What is the effect of CSR firm-fit on the relationship between CSR-performance and employee motivation?

(1)

Faculty of Economics and Business

MSc. in Business Administration – Strategy Track

Master thesis

What is the effect of CSR firm-fit on the relationship between

CSR-performance and employee motivation?

Name: Romy ten Nijenhuis Student number: 10001018

Thesis Supervisor: Panikos Georgallis Date: 22-06-2018 - Final Version.

(2)

1

Statement of originality

This document is written by Student Romy ten Nijenhuis 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.

(3)

2

ABSTRACT

In response to pressure from society, firms are showing more interest in the effects of CSR on firm performance. This study is focused on the effect of CSR on a specific form of firm performance, namely; employee motivation. Less research has been done on how different types of CSR activities can influence employee motivation, this could explain some of the contradicting results. For this reason, this study examines what the effect of CSR firm-fit is on

the relationship between CSR performance and employee motivation. Building on the social

identity theory, I expect employees to identify more with firms that have aligned their CSR engagement with their core business. Firstly, it is therefore expected that CSR performance has a positive relationship with employee motivation. Secondly, it is expected that engaging in CSR activities, in line with the core business, will have a positive moderating effect on the relationship between CSR performance and employee motivation. The sample consists of 50 publicly listed firms involved in different industries and originated from different countries. Panel data was collected for a period of five years per firm, from 2012 until 2016. Results showed a negative influence of CSR performance on employee motivation and a negative moderating effect of CSR firm-fit, both contradicting to the expectations. However, results do show a positive influence of CSR firm-fit on employee motivation. A first step was made to study the possible influence of CSR firm-fit on employee motivation, however the

(5)

4

1. Introduction

"Corporate social responsibility is a hard-edged business decision. Not because it is

nice to do or because people are forcing us to do it, …, but because it is good for our business. We need a constant flow of talented people. More and more people are looking at

companies and ask themselves if this is an organization whose values they share. This is a

hard-edged business issue.” – Niall FitzGerald, Former CEO, Unilever.

Niall FitzGerald might have been one of the first CEO’s to recognize the importance

of CSR engagement for employees, when this interview was given in 2003 (Elliott, 2003). Currently, society is successfully putting pressure on firms to engage in Corporate Social Responsibility (CSR), which creates difficulty in finding a firm that is not trying to engage in CSR (Utting, 2005). Most firms agree (or give in), to the importance of looking at the triple bottom line; People, planet, and profit, or; society, the environment and the economy (Marcus & Fremeth, 2009). Porter and Kramer (2006) already argued that CSR engagement should be seen by managers as shared value because it will stimulate financial growth and thus give strategic value. This is confirmed by the numbers in the yearly CEO survey from PWC; 64% out of 1.409 CEOs, said they are increasing their investment in CSR in the next years and 52% believed that “creating value for wider stakeholders helps profitability” (PWC, 2016).

The influence of CSR has been widely researched and results have shown that among other things, CSR can indirectly influence reputation, consumer loyalty, stakeholder relations and financial performance (Aguinis & Glavas, 2012). However, results of the influence of CSR on overall firm performance have been contradicting (Aguinis & Glavas, 2012; Surroca, Tribó, & Waddock, 2010). This is in disagreement with the current practical understanding of CEO’s and strategy as is quoted above. It is therefore important to focus on the mechanics by which CSR translates to firm performance. Employee motivation is such an aspect that has been researched to some extend (Gond, El-Akremi, Igalens, & Swaen, 2010) but research has

(6)

