natural environment and are prepared to pay extra for products which do not have a negative effect on the environment (Klassen & McLaughing, 1996). Academics have investigated the effect which CSR (and the disclosure and implication of CSR activities) has on firms, either by observing financial performance (Harjoto & Jo, 2015; King & Lenox, 2002; Nollet et al., 2016; Trumpp & Guenther, 2017), social and environmental performance (Russo & Harrison, 2005) or another measure of interest, e.g. the cost of capital, (long-term) executive compensation and executive compensation structure (Dhaliwal et al., 2011; Berrone & Gomez-Mejia, 2009; El Ghoul et al., 2011; Mahony & Thorne, 2005; Mahony & Thorn, 2006). However, there have been no conclusive answers regarding the effect of CSR on the financial performance and value of a firm. This study relies on two different lines of reasoning regarding the effect of CSR activities on firm variables, a value-reducing and a value-enhancing view. Proponents of the value-reducing view see CSR activities as not pursuing profit/value maximization while proponents of the value-enhancing view see these activities as a way to make processes more efficient and as a way to create new capabilities, resulting in positive financial returns.

The value-reducing view focuses on the trade-off between environmental performance and financial performance (Klassen & McLaughing, 1996) and is based on agency theory. Friedman (1962, 1970) sees the commitment of time and resources towards CSR as a possible agency cost. Many studies which investigate (negative) effects of CSR on financial performance and other financial variables call upon Friedman’s view of the negative effect of

(5)

4 CSR on financial performance (e.g. King & Lenox, 2002; Margolis & Walsh, 2003; Harjoto & Jo, 2015). Committing resources towards CSR is, according to Friedman, not automatically in agreement with the maximization of shareholder wealth and profit making. Hong et al. (2016) describe this as “self-interested behaviour by individual managers at the expense of the

firm’s shareholders” (p. 200). Recent studies show that investment in CSR may be used to

prioritize personal interest over those of the shareholders (Borghesi et al., 2014; Brown et al., 2006, Jiraporn & Chintrakan, 2013; Krüger, 2015). CSR activities may take place to enhance the executive’s reputation instead of maximizing shareholders´ wealth (Barnea & Rubin, 2010). Another way of ‘personal use’ of CSR which the executives can use is to please stakeholders and consequently entrench oneself within their function within the firm (Cespa & Cestone, 2007). Investment in new technology also may be substantial when altering processes within the firm to become more environmental friendly (Walley & Whitehead, 1994). Other reasons include that savings from emission reduction may take time to be realized and reorganizations (e.g. renegotiation of contracts) may have to take place (White et al., 1993).

The second view sees CSR activities as a value-enhancing opportunity. Orlitzky et al. (2003) call upon stakeholder theory to explain the increase in future financial performance when CSR activities are enacted within firms. They argue that “satisfaction of various stakeholder

groups is instrumental for organizational ﬁnancial performance” (Orlitzky et al., 2003, p.

405). Further elaboration reveals that investments in Corporate Social Performance may help firms develop new capabilities, competencies and resources which manifest in the company culture, structure, HR and technology. These competencies lead towards a more efficient use of resources and in turn increase the financial performance of the firm. More capable employees may be interested in firms in which CSR activities take place, in turn enhancing the efficiency and improving financial performance. The reputation of the firm is also enhanced when the firm invests in CSR activities, this reputation enhancement may in turn result in increased sales and higher financial performance.

A way to incentivize executives to invest resources towards CSR activities is by linking executive compensation with CSR-related targets and measures (Hong et al., 2016; Russo & Harrison, 2005). Behaviour of executives is directed towards CSR activities since part of their compensation is based on CSR-related targets and measures. Following the previous discussing, linking CSR measures and targets to executive compensation may either increase

(6)

5 or decrease firm performance through the resources which are being used for CSR purposes and the other mentioned effects. Hong et al. (2016) examine CSR-based compensation for executives and find common descriptions for CSR-based compensation in proxy statement such as “Community, Diversity, Greenhouse Gas Emission Reduction, Product Safety and

Sustainability” included in the compensation scheme of executives. Altering compensation

structures will lead to executives working towards a specific goal for the organization (Mahony & Thorn, 2006). Executive compensation consists of a fixed part (salary), short-term variable incentives and long-term variable incentives (Kane, 2002; Mahony & Thorne, 2005; Mahony & Thorn, 2006). Since CSR targets and measures are performance based, CSR based compensation will be part of either the short-term or long-term variable incentives. CSR activities have shown to increase when CSR-based measures and targets are included within the compensation scheme of executives (Hong et al., 2016)

The possible positive or negative effect which linking CSR with executive compensation has on firm performance will also influence other financial variables. This study focuses on the effect of CSR-related executive compensation on the Cost of Equity Capital. Chen et al. (2013) stress the importance of the Cost of Equity Capital (hereafter: COEC), stating that it is “a more direct yardstick of corporate investment and financing decisions than firm valuation”

(p. 850). Firms use the COEC when determining where to allocate their resources since the

cost of capital summarizes the trade-of between risk and return. This is done by using the cost of capital to calculate the future cash flow of the projects. If the value-reducing (value-enhancing) view is supported, incentivizing executives to allocate resources towards CSR is seen as reducing (increasing) financial performance, in turn increasing (decreasing) the COEC. The goal of this study is to provide further evidence regarding the effects of CSR and CSR-related executive compensation. Keeping the aforementioned discussion in mind, the research question of this research is:

“How does CSR-related executive compensation influence the Cost of Equity Capital?”

A sample of 111 firms from fiscal 2015 and 94 firms from fiscal 2016 have been examined with regard to CSR-related targets and measures from 2016 proxy statement which provide information regarding fiscal 2015 executive compensation. No significant relation was found between CSR-related compensation and the COEC. This implies that investors do not expect an effect of CSR-related executive compensation on future earnings per share since the

(7)

6 implied COEC does not change significantly. CSR-related executive compensation has found to be negatively associated with financial performance, this was robust for the social dimension of CSR-related compensation, not for the environmental dimension. This implies that inclusion of social CSR-related targets and measures in the executive compensation leads to activities which reduce the financial performance of the firm. Weak evidence was found that the reduction in financial performance due to the inclusion of social CSR-related compensation targets leads to an increase in the COEC.

This research makes several contributions to the academic literature and theory. First, the two competing views regarding the positive or negative effect of CSR-related activities have not yet brought a concise answer. This research extends current knowledge regarding these views by extending this towards CSR-related executive compensation. Second, this research employs a variable which reflects CSR-related compensation more precisely than previous attempts to do so by extending CSR in a social and environmental dimension (e.g. Hong et al, 2016). This research adds to the knowledge of the effect of the corporate governance tool of executive compensation by investigating previously not investigated dimensions of executive compensation in depth. Third, this research aims to contribute to the understanding of the COEC and capture a new variable which influences the COEC. This is of importance since the COEC is a key measure when valuing firms as well as corporate investments.