5 increased since 2010 and still shows contradicting results (Aguinis & Glavas, 2012; Farooq, Payaud, Merunka, & Valette-Florence, 2014; Hansen, Dunford, Boss, Boss, & Angermeier, 2011). This could be caused by the broad concept of CSR and employee motivation. Employee motivation and the reduction of turnover rate can both reduce costs and thereby have a positive influence on firm performance (Aguinis & Glavas, 2012; Hansen et al., 2011). Firm performance can also be influenced by CSR through organizational commitment and reduced recruitment costs (Greening & Turban, 2000). Another study has found that CSR could lead to the acceptance of lower wages (Burbano, 2016). A cautious assumption can be made that when wages are lower, this could increase a firm's financial performance since the bulk of a company’s wage bill is represented in the pay to non-executives (Larkin, Pierce, & Gino, 2012). From this view, it can be concluded that the motivation and commitment of employees is a mediating factor between CSR and firm performance, since reducing turnover rate and acceptance of lower wages can reduce costs, and factors such as higher work motivation can increase profits. However, the mechanics behind these relationships are not yet fully understood. Therefore, more attention needs to be given to the mechanics to fully understand the relationship between CSR and employee motivation.

One of the ways in which CSR can influence employees, can be appointed to the social identity theory. Employees identify with the firm they work for, therefore part of their motivation can be explained by the identity and actions of a firm (Stuart, 2002). Following this theory, CSR could have a bigger influence on employees when the CSR engagement is in line with the firm. When an employee identifies him or herself with a firm the employee can thus identify more with the CSR engagement. This theory is supported by marketing research, where studies have shown that consumers are more motivated when the CSR engagement of a firm, is in line with the firms’ core business, the CSR firm-fit (Einwiller, Fedorikhin, Johnson, & Kamins, 2006). As Utting (2005) said, it is difficult to find

(7)

6 (corporate) firms that are not trying to engage in CSR. In a society that puts pressure on all firms to engage in CSR, the motives to engage in CSR can be less intrinsic, and thus the type of CSR can be more random (Jong & Meer, 2017). This might cause the CSR engagement and firm not to be in line, and thus not to be in line with the identification of the employee. This misalignment might explain the current contradicting results of the influence of CSR on employee motivation. Therefore, in this thesis, I will strive to answer the following question (Figure 1): What is the effect of CSR company-fit on the relationship between CSR-performance and employee motivation?

This research will give two theoretical contributions to the growing field of CSR research. Firstly, the main focus is to strive for more clarity on the relationship between CSR engagement and employee motivation and commitment, by focusing on the moderating influence of CSR firm-fit. Some research has been done on the CSR firm-fit, however, the relationship with employee motivation has not yet been explored (Liu, Liston-Heyes, & Ko, 2010). Therefore, I will combine existing theories on CSR firm-fit on the effects of CSR engagement and motivation from the marketing research field to the strategy research field. Secondly, it will add to the existing research available regarding the influence of CSR on employee motivation. Research available on the effect of a firm’s CSR performance is mainly on an organizational level, related to measures such as financial performance. Less research is available on the effect of CSR on an individual level, such as employee motivation (Aguinis & Glavas, 2012). By looking at the effect of a firm’s CSR performance on employee motivation, I integrate the organizational and individual level, resulting in a multilevel analysis on the effects of CSR performance.

This thesis will also provide a practical contribution. It will provide new insights for managers on how to form the CSR strategy in their business portfolio. Alligning CSR with

(8)

7 the firm, and thus creating a CSR firm-fit could stimulate employee motivation, reduce costs and overall might even improve firm performance.

In the next chapter, an overview of the existing literature is given. This is followed by a visualized model of the theoretical framework (Figure 1).

2. Literature review

First, I will give a short introduction on CSR. Secondly, I will give an overview of existing literature studying the relationship between CSR and employee motivation. This is followed by a description of the connection between this relationship and CSR firm-fit, and a definition of this concept. Following the literature review, a visualized model of the theoretical framework is given.

2.1 Corporate Social Responsibility

In the last decade research on Corporate Social Responsibility (CSR) has developed quickly. In this paper, I will use the definition of CSR from Aguinis where CSR is defined as: “context-specific organizational actions and policies that take into account stakeholders’

expectations and the triple bottom line of economic, social, and environmental performance.”

(Aguinis, 2011, p. 855). The triple bottom line directs back at the different ways CSR has been categorized. The best-known categorization of CSR is the pyramid of Carroll (1991). The pyramid consists of 4 types of CSR responsibilities; philanthropic, ethical, legal and economic (Carroll, 1991). Based on this, Kourula and Halme (2008) introduced a more action-orientated categorization, which allows better possibilities for assessment. They introduced three dimensions, the first dimension is Philanthropy which mainly involves charity and voluntarism. The second dimension is Corporate responsibility integration, which involves “conducting existing business operations more responsibly” (Kourula & Halme, 2008). The third dimension is Corporate responsibility innovation, in which the main focus is

(9)

8 on solving social and environmental problems or developing new business models (Kourula & Halme, 2008). In this thesis, I will focus on all three categories.

In the business strategy research field, firms have mainly been interested in the effect of CSR on firm performance. However, research of known academics on this topic is inconsistent. Aguinis & Glavis (2012) examined an overview of CSR research of the last couple of decades and found contradicting results. On one side there are several economists that argue that the business of business is business (Friedman, 2007). This is supported by evidence found that CSR has a negative influence on firm performance (Aguinis & Glavas, 2012; Brammer & Millington, 2008). On the other hand, there is also evidence found claiming that CSR can have a positive influence on firm performance (Aguinis & Glavas, 2012). Which is in line with Freeman (1999) who argued that firms should not just satisfy shareholders, but also the stakeholders, thereby including the employees. This inconsistency has been linked to several moderating and or mediating factors. For example, the correlation of CSR with R&D (McWilliams & Siegel, 2001), stakeholder relations, firm size, government regulations or resources (Aguinis & Glavas, 2012). Therefore, to get a better idea of the influence of CSR on firm performance, the concept of firm performance needs to be specified to see how it is influenced by CSR and when this leads to increased performance. Employee motivation is an important element of firm performance, and multiple studies have shown a positive relationship between CSR and employee motivation (Aguinis & Glavas, 2012; Burbano, 2016; Farooq et al., 2014; Hansen et al., 2011). I will focus on the relationship between CSR and employee motivation, therefore in the next section the concept of employee motivation will be defined.

2.2 Employee Motivation and Commitment

Barney (1991) argued that based on the resource-based theory, employees are a specific resource, and can be a sustainable competitive advantage for a firm. Therefore, it is

(10)

9 important for a firm to take care of its employees and keep them motivated. Employee motivation is a very broad concept that can be divided into different sub-concepts. It can be defined as “an individual's degree of willingness to exert and maintain an effort towards

organizational goals” (Franco, Bennett, & Kanfer, 2002, p. 1255). In the past employee

motivation and commitment were seen as separate concepts. However, Locke (1997) created a model were both concepts were integrated as employee motivation. The integrated model of employee motivation and commitment has been accepted widely and influences the way employee motivation can be measured (Meyer, Becker, & Vandenberghe, 2004; Nohria, Groysberg, & Lee, 2008).

In this study, the model of Locke (1997) will be used by defining employee motivation as a combination of employee satisfaction and commitment. Commitment can be divided into different forms, but for the purpose of this study I will look specifically at “affective commitment” (Allen & Meyer, 1990). Affective commitment is defined as “an

affective or emotional attachment to the organization such that the strongly committed

individual identifies with, is involved in, and enjoys membership in, the organization”(Allen

& Meyer, 1990, p. 2). The extent of employee commitment is thus dependent on the identification with the firm. This definition is derived from the social identity theory. To understand the relationship between CSR and employee motivation it is necessary to divagate and zoom in on the social identity theory in the next section.

2.3 Social Identity Theory and CSR

According to the social identity theory “people are more likely to identify with an

organization when they perceive its identity to be enduring, distinctive, and capable of

enhancing their self-esteem” (Bhattacharya & Sen, 2004, p. 15). Organizational commitment

fits into this framework, since it is also dependent on the identification with the firm, and thus influenced by the behavior of the firm or organization itself. This phenomenon is based on

(11)

10 the fact that identification of an employee with a firm is partially answered by the question

“Who am I” (Ashforth & Mael, 1989, p. 21). If an employee feels that the identity of the firm

represents, or is similar to his or her own identity, organizational commitment will be higher. Reason for this is, is when an employee identifies with a firm, they depend more on the firm. Which consequently causes the employee to perform better because they are dependent on the firm to do well, and therefore more motivated to do well themselves (Dutton, Dukerich, & Harquail, 1994). Thus, creating more organizational commitment.