Practical contributions include increased knowledge concerning executive compensation, the effects of executive compensation as well as increased knowledge surrounding the estimation of the COEC.

This paper is structured as follows. The section 2 addresses the literature review and shows the development of the hypotheses. Section 3 describes the methodology, which include the operationalization of the variables which are used. Section 4 shows the results of the analyses and section 5 concludes the paper with a discussion, limitations, implications and ideas for potential future research.

(8)

7

2. Literature Review and Hypothesis Development

2.1. Executive Compensation

Corporate governance tools are used by firms to align behaviour of executives with behaviour which is beneficial for the firm’s shareholders by means of financial returns on the investments which they made (Shleifer & Vishny, 1997). Executive compensation (structure) is one corporate governance tool which discourages executives to deviate from the wanted behaviour via adverse selection or moral hazard (Beronne & Gomez-Meija, 2009; Shleifer & Vishny, 1997) Executive compensation has to be related to informativeness, riskiness and controllability of the performance targets (Beronne & Gomez-Meija, 2009). Well defined compensation structure and compensation components will help to align the actions of executives with the share- and stakeholder’s interests, leading to goal congruence between executives and share- and stakeholders (Connelly et al., 2011). Goal congruence between the executives and the share- and stakeholders is necessary to adjust behaviour of executives so that this will benefit both the executive and the share- and stakeholders.

The corporate governance tool of executive compensation structure may be tied to CSR measures and targets to incentivize executives to devote time and resources towards CSR activities. Agency theory states that agents (executives) are self-interested and do not necessarily prioritize organizational goals. For this reason, incentives should be which correspond with organizational objectives to increase the likelihood that these objectives are achieved. Thus, according to agency theory, executives will give explicit attention towards elements which are included within their compensation structure. Including CSR based measures and targets in the compensation structure thus directs the effort of executives towards CSR activities since part of their compensation is based on this measure (Russo & Harrison, 2005). An easy to measure incentive approach will further enhance the usefulness of the incentive structure. Hong et al. (2016) test this by using a control hypothesis which investigates whether CSR-related executive compensation in fact increases CSR activities. They found that including CSR-based measures and targets in the executive’s compensation structure does increase the firm’s CSR activity levels. This suggests that executive effort and attention is directed towards CSR by including CSR based measures in their compensation structure. Russo & Harrison (2005) investigated the relation between environmental

(9)

8 performance and compensation based on environmental. This study measured environmental performance by looking at the emission of electric plants. Surveys were used to gather data about the tie between compensation and environmental performance. They found that a tie between environmental performance and compensation increased environmental performance for facility managers. This also suggests that the manager’s attention includes environmental practises when this is included within their compensation structure when executives otherwise would not have given (explicit) attention towards environmental practises. Thus, incentivizing CSR practises leads to the fact that facility managers is likely to direct their attention towards these CSR practises to acquire part of their performance compensation. This result did not hold for environmental quality managers which were employed in the electric plants. A possible explanation for this division is that environmental performance managers already try to minimize emissions because this is related to the activities which are required for the profession whilst facility managers do not do so in such a manner.

2.2. The Value-Reducing View on CSR

Research on the effect of CSR on financial performance has not yet brought a conclusive answer regarding whether there is a positive or negative effect. Friedman´s (1962; 1970) agency cost prediction regarding CSR activities is one of the most prominent arguments which were made for the negative relation between CSR activities and financial performance. Friedman argues that CSR activities are beyond the scope of firms since firms are meant to generate profit and CSR activities do not necessarily correspond with profit making. As mentioned in Fisher-Vanden & Thorburn (2011), Molloy et al. (2002) found that investors see environmental investments to raise costs except when these investments are made to reduce non-compliance penalties which the firm would receive when these investments were not made. This study shows this by means of negative abnormal returns for firms which partake in climate change programs on a voluntary base. Borghesi et al. (2014) investigate the reason why executives invest in socially responsible investment opportunities. Managers can have different reasons why they invest in CSR activities other than financial measures on CSR activities. One of these reasons is that the personal and professional reputation of the executive may be enhanced by investing in social or environmental friendly initiatives. Another reason is that executives feel like they have the moral obligation to invest in social or environmental friendly activities. These motivational reasons for investing (or sometimes over-investing) in CSR activities are not in line with the profit-maximizing thought which

(10)

9 shareholders typically have according to agency theory. Barnea & Rubin (2010) show that high CSR investments may come from personal reputation enhancing motives rather than a profit maximizing motives. CSR expenditures may exceed the level which optimizes firm value (over-investment) for the reason that a higher CSR rating may enhance the executive’s individual reputation. Thus, over-investment in CSR activities may come from reputation enhancing motives rather than financial motives. This may lead to a decrease in financial performance, firm value and shareholder value. Cespa & Cestone (2007) and Prior et al. (2008) mention that managers which do not live up to expectations can use CSR investments as a way to please external stakeholders and use this to entrench themselves in their current position. This leads to the fact that inefficient managers remain within the firm when they should be replaced with a more efficient manager to increase financial performance of the firm. Consequently, the investment in CSR activities has a negative effect on the potential financial performance which the firm could have when a more capable manager would be incorporated in the company.

The formerly mentioned consequences on firm value and financial performance respectively all have in common that executives use firm resources for purposes which do not lead to profit maximization. In typical for-profit organizations, the shareholders expect the executive to allocate resources towards profit-maximizing goals. When the executive does not use the resources for this purpose, this is called the agency problem. The executive (agent) has an advantage over the shareholders (principal) in terms of information availability. This can lead to the agent using the principal’s resources for purposes other than the principle’s purpose (making profit and increasing value). These other uses of resources include the formerly mentioned reputation enhancement and entrenchment when investigating CSR activities.

As mentioned before, incentivizing executives to commit resources towards CSR activities by including CSR-related targets and measures in their compensation structure helps to direct their attention towards these CSR activities since part of their compensation is based on CSR (Hong et al., 2016; Russo & Harrison, 2005). The directing of attention from executives towards CSR activities will likely increase the level of CSR activities within the companies. Incentivizing executives to spend time and resources towards CSR activities by means of tying executive compensation with CSR-related measures and targets may have negative consequences for financial performance according to this view. Compensating executives for CSR investments and activities, which are not necessarily in line with profit and value

(11)

10 maximization, can thus be seen as an agency cost (Friedman, 1962; 1970) Impairment of (short-term) financial performance may lead to the situation that providers of equity capital may be inclined to increase the COEC to make up for the more uncertain financial nature of these investments. This increased risk leads to the fact that providers of equity capital are less certain of their financial return. These investors will want a higher return (COEC) since they could most likely find another firm with less risk which could provide the same return. Increased firm risk will thus lead to a higher COEC. The beta component of the COEC accounts for the perceived risk of the firm, a higher perceived firm risk thus increases the beta which in turn increases the firm’s COEC. This way of thought leads to the following hypothesis:

H1a: CSR-related executive compensation is positively related to the Cost of Equity Capital.

2.3. The Value-Enhancing View on CSR

Hypothesis 1a assumes there are possible negative consequences from compensating executives for CSR practises on the financial performance and subsequently on the COEC of a firm. However, there is a body of literature which finds a positive effect of CSR activities on the firm’s financial performance and firm value. Orlitzky et al. (2003) investigate the relation between Corporate Social Performance (CSP) and Corporate Financial Performance (CFP). They find that this relation is universally positive across different industries and contexts. Different effect sizes, “from highly positive to modestly positive”, are explained with several contingencies, e.g. firm reputation effects and CSP disclosure. The positive effect is explained by the developments of new capabilities, competencies and resources which manifest in the company culture, structure, HR and technology. New capabilities are developed because engaging in CSP, especially preventive CSP efforts, requires “significant employee

involvement, organization-wide coordination and a forward-thinking managerial style”

(Orlitzky et al., 2003, p.406; Shrivastava, 1995). This will in turn lead to better scanning skills, enhanced processes and information systems which will help the organization prepare for future changes. This leads to more efficient use of resources (Majumdar & Marcus, 2001). They also call upon reputation enhancement of the firm as a way through which CFP is positively influenced by CSP. Malik (2014) shows that CSR activities relate to better financial performance and higher firm value through market benefits (e.g. extended market share, increased customer loyalty), employee benefits (e.g. increased morale), operational benefits

(12)

11 (e.g. improved reputation and efficiency) and capital market benefits (e.g. increased markets return)

Hart & Ahuja (1996) investigated pollution prevention as a way of emission reduction and the effect which this had on operating performance on data from 1988 and 1989. They found that operating performance (measured as ROA, ROS and ROE) significantly benefitted from pollution prevention in following years. The authors argue that operating performance is enhanced by the lower costs of raw material, compliance, disposal and liabilities. Adjusting processes to make them more environmental friendly may lead to discovery of inefficiencies within these processes and subsequently improvement, leading to enhanced efficiency. Preventing pollution on a higher level (when firms are close to “zero pollution”) tends to get more expensive, however, the costs of reducing emission rarely exceeded the benefits. These results are applicable to firms with relatively poor environmental performance. El Ghoul et al. (2011) investigate 2.809 unique firms from 1992 to 2007 and find that the COEC is significantly lower for firms which perform well CSR-wise. According to the authors, this is because low ranking CSR firms have a higher perceived risk and a reduced investor base. They furthermore find that the “employee relation, environment policy and product strategy” dimensions of CSR influence the reduction in the COEC whilst “community relation, diversity and human rights” do not. Connors & Silva-Gao (2008) investigate the effect of environmental performance on a firm’s COEC and find a significant positive relation between emission and COEC. The positive relation between emission and COEC is due to the risk factor included in the COEC. This implies that investors think that potential costs (e.g. liabilities or compliance costs) might arise from the high emission compared to when emission is lower. This also holds when they control for financial performance (through ROE). King & Lenox (2002) find a significant negative relation between a firms emission and financial performance for 614 firms with 2.837 firm-year observations between 1991 and 1996, indicating that waste prevention results in better financial performance (measured as ROA) for the firm. Waste prevention may lead to (unexpected) innovation of de production process and encourages development of the worker’s skills, making the firm more efficient in their operations. A meta-analysis study by Margolis & Walsh (2003) of 127 studies regarding CSR and financial performance found that studies mostly point towards a positive relation between Corporate Social Performance and Corporate Financial Performance.

(13)

12 Nollet et al. (2016) and Trumpp & Guenther (2017) find evidence for both the positive and the negative relationship, implying a U-shaped relation between CSR investments and Corporate Financial Performance. Before a certain threshold of CSR investments is met, financial performance is negatively influenced, when that threshold is exceeded, financial performance is positively influenced.

Since CSR activities may have positive consequences on the firm’s financial performance and value, tying executive compensation with CSR-related measures and targets may lead the executives to use resources towards CSR purposes and in turn have a positive influence on financial performance and other financial variables. Beronne & Gomez-Meija (2009) also mention that rewarding social performance at the top may lead to higher firm value through institutional legitimacy. Legitimate firms can attract better employees (Buysee & Verbeke, 2003), are less likely to have legal sanctions, environmental remediation expenses and reduce chances of organizational failure (Singh et al., 1984). The reduced risk through enhanced financial performance, increased legitimacy and enhanced reputation may lead to a reduced COEC. This leads to the following hypothesis:

H1b: CSR-related executive compensation is negatively related to the Cost of Equity Capital.

2.4. Mediation through Financial Performance

Both H1a and H1b examine the direct effect of CSR-related executive compensation on the COEC while controlling for financial performance and several other variables. As formerly mentioned, CSR activities may have a negative or positive effect on financial performance. It may be possible that incentivizing these CSR activities by means of tying CSR-related measures and targets to executive pay has an effect on the financial performance of a firm through the CSR activities. Since the financial performance of a firm influences its COEC, this might be another way through which incentivizing executives to use resources for CSR purposes influences the COEC. The view which investigates the negative effects of CSR expects a decline in financial performance due to CSR and subsequently the COEC will rise. My expectations are that the same effects might be expected for CSR-related executive compensation since this may reinforce CSR activities (Hong et al, 2016; Russo & Harrison, 2005) and subsequent effects hereof. This leads to the following hypothesis.

(14)

13 H2a: CSR-related executive compensation is positively related to the Cost of Equity Capital through a decrease in financial performance.

The body of literature which investigates the positive effects of CSR such as the aforementioned reputation enhancement of the firm, legitimization of the firm and increased firm innovation expects the financial performance of the firm to increase due to the CSR activities. Since these CSR activities are incentivized by the inclusion of CSR-related executive compensation in the executive’s compensation structure, I hypothesize based on this view that incentivizing CSR by means of CSR-related executive compensation increases the financial performance and consequently lowers the COEC.

H2b: CSR-related executive compensation is negatively related to the Cost of Equity Capital through an increase in financial performance.

Figure 1 gives a graphical overview of the different hypotheses and the expected effect which the variables have on each other. Hypothesis 2a expects a negative relation between CSR related compensation and financial performance which results in an increase in the COEC. This is stated as a negative relation between financial performance and the COEC since a decline (increase) in financial performance is hypothesized to lead to an increase (decline) in the COEC in H2a (H2b).