The relationship between social identity theory and organizational commitment and employee motivation has been confirmed by several scholars (Brammer, Millington, & Rayton, 2007; Dutton et al., 1994; Ellemers, Kortekaas, & Ouwerkerk, 1999). Carmeli, Gilat and Waldman (2007) explored the influence of CSR on the intensity of identification with a firm. A survey was used to assess the influence of perceived CSR engagement and perceived financial performance on the organizational identification. Carmeli et al. (2007) found that perceived CSR engagement is positively related with organizational identification. The results of this study show the importance of the influence of CSR engagement on organizational identification. Carmeli et al. (2007) also found that organizational identification is positively related to employee performance. Based on these results and the definition of organizational commitment, it can be concluded that CSR engagement influences the organizational identification of employees, and therefore effects employee motivation. In the next section evidence for this relation will be discussed.

2.4 CSR and Employee Motivation and Commitment

Employee commitment is influenced by the behavior of the firm itself. This implies that certain behavior of a firm, like CSR engagement, can influence employee commitment and therefore the motivation through the effect of organizational identification. This is

(12)

11 illustrated in a highly cited article of Ambec and Lanoie (2008). They found that CSR can increase revenue or reduce costs by reducing the cost of illnesses, absenteeism, recruitment, and turnover (Ambec & Lanoie, 2008). Recently more researchers have found that CSR can increase organizational commitment and thus reduce turnover (Farooq et al., 2014; Hansen et al., 2011). The influence of CSR on employee commitment has been researched for internal and external CSR separately (Farooq et al., 2014). Internal CSR engagement is directly related to the employee (Cohen, 2017). For example; recycling in the office, long-term employment security or encouraging employees through training. External CSR engagement of a firm is not about the actual job that is performed, but aimed towards other stakeholders (Delfgaauw & Dur, 2007). Examples of external CSR engagement are donating to other parties or causes, setting up neighborhood initiatives or setting up free (or low-cost) education programs for specific social groups (Hameed, Riaz, Arain, & Farooq, 2016). Some empirical evidence is available on how having internal CSR, like having social responsible work, can as a non-pecuniary incentive, motivate employees intrinsically (Larkin et al., 2012). Based on the social identity theory this can be expected since internal CSR engagement will be perceived as a firm to be responsible in terms of employee treatment. This will strengthen the identification of an employee with the firm, and thus lead to higher employee commitment.

Other research also indicates that external CSR engagement of a firm can influence employees (Brammer et al., 2007; Burbano, 2016). Burbano (2016) examined the effects of employer CSR engagement on the demands of workers on wages. The results showed that applicants that were informed of the employer’s social responsibility were willing to accept lower wages and thus have a higher intrinsic motivation. Brammer et al. (2007) examined the effect of external CSR engagement on employee commitment and satisfaction in the social identity framework and also found a positive relationship. These studies confirm the

(13)

12 introductory statement of Niall FitzGerald; an employee as a stakeholder is important to increase firm performance (Elliott, 2003).

Based on the social identity theory which explains how employees are affected by the actions of a firm it is assumed that CSR engagement influences employee motivation. CSR engagement has the purpose of doing good and considering all stakeholders, instead of just the shareholders. This leads to the assumption that employees will positively identify themselves with a firm that engages in CSR. Consequently, when an employee identifies with a firm, they depend more on the firm. Which causes the employee to perform better, because they are dependent on the firm to do well, and therefore more motivated to do well themselves (Brammer et al., 2007; Burbano, 2016; Farooq et al., 2014; Hansen et al., 2011; Larkin et al., 2012). Before continuing to the effect of CSR firm-fit, this I will first verify the relationship between CSR performance and employee motivation. Therefore, the following baseline hypothesis is formulated:

H1: External CSR engagement has a positive relationship with employee motivation

2.5 CSR Firm-Fit

One of the fields that focusses of CSR on an individual level, is the effect of the fit of a firm’s CSR on consumer behavior. The fit of the CSR of a firm is defined as: “the

perceived fit between the corporate mission and the social initiative” (Becker-Olsen,

Cudmore, & Hill, 2006, p. 46). The correlation between the CSR firm-fit and the consumer was believed to be negative at the beginning of the emergence of this research field (Drumwright, 1996). Firms chose unrelated causes when engaging in CSR, because they feared that choosing related causes was seen as opportunism by consumers. According to Drumwright consumers see a firm with high CSR firm-fit as exploiting the cause for their own purposes and thus not as engaging in CSR for the right reasons (Drumwright, 1996).

(14)

13 However more recently most researchers have found contradicting evidence that firms that engage in CSR that is perceived to be related to their corporate mission will be seen as appropriate and can reinforce a firm’s market position (Becker-Olsen et al., 2006). Consumers see a company that chooses a cause with a high fit as an expert in their business because they are overall aligned (Hoeffler & Keller, 2002). Additionally, researchers found that when CSR engagement of a firm is perceived as a low fit with the firm's mission, it is seen as insincere, hypocritical and can lead to negative assessment by consumers (Becker-Olsen et al., 2006; Wagner, Lutz, & Weitz, 2009). Öberseder et al. (2014) confirmed the more recent viewpoint and recently developed a scale for firms to see if their CSR engagement corresponds with the perceived CSR firm-fit of the consumer. To illustrate,

Home-Depot engaging in CSR related to homelessness is a high fit with their core business. Likewise, Ford Motor Company engaging in CSR on vehicle safety for children is also a high fit. However, Ford Motor Company stepping up for homelessness is seen as a low-fit and by some therefore even as selfish or profit-driven CSR (Becker-Olsen et al., 2006).

The relationship between CSR firm-fit and consumer behavior can be partly explained by the social identity theory (Bhattacharya & Sen, 2004). As discussed above, employees identify with a firm. In the same way, consumers also identify with a firm (Bhattacharya & Sen, 2003; Scott & Lane, 2000). When a consumer identifies with a firm and a new CSR activity of the firm is not in line with the core business, and thus identity, of the firm the strength of identification of the consumer might decrease. Having a high CSR firm-fit can be seen as more enduring, or consistent, accordingly, this will cause consumers to increase the identification with the firm. Since organizational identification has been connected to employee motivation and organizational commitment (Ashforth & Mael, 1989), one might wonder if the perceived fit of CSR engagement also influences the effect of CSR engagement on employee behavior. Although extensive research has been carried out on the perceived fit

(15)

14 of CSR on consumer behavior, limited research exists on the effect of perceived fit of CSR engagement on employee behavior (Liu et al., 2010).

Liu et al. (2010) examined the effect of the perceived fit of internal CSR engagement on employee behavior. The social identity theory established that internal CSR engagement will strengthen the identification of an employee with the firm, and thus lead to higher employee commitment. Focusing on internal CSR, the same effect can be expected with a perceived CSR firm-fit, since internal CSR is by definition related to the firm's core business, and thus has a firm-fit. In contrast, since external CSR engagement does not have to be in line with the firm's core business, the effect of CSR firm-fit on the relation between external CSR engagement and employee motivation is not obvious and will therefore be discussed below.

Expanding on previous research I expect that employees will have a stronger organizational identification with firms that have a high CSR firm-fit. An employee has a stronger sense of identification with a firm if the firm engages in CSR in general. Based on the social identity framework it is expected that when a firm is more aligned and shows appropriate behavior, such as having a high CSR firm-fit, employees have a stronger organizational identification. When the organizational identification is stronger, the employee depends more on the firm. This leads to the employee to perform better, since they are dependent on the firm to do well, and therefore more motivated to do well themselves. Therefore, it is expected that when the CSR firm-fit is high, the relationship between external CSR engagement on employee motivation is stronger and vice versa. Thus, the second hypothesis is proposed as following:

H2: The relationship between external CSR engagement and employee motivation is positively moderated by the perceived CSR firm-fit.