+ (H1a) - (H1b)

- (H2a) - (H2a)

+ (H2b) - (H2b)

Figure 1: Graphical overview of the conceptual model

CSR-related executive compensation

Cost of Equity Capital

(15)

14

3. Methodology

The research approach applied to this study is archival research. Archival research is the most suitable approach since this helps to capture the several (financial) variables which are needed to test the hypotheses. To test the hypotheses, a random sample of 119 firms was selected from the Standard and Poor’s 500 index (S&P 500). Unavailability of the variables, which will be further discussed below, led to the fact that the final population of the S&P 500 index was 428 firms in this study, of which 119 were randomly selected to test the hypotheses. Random selection of the firms was used to exclude possible selection bias. The S&P 500 index has been used in multiple studies which investigate relating areas, e.g. emission reduction (Hart & Ahuja, 1996), social performance on the cost of capital (Sharfman & Fernando, 2008), the relation between corporate governance and environmental performance (Walls et al., 2012) and the effect of CSR on corporate governance and executive compensation (Hong et al., 2016). The S&P 500 index focusses on large firms which covers almost 80% of the available market capitalization of the US (McGraw Hill Financial, 2017). Hart & Ahuja (1996) also use the S&P 500 index in their research since evidence (Rice, 1993) suggests that environmental performance and emission levels differ greatly among these firms. Firms which are included within the S&P 500 index also have to submit a ‘proxy statement’ (DEF 14A form) which is an “Official notification to shareholders of matters to be brought to a vote”. This form includes a component which focusses on the executive compensation. Information regarding the build-up and the measures used within executive compensation will be used to test the hypotheses.

3.1. Cost of Equity Capital

Data has been retrieved from multiple sources and has been combined to test the hypotheses. The implied Cost of Equity has been used in this study as a proxy for the COEC (PEG_EAST_2). This has been calculated by using Easton’s (2004) model. Easton’s model to calculate the implied Cost of Equity has been used in multiple related studies as a proxy for the COEC (Dhaliwal et al., 2011; Connors & Silva-Gao, 2008; Chen at al., 2015; Harjoto & Jo, 2014). Botosan & Plumlee (2005) mention that estimates of expected return based on Easton’s (2004) model, dividend forecasts and Value Line target prices are superior to other formulas because they include variables related to risk proxies, beta and leverage in a

(16)

15 consistent and predictable manner (Connors & Silva-Gao, 2008). Easton’s (2004) formula is stated in equation (1).

𝑃

_𝑡

=

𝐹𝐸𝑃𝑆𝑡+2+𝑟𝑚𝑝𝑒𝑔𝐷𝑃𝑆𝑡+1−𝐹𝐸𝑃𝑆𝑡+1

𝑟_{𝑚𝑝𝑒𝑔}2 (1)

𝑊ℎ𝑒𝑟𝑒: 𝐹𝐸𝑃𝑆_𝑡+2 > 𝐹𝐸𝑃𝑆_𝑡+1> 0

𝐴𝑛𝑑: 𝐷𝑃𝑆_𝑡+1= 𝐷𝑃𝑆_𝑡

In Easton’s (2004) formula, FEPS stands for Forecasted Earnings Per Share. 𝐹𝐸𝑃𝑆_𝑡+1 and 𝐹𝐸𝑃𝑆_𝑡+2 respectively mean the Forecasted Earnings Per Share for one year ahead and the Forecasted Earnings Per Share for two years ahead. 𝑟2_{𝑚𝑝𝑒𝑔} stands for the Cost of Equity Capital. 𝑃_𝑡 stands for the share price at time t. 𝐷𝑃𝑆_𝑡+1 stands for the forecasted Dividend Per Share for one year ahead, however, in this formula 𝐷𝑃𝑆_𝑡+1 is equal to the dividend for year t, 𝐷𝑃𝑆_𝑡 (Dhaliwal et al., 2011; Easton, 2004). 𝐷𝑃𝑆₀ refers to the dividend per share from year t-1 (Dhaliwal et al., 20t-1t-1).

Following Dhaliwal et al. (2011), the 𝐹𝐸𝑃𝑆₁, 𝐹𝐸𝑃𝑆₂, 𝑃₀and 𝐷𝑃𝑆₁ were taken from the month of June. The forecasted earnings per share for one- (𝐹𝐸𝑃𝑆₁) and two-year (𝐹𝐸𝑃𝑆₂) ahead were taken from the month June of the year 2015 from the I/B/E/S database. The price of the shares (𝑃₀) was taken from the same day in June as the forecasts were published, this was collected from the CRSP database. As mentioned before, 𝐷𝑃𝑆₀ refers to the dividends per share from the year t-1, thus, 𝐷𝑃𝑆₀ was collected by using the Compustat database and selecting the fiscal year of 2014.

3.2. CSR-related Executive Compensation

The independent variable used in this study is CSR-executive compensation. Following the methodology of Hong et al. (2016), proxy statements of 2016 were examined for a random sample of 119 firms from the S&P 500 index. 2016 proxy statements describe the compensation, the compensation structure and the compensation process for named executive officers (NEO’s) for the fiscal year of 2015. These 2016 proxy statements review pre-established targets (pre-established in the early months of fiscal 2015) to calculate, analyse and discuss the NEO’s performance in fiscal 2015 to come to the compensation of the NEO’s. Thus an analysis between 2016 proxy statements (which describe 2015 performance based on

(17)

16 pre-established targets) and financial performance and COEC using financial data of fiscal 2015 seems to be appropriate.

The Compensation Discussion and Analysis section of the proxy statements describe the compensation and its structure in detail. This section of the proxy statements was used to find whether executive compensation included CSR dimensions. As mentioned previously, CSR may consist of social and environmental activities, for this reason, a distinction is made within the measurement of CSR-related executive compensation. The dimensions of CSR included in this study are the environmental dimension (CSR_ENV) and the social dimension (CSR_SOC). Splitting CSR in two dimensions might give more insight in the origin of the effects on COEC and financial performance. Following Hong et al.’s (2016) method for coding and the reasoning behind this, dummy variables were used to code whether CSR-based targets and measures were included in the executive’s compensation within the firm. This coding focussed on short-term variable compensation since individual performance goals and other non-financial targets are usually included in the short-term variable component. Long-term variable compensation mainly includes market-related variables such as shareholder return and vested stocks. If firms were found to compensate executives based on the social dimension of CSR, the variable ‘CSR_SOC’ was coded with a one, and coded zero if the executives were not compensated based on the social dimension of CSR. The most common examples of inclusion of the social aspect of CSR based compensation included “Safety, Diversity, Ethical Business Practises, (Corporate) Social Responsibility and Health”. Appendix 2 of Hong et al. (2016) was used as a reference for the coding process. The same principle applied to the environmental dimension of CSR. The variable ‘CSR_ENV’ was coded with a one if executives were compensated based on the environmental dimension of CSR and coded zero if the executives were not compensated based on the environmental dimension of CSR. Examples of environmental CSR-related measures which were found are “Environment(al Footprint reduction), Sustainability (goals), Emission Reduction and Green Energy”. An overarching dummy variable, ‘CSR’, was used to measure whether executives were compensated on either one of these dimensions. This way, the effect of CSR based compensation in total, as well as the effect of the social and environmental dimension can be measured separately against the COEC and financial performance of the firms.

(18)

17

3.3. Financial Performance

Hypothesis 2a and Hypothesis 2b include the mediating variable ‘financial performance’. Financial performance was measured using ‘Return on Assets’ (ROA) and ‘Return on Equity’ (ROE). ROA shows the return as a percentage of the total assets of a firm while ROE shows the return as a percentage of the stockholders’ equity. Thus, the difference between ROA and ROE is that ROA includes equity and debt while ROE includes only the equity component. Both variables show the financial performance of the firm and are used in multiple studies as a measure of financial performance (King & Lenox,2002; Nollet et al., 2015; Trumpp et al., 2016; Beronne & Gomez-Meija; 2009). Cochran and Wood (1984) mention that these variables capture the internal efficiency of the firm. These variables are subjected to the allocation of resources towards specific projects and reflect ‘internal decision-making

capabilities and managerial performance rather than external market responses to organizational (non-market) actions’ (Orlitzky et al., 2003; p.408). Thus the effect of the

decision of executives to allocate resources towards CSR activities based on the CSR-related compensation is likely to be measured by these two measures of financial performance more than another unit of measurement. ROA and ROE are calculated by gathering data in the Compustat database regarding the fiscal year 2015. The variables used to calculate this are ‘Total assets’, ‘Stockholders’ Equity’ and ‘Net Income (Loss)’.

3.4. Control Variables

One of the control variables within this study is the financial leverage (FIN_LEV), measured as interest bearing long-term debt divided by the total assets, also used in the study of Connors and Silva-Gao (2008) who calculated the COEC using several methods including Easton’s (2004) method. Modigliani & Miller (1958) mention that the expected rate of return on equity capital will increase when there is more debt within the firm´s capital structure. The interest bearing debt of the fiscal year of 2015 has been collected by means of the Compustat database. Missing interest bearing debt values of 51 firms have been hand-collected by investigating 2015 annual reports to increase the initial range of firms which could be selected for the random sample. Board size (BOARD_SIZE) and Board independence (BOARD_IND) are also included as control variables, since the board of executives has a big influence on decisions which could impact the COEC and the financial performance of a firm. Chen et al. (2013) mention that board independence influences CEO pay disparity and consequently influences the COEC by using the managerial power hypothesis. Following Chen et al.

(19)

18 (2013), Board size is also included as a control variable. Board independence and Board size are collected by means of the Asset4 database for the fiscal year 2015. This study also controls for the industry in which the firms operate (SIC_CODE). Industry codes have been gathered by using the Compustat database and are based on the first two digits of the Standard Industry Classification Code (SIC). This measure is included to control for industry risk which is not dependent on the capital structure of firms (Dhaliwal et al., 2011). Following Barnea & Rubin (2010) firm size (FIRM_SIZE) is included as a control variable since the size of the firm influences expected CSR activities. Effects of the size of the firms on the financial performance and COEC will be controlled by using this measure. Table 1 gives an overview of all the variables, their description and the data source from which they were collected.

To control for a possible lagged effect of CSR activities which may be initiated by the 2015 compensation structure which includes CSR-measures, an analysis may take place by means of using financial data from the fiscal year of 2016. Since not all firms which are present in the initial sample have submitted an annual report or submitted a 10-K form at the moment of writing, the size of this sample has declined to 94. The method how variables were gathered was identical for the lagged COEC and the lagged ROA and ROE. Control variables were also lagged to the fiscal year of 2016. Data regarding BOARD_SIZE and BOARD_IND were not fully available for S&P500 firms for fiscal 2016. For this reason, available firms have been adjusted whilst firms which did not have data available regarding these two variables remained at the same value as fiscal 2015.

(20)

19

Table 1: Variable descriptions and data source.

Variable: Description: Data source:

PEG_EAST_2

The implied Cost of Equity calculated by using Easton’s (2004) formula, used as a proxy for the Cost of Equity.

Compustat, I/B/E/S and the CRSP database

CSR

Dummy variable which equals one if a firm compensates executives on social or

environmental CSR targets in the short-term variable incentives and zero otherwise.

2016 proxy statements

CSR_ENV

Dummy variable which equals one if a firm compensates executives on environmental CSR targets in the short-term variable incentives and zero otherwise.

CSR_SOC

Dummy variable which equals one if a firm compensates executives on social CSR targets in the short-term variable incentives and zero otherwise.

ROA Return on Assets, Net Profit (Loss) divided by

Total Assets. Compustat

ROE Return on Equity, Net Profit (Loss) divided by

Stockholders’ Equity. Compustat

FIN_LEV Interest bearing long term debt divided by total

assets. Compustat

BOARD_SIZE The total number of board members in the board

at the end of the fiscal year. Asset4 BOARD_IND Percentage of independent board members as

reported by the company. Asset4

SIC_CODE

Industry code, using the first two digits of the Standard Industry Classification Code to classify the firms in an industry.

Compustat FIRM_SIZE Total assets as a proxy for firm size, shown in

(21)

20

4. Results

4.1. Descriptive Analysis

An initial exploratory analysis of the variables of fiscal 2015 and the lagged year of fiscal 2016 led to the fact that multiple extreme outliers were excluded from further analysis. These extreme outliers were rejected due to the fact that these led to unsuitable COEC, ROA and ROE. The original ROE for fiscal 2015 for example, had a mean of 23,87% with a standard deviation of 82,18%, a minimum of -485,66% and a maximum of 518,18%.Consequently, extreme outliers for fiscal 2015 and fiscal 2016 were identified and rejected for the following variables: “PEG_EAST_2, ROA and ROE” since these are key variables for the hypotheses. This led to a reduction in sample size for fiscal 2015 from 119 firms to 111 firms and for fiscal 2016 from 108 firms to 94 firms. Changes in the means of the CSR, CSR_ENV and CSR_SOC variables due to reduction in sample sizes were minimal (all less than 1% except for CSR_ENV for fiscal 2016, which declined from a mean of 0,120 to a mean of 0,106). Table 2 and 3 show the descriptive statistics for fiscal 2015 and fiscal 2016 respectively.