(16)

15

Combining the two hypotheses I will answer the main question of this research: What is the effect of CSR firm-fit on the relationship between CSR performance and employee motivation.

2.6 Conceptual framework visualized

A visualized model of the conceptual framework can be found below.

This paper will not focus on the relationship between employee motivation and firm performance. However, to illustrate the importance of the research field this relationship was included in the model.

Figure 1: Model - What is the effect of CSR firm-fit on the relationship between CSR performance and employee motivation? Black lines represent the relationships focused on in this research paper

(17)

16

3. Research Design

In this chapter, I will first give a description of the process of data collection for the research question. Secondly, I will present a definition of all variables and measures.

3.1 Data Collection

To answer the research question on what the effect of CSR firm-fit is on the relationship between CSR engagement and employee motivation, a quantitative research method was applied. I used a combination of data analysis and database research to collect all data. The data on CSR engagement and employee motivation is secondary data, taken from the Asset4 database using DataStream. The decision to take this data from a database was based on the complexity of the concept of employee motivation. Employee motivation is a concept that needs to be measured in a longitudinal study. The data contains only publicly listed firms (PLF’s), due to accessibility but also due to the public access to extra information such as the CSR engagement plans. I believe that the combination of the public availability of this data and the large sample will increase the external validity and reliability.

The starting point to collect data from Asset4 was the active universe list. This list contains 6022 firms in the world that are currently providing data for this database. Data was collected for latest available, five consecutive years, 2012-2016. Of the 6022 firms in the first sample, CSR-scores were available for 5125 firms between 2012-2016. Of these 5125 firms, 1490 firms also had data on employee motivation for at least one year in this period of time. Due to time restrictions on manually collecting the data for the moderator CSR firm-fit, I made a random selection of 50 firms from the 1490 firms left for the analysis of this research paper (See Appendix 1). The firms originated from a large variety of countries, increasing the generalizability (See Appendix 2). The 50 firms over 5 years will provide 250 observations.

(18)

17 3.2 Variables

3.2.1 Independent Variable

The independent variable is CSR performance. CSR performance has been measured in the past by other scholars looking at three categories from the Asset4 database: environmental score, social score and corporate governance score (Barnett, Hartmann, & Salomon, 2017; Cheng, Ioannou, & Serafeim, 2014; Eccles, Ioannou, & Serafeim, 2014). An average of these three defined the CSR performance of a firm. However, in this study, only a measure on external CSR engagement is needed. Corporate governance is a measurement that is defined by internal CSR, such as gender diversity in the board or CEO compensation and by standardized processes such as stakeholder voting (Harford, Mansi, & Maxwell, 2008). Therefore, I only used the environmental and social score to create an average score that defines the external CSR performance of a firm. Since all scores are equally weighted the

CSR score was created using the following formula:

𝐶𝑆𝑅 𝑆𝑐𝑜𝑟𝑒 = (𝐸𝑛𝑣𝑖𝑟𝑜𝑛𝑚𝑒𝑛𝑡𝑎𝑙 𝑆𝑐𝑜𝑟𝑒+𝑆𝑜𝑐𝑖𝑎𝑙 𝑆𝑐𝑜𝑟𝑒)

2 (“Thomson Reuters ESG Scores,” 2018). 3.2.2 Dependent Variable

The dependent variable is employee motivation. Employee motivation was measured in two constructs; employee commitment and overall satisfaction of the employee on the firm following the definition and model of Locke (1997). The first construct, employee commitment, was collected as employee turnover rate. Turnover has proven as a good measure of turnover intention and therefore employee motivation (Hancock, Allen, Bosco, McDaniel, & Pierce, 2013; Meyer et al., 2004). To improve readability of the results and create a hypothesis with a positive relationship between CSR score and Turnover Rate, Turnover Rate was reverse coded. Therefore, from here on, it will be presented as Retention

(19)

18 The second construct of employee motivation, the satisfaction of the employee, was measured using the satisfaction score. Job satisfaction can be defined as “the pleasurable

emotional state resulting from the appraisal of one’s job as achieving or facilitating one’s job values.” (E. A. Locke, 1969, p. 316).