Table 2: Descriptive statistics for fiscal year 2015

Variable N Mean Median Std. Dev. Min Max

PEG_EAST_2 111 9,28% 9,28% 2,05% 4,47% 15,87% FIRM_SIZE 111 114.343 29.281 326.350 2.128 2.351.698 ROA 111 5,24% 4,8% 4,84% (9,64%) 18,40% ROE 111 15,73% 14,03% 12,64% (25,21%) 57,72% FIN_LEV 111 0,246 0,242 0,149 0,000 0,766 BOARD_SIZE 111 11,4 11 1,93 7 17 BOARD_IND 111 84,65 % 88% 9,39% 53,85% 94,12% CSR 111 0,414 0 0,495 0 1 CSR_ENV 111 0,108 0 0,312 0 1 CSR_SOC 111 0,405 0 0,493 0 1

Table 3: Descriptive statistics for fiscal year 2016 (1-year lag)

Variable N Mean Median Std. Dev. Min Max

PEG_EAST_2 94 9,39% 9,33% 2,19% 2,93% 14,36% FIRM_SIZE 94 133.650 31.645 371.292 2.238 2.490.972 ROA 94 5,38% 4,81% 4,18% (0,91%) 21,88% ROE 94 16,53% 13,70% 9,89% (9,13%) 48,04% FIN_LEV 94 0,222 0,218 0,134 0,000 0,520 BOARD_SIZE 94 11,4 11 1,96 7 16 BOARD_IND 94 85,50% 90% 8,40% 53,85% 94,12% CSR 94 0,415 0 0,495 0 1 CSR_ENV 94 0,106 0 0,310 0 1 CSR_SOC 94 0,404 0 0,493 0 1

(22)

21 Table 4 shows the results of an independent samples T-test between firms which do have CSR-related targets and measures within their compensation program and firms which do not have CSR-related targets and measures within their compensation program. The independent samples T-test examines whether there is a significant difference in means between two groups, in this case, firms with and without CSR based compensation components. The independent samples T-test has been conducted for all relevant variables for the hypotheses as well as the control variables. The left side of table 4 shows the independent T-test statistics for fiscal 2015 variables whilst the right side of the table shows the results of the independent T-test for fiscal 2016 financial variables.

Table 4: Independent samples T-test for 2015 and 2016.

Variables 2015 No CSR contract mean CSR contract mean T stat Variables 2016 No CSR contract mean CSR contract mean T stat PEG_EAST_2 9,14% 9,48% -0,846 PEG_EAST_2 8,96% 10,00% -2,323* FIRM_SIZE 62.643 187.397 -2,011* FIRM_SIZE 66.772 227.965 -2,112* ROA 6,00% 4,17% 1,985* ROA 6,13% 4,31% 2,117* ROE 15,96% 15,42% 0,220 ROE 17,3% 15,4% 0,905 FIN_LEV 0,253 0,236 0,597 FIN_LEV 0,229 0,212 0,598 BOARD_SIZE 11,2 11,7 -1,438 BOARD_SIZE 11,2 11,7 -1,228 BOARD_IND 84,06 85,47 -0,780 BOARD_IND 85,90% 84,94% 0,330 * Correlation is significant at the 0.05 level (2-tailed)

Table 4 shows the results for the overarching CSR variable which is denoted as a one if either CSR_ENV or CSR_SOC is also indicated as a one. Further investigation of unreported findings of independent samples T-tests for both CSR_ENV and CSR_SOC made it clear that the significances which are shown in table 4 are the result of the CSR_SOC variable. Tests on the CSR_SOC variable reveal the same significant differences as tests on the overarching CSR variable. There are no significant differences between the two sub-groups when CSR_ENV is used for their distinction. To investigate the implied correlations within table 4, a correlation analysis has been executed for both 2015 and 2016. The results of this analysis are shown in table 5 for the year 2015 and table 6 for the year 2016.

As can be seen in table 5, CSR and CSR_SOC are significantly related at the 5% level with FIRM_SIZE and ROA, implying that bigger firms are more likely to include a social aspect of CSR in the compensation of the executives. CSR and CSR_SOC are negatively related to ROA, implying that compensating executives for the social aspect of CSR decreases financial performance (in line with the reasoning of hypothesis 2a). Since CSR and CSR_SOC are significantly associated with ROA, but not with ROE, ROA will be used to test H2a and H2b.

(23)

22 ROA is also more significantly related to PEG_EAST_2 than ROE in fiscal 2015. Although table 5 shows some significance in regards to the effect of CSR-related measures on financial performance, this does not take in account the interrelated relationships between control, dependent and independent variables. Table 6 shows the correlation for 2016, which shows correlation at the 5% significance level between CSR(_SOC) and PEG_EAST_2.

Table 5: Correlation matrix for variables of fiscal year 2015.

PEG_ EAST_2 FIRM_ SIZE RO A RO E FIN_ LEV BO ARD _ SIZE BO ARD _ IND C SR C SR_ ENV C SR_ SO C PEG_ EAST_2 1 FIRM_ SIZE ,195** ,40 1 RO A -,218** ,022 -,229** ,016 1 RO E -,204** ,032 -,152 ,110 0,822*** 0,000 1 FIN_LEV -,217** ,022 -,220** ,020 0,125 0,192 ,223** ,018 1 BO ARD_ SIZE ,176* ,065 ,276*** ,003 -0,143 0,135 -,063 ,508 -,074 ,442 1 BO ARD_ IND ,133 ,166 ,133 ,162 -0,062 0,518 ,047 ,623 ,032 ,741 ,146 ,127 1 C SR ,081 ,399 ,189** ,047 -0,187** 0,050 -,021 ,826 -,057 ,552 ,136 ,153 ,074 ,437 1 C SR_ ENV ,099 ,300 -,092 ,339 -0,035 0,716 ,025 ,791 ,120 ,209 ,032 ,737 ,067 ,486 ,414*** ,000 1 C SR_SOC ,066 ,489 ,196** ,039 -0,193** 0,043 -,040 ,680 -,056 ,561 ,160 ,094 ,064 ,503 ,982*** ,000 ,363*** ,000 1

T he first line of each variable shows the Pearson’s correlation coefficient while the second line shows the two-tailed significance. * Correlation is significant at the 0.10 level (2-tailed)

** Correlation is significant at the 0.05 level (2-tailed) *** Correlation is significant at the 0.01 level (2-tailed)

(24)

23

Table 6: Correlation matrix for variables of fiscal year 2016.