Together employee motivation is therefore represented in Retention Rate and Satisfaction Score of the employee, both extracted from the Asset4 database. External CSR engagement is expected to increase employee overall satisfaction and employee retention, thus having a positive relationship.

3.2.3 Moderating Variable

The CSR firm-fit was operationalized as the similarity between the main business of the firm and the firms’ CSR engagement. First, I obtained the main business of the firm from the “About us” section on the shareholder's page of the official firm website. Secondly, I collected information on CSR engagement of the firms by downloading all separate CSR reports per year when available. If a firm or year, did not have a separate CSR report, annual reports were downloaded containing similar information.

To evaluate which CSR activities should be included in the analysis of firm-fit, first I made a distinction by extracting only external CSR activities. Thereafter the external CSR engagement was selected using three criteria; 1) it had to be voluntarily and not forced by law, 2) the firm itself had to participate in the activity and 3) the activity must be concrete (de Jong & van der Meer, 2017). Of all external CSR activities that were found in the

sustainability or annual firm reports, I only took the main CSR activities into account. Main activities include big projects the firm creates awareness for in the news and within the firm. Smaller activities which had less elaboration in the CSR report, such as random donations or a sponsorship of a local schools’ soccer team, were not included. This was partly done to reduce complexity, but also because employees are likely to not be aware of all CSR

(20)

19 activities. The main activities should be better known by most employees in comparison to smaller CSR activities and therefore have more influence on the employee motivation (Raub & Blunschi, 2014).

After documenting per year per firm what the core business and the main CSR activities were, I made the first CSR firm-fit analysis. I categorized each observation in one out of three categories; no fit, low fit, and high fit. CSR engagement was analyzed per year separately, creating five CSR firm-fit values per firm, one for every year. To measure the reliability of this analysis, I set up a document with guidelines on how to categorize the CSR

firm-fit. The guidelines include a clear definition of CSR, the mission and main business of

each firm and what the CSR engagement of the firm incorporates. Next this was tested in small batches of 3 firms over 5 years by two people and the guidelines were perfected (Appendix 3). After perfecting the guidelines, two people were asked to each categorize 10 firms. Thereafter the categorization of the data by myself and by the others was analyzed on inter-rater reliability. The results of the inter-rater reliability were 17 out of 20, an 85% overlap in ratings. Therefore, it was concluded that the CSR firm-fit index was reliable enough to include in the data analysis.

3.2.4 Control Variables

A few other variables were taken into account to control for their influence on CSR performance and employee motivation. Firstly profit, revenue and industry grouping will all be included in the analysis. The amount of profit can give a firm more options in the way of investing in CSR (Waddock & Graves, 1997). Revenue was chosen as a variable to represent the size of a firm. Due to the bigger visibility of larger firms, they tend to invest more in CSR engagement compared to smaller firms, especially external CSR activities (Baumann-Pauly, Wickert, Spence, & Scherer, 2013). The last firm characteristic, industry grouping, has been proven to have influence on CSR. Controversial industries such as oil and gas or the tobacco

(21)

20 industry are less social responsible than customer good industries (Hillman, Keim, & Schuler, 2004). Dummy variables were computed for industry grouping, where “Basic materials” was chosen as a baseline.

Secondly, the percentage of women employees will be controlled for. Women are more positively influenced by the CSR engagement of the firm they work for than men (Brammer et al., 2007; Burbano, 2016). It is suggested that this could be caused by the gap between men and woman that is still present on the work floor (Brammer et al., 2007). The fact