PEG_ EAST_2 FIRM_ SIZE RO A RO E FIN_ LEV BO ARD _ SIZE BO ARD _ IND C SR C SR_ ENV C SR_ SO C PEG_ EAST_2 1 FIRM_ SIZE ,272*** ,008 1 RO A -,253** _,014 -,282*** _,006 1 RO E -,210** _,043 -,218** _,034 ,679*** _,000 1 FIN_LEV -,184* _,077 -,251** _,015 ,130 _,212 ,199* _,055 1 BO ARD_ SIZE ,249** ,016 ,265*** ,010 -,223** ,030 -,079 ,452 ,038 ,715 1 BO ARD_ IND ,027 ,794 ,143 ,170 -,130 ,213 ,054 ,603 ,070 ,504 ,135 ,193 1 C SR ,235** _0,022 ,215** _,037 -,216** _,037 -,094 _,368 -,062 _,552 ,135 _,194 -,056 _,590 1 C SR_ ENV ,042 ,689 -,097 ,351 ,044 ,675 ,102 ,326 ,099 ,341 ,052 ,617 ,026 ,801 ,410*** ,000 1 C SR_SOC ,239** ,020 ,223** ,030 -,226** ,028 -,,134 ,197 -,053 ,614 ,162 ,118 -,068 ,514 ,978*** ,000 ,349** ,001 1 T he first line of each variable shows the Pearson’s correlation coefficient while the second line shows the two -tailed significance.

* Correlation is significant at the 0.10 level (2-tailed) ** Correlation is significant at the 0.05 level (2-tailed) *** Correlation is significant at the 0.01 level (2-tailed)

4.2. Regression Analysis

H1a hypothesized a positive relation between CSR and PEG_EAST_2 while H1b hypothesized a negative relation between CSR and PEG_EAST_2. A linear regression has been conducted to test both H1a and H1b. Results from the linear regression are shown in table 7. A distinction has been made for fiscal 2015 (models 1-4) and for the lagged year of fiscal 2016 (models 5-8). All models show the regression coefficients, and the standard errors within the brackets. Models (1) and (5) show the results for the control variables while using PEG_EAST_2 as dependent variable. Model (2) and (6) show the results when including the overall CSR variable in the linear regression. Model (3) and (7) show the results for CSR_ENV while model (4) and (8) show the results including CSR_SOC. Separate models have been included for CSR_ENV and CSR_SOC so a clear distinction can be made both for these CSR dimensions. ROA has been taken as a proxy of financial performance since this has been shown to be significantly associated with PEG_EAST_2 as well as the social dimension of CSR in both fiscal 2015 and fiscal 2016. In addition, ROA correlates more significantly with PEG_EAST_2 in both years than ROE (see table 5 and 6).

(25)

24

Table 7: Linear regression analysis for fiscal 2015 and fiscal 2016

Fiscal year 2015 2016 Dependent variable PEG_EAST_2 PEG_EAST_2 Model (1) (2) (3) (4) (5) (6) (7) (8) CSR -0,076 (0,412) 0,676 (0,470) CSR_ENV 0,662 (0,654) 0,505 (0,753) CSR_SOC -0,147 (0,411) 0,662 (0,472) FIRM_SIZE 5,764 (0,000) 5,969 (0,000) 6,038 (0,000) 6,154 (0,000) 9,034 (0,000) 7,296 (0,000) 9,293 (0,000) 7,327 (0,000) ROA -0,079* (0,043) -0,081* (0,044) -0,071 (0,044) -0,083* (0,044) -0,085 (0,058) -0,065 (0,060) -0,082 (0,059) -0,066 (0,059) FIN_LEV -2,453* (1,305) -2,456* (1,311) -2,590* (1,312) -2,457* (1,311) -2,171 (1,715) -2,116 (1,705) -2,201 (1,721) -2,157 (1,705) BOARD_ SIZE 0,092 (0,105) 0,092 (0,105) 0,095 (0,105) 0,094 (0,105) 0,202* (0,119) 0,197* (0,118) 0,200* (0,119) 0,191 (0,118) BOARD_IND 0,021 (0,20) 0,021 (0,021) 0,020 (0,021) 0,021 (0,021) -0,008 (0,026) -0,003 (0,027) -0,008 (0,027) -0,003 (0,027) SIC_CODE -0,010 (0,012) -0,011 (0,013) -0,006 (0,013) -0,011 (0,012) -0,002 (0,014) 0,003 (0,015) 0,001 (0,015) 0,002 (0,014) INTERCEPT 7,871*** (2,260) 7,933*** (2,296) 7,612*** (2,274) 7,978*** (2,290) 8,707*** (2,786) 7,741*** (2,849) 8,532*** (2,807) 7,819*** (2,842) n 111 111 111 111 94 94 94 94 F-statistic 2,483** 2,113** 2,275** 2,219** 2,547** 2,505** 2,233** 2,488** 𝑹𝟐 _0,125 _0,126 _0,134 _0,126 _0,149 _0,169 _0,154 _0,168 T he table reports the regression coefficient with the standard error in the brackets

* Significant at the 0.10 level ** Significant at the 0.05 level *** Significant at the 0.01 level

Table 7 shows no significant relationship between CSR, CSR_ENV or CSR_SO C on PEG_EAST_2, indicated in models (2), (3) and (4) for fiscal 2015 with betas of -0,076, 0,662 and -0,147 and p-values of 0,854, 0,314 and 0,721 respectively. Models (6), (7) and (8) show the same result for the lagged year of fiscal 2016 with betas of 0,676, 0,505 and 0,662 and p-values of 0,154, 0,504 and 0,164 respectively. These results indicate that neither H1a nor H1b can be accepted. All models for fiscal 2015 show a significant relationship between FIN_LEV and PEG_EAST_2 and all models except model (3) show significant relations between ROA and PEG_EAST_2. No models for the lagged year of fiscal 2016 show these relations. This implies weak evidence for the fact that ROA and FIN_LEV influence COEC both in a negative manner. A higher ROA will result in a lower COEC and a higher financial leverage will also lead to a lower COEC, the latter of these two results is counterintuitive.

4.3. Mediation Analysis

H2a hypothesized a mediation model in which there is a negative relation between CSR and financial performance, which leads to a rise in the COEC. H2b hypothesized a mediation

(26)

25 model in which there is a positive relation between CSR and financial performance, which leads to a decline in the COEC. A mediation analysis has been conducted to test both H2a and H2b; its results are shown in table 8. As stated before, ROA has been taken as financial performance since this has shown correlation with CSR as well as with PEG_EAST_2. Table 8 shows the three steps of the mediation analysis. The first step investigates the direct relation between CSR variables and PEG_EAST_2. The second step investigates the so-called ‘Path A’, this is the relation between the CSR variables as independent variable and ROA as the dependent variable. Step 3 investigates the so-called ‘Path B’, this is the relation between ROA as independent variable and PEG_EAST_2 as dependent variable while controlling for the CSR variables. A full mediation is realized when path A is significant and path B is significant while there is no direct relation between the CSR variables and PEG_EAST_2. This is shown in table 8 when the CSR variables are significant in step 2 and ROA is significant in step 3 while the CSR variables are not significant in step 3. Partial mediation is realized when path A and path B are significant while the relation between the CSR variables and PEG_EAST_2 is also significant in step 3. Table 8 includes several panels to examine the effect of the different CSR variables. Panel A investigates the relations while using the overarching CSR variable while panel B and panel C investigate the relations for CSR_ENV and CSR_SOC respectively. The effect of the different dimensions of CSR can be examined more thoroughly in this manner. Table 8 shows the result of the mediation analysis for the fiscal year of 2015. Results of the mediation analysis of the lagged year of fiscal 2016 are not tabulated in this study.

Table 8 shows that in none of the cases, there is a direct link between CSR variables and the COEC (step 1). Step 2 shows path A, which tested the regression of the CSR variables on ROA. This was significant in all cases, with p-values of 0,016, 0,059 and 0,019 for CSR, CSR_ENV and CSR_SOC respectively. Both the environmental and the social dimension of CSR have a negative effect on ROA when they are linked to executive compensation, shown by the negative beta in step 2. Step 3 shows path B, which tested the regression of ROA on the COEC, whilst controlling for the CSR variables. This was significant while controlling for CSR and CSR_SOC with p-values of 0,071 and 0,065 respectively. ROA is negatively related to the COEC when controlling for CSR(_SOC), indicating that a decline in ROA relates to a rise in COEC. ROA does not influence the COEC in a significant manner when controlling for CSR_ENV. Concluding, CSR, CSR_ENV and CSR_SOC are negatively related to ROA and ROA is negatively related to PEG_EAST_2 when controlling for CSR and CSR_SOC.

(27)

26 These findings suggest that CSR-related compensation lowers the financial performance (ROA), and this reduction in financial performance increases the COEC (PEG_EAST_2) since this is a negative relation. This is in accordance with hypothesis 2a.

Table 8: Mediation analysis for fiscal 2015, n=111

Dependent Variable PEG_EAST_2 (a) ROA PEG_EAST_2

(Panel A) Step 1: Step 2: Step 3:

Independent variable CSR 0,100 (0,405) -2,171** (0,885) -0,076 (0,412) ROA -0,081* (0,044) Control variables FIRM_SIZE 6,698 (0,000) -9,016 (0,000) 5,969 (0,000) FIN_LEV -2,604* (1,323) 1,826 (2,891) -2,456* (1,311) BOARD_SIZE 0,119 (0,106) -0,334 (0,231) 0,092 (0,105) BOARD_IND 0,023 (0,021) -0,021 (0,045) 0,021 (0,021) SIC_CODE -0,002 (0,012) -0,110*** (0,026) -0,011 (0,013) 𝑹𝟐 _0,097 _0,223 _0,126 F-statistic 1,871* 4,981*** 2,113** (Panel B) Independent variable CSR_ENV 0,857 (0,647) -2,753* (1,442) 0,662 (0,654) ROA -0,071 (0,044) Control variables FIRM_SIZE 7,196 (0,000) -1,635 (0,000) 6,038 (0,000) FIN_LEV -2,769** (1,318) 2,527 (2,933) -2,590* (1,312) BOARD_SIZE 0,121 (0,105) -0,361 (0,233) 0,095 (0,105) BOARD_IND 0,022 (0,021) -0,020 (0,046) 0,020 (0,021) SIC_CODE 0,002 (0,012) -0,113*** (0,027) -0,006 (0,013) 𝑹𝟐 _0,112 _0,206 _0,134 F-statistic 2,183** 4,499*** 2,275** (Panel C) Independent variable CSR_SOC 0,027 (0,405) -2,106** (0,887) -0,147 (0,411) ROA -0,083* (0,044) Control variables FIRM_SIZE 6,926 (0,000) -9,351 (0,000) 6,154 (0,000) FIN_LEV -2,611* (1,324) 1,871 (2,896) -2,457* (1,311) BOARD_SIZE 0,120 (0,106) -0,319 (0,231) 0,094 (0,105) BOARD_IND 0,023 (0,021) -0,023 (0,045) 0,021 (0,021) SIC_CODE -0,002 (0,012) -0,108*** (0,026) -0,011 (0,012) 𝑹𝟐 _0,097 _0,220 _0,126 F-statistic 1,860* 4,902*** 2,129**

T he table reports the regression coefficient with the standard error in the brackets * Significant at the 0.10 level

** Significant at the 0.05 level *** Significant at the 0.01 level

An untabulated mediation analysis of the variables of fiscal 2016, for the one-year lag, indicates a direct positive relation between CSR(_SOC) and PEG_EAST_2 when not controlling for ROA, with a beta of 0,796 and 0,783 and a p-value of 0,085 and 0,092 respectively. Analysis of path A, which tested the regression of the CSR variables led to the

(28)

27 fact that CSR and CSR_SOC were significantly negative associated with ROA. The respective betas were -1,862 and -1,838 and the p-values were 0,026 and 0,029 respectively. This was not significant for CSR_ENV, which had a p- value of 0,433. These results indicate that compensating executives based on the social dimension of CSR lowers financial performance (ROA) in the year of compensation itself and the lagged year. Path B was found to be not significant in any of the untabulated analyses including the variables of fiscal 2016. No lagged results were found which indicated a relation between ROA and COEC while controlling for CSR(_SOC), suggesting that the statistical strength of path B may be rather low.

The analyses of 2015 and 2016 show that in both cases the presence of CSR and CSR_SOC was significant with a decline in the ROA. Indicating that compensating executives for the social dimension of CSR lead to activities which reduce the ROA.

Summarizing the results, H1a and H1b were tested by means of a linear regression analysis, shown in table 7. This table shows that none of the CSR variables were significantly related to the dependent COEC variable “PEG_EAST_2”. This is the case for both the fiscal year of 2015 and the lagged year of fiscal 2016. H2a and H2b sought the relation between the CSR variables and the COEC through financial performance, which is measured for both years using ROA. H2a and H2b have been tested by using a mediation analysis, which is tabulated in table 8. Table 8 shows that H2a is weakly supported in 2015, CSR and CSR_SOC led to a reduction in ROA, and since ROA is negatively related to the COEC, this reduction leads to an increase in the COEC. In all of the fiscal 2015 cases however, CSR, CSR_ENV and CSR_SOC were significant and negatively related to ROA, implying that including CSR-related targets and measures within executive compensation leads to a reduction in ROA. This was also found within fiscal 2016 for CSR and CSR_SOC. ROA was negatively related to PEG_EAST_2 while controlling for CSR and CSR_SOC, but not CSR_ENV in fiscal 2015. No relation was found between ROA and PEG_EAST_2 within fiscal 2016. This implies weak evidence that the COEC increases due to a reduction in the financial performance of a firm.

CSR-related Executive Compensation and the Cost of Equity Capital – An Empirical Analysis

Master’s Thesis A&C, Controlling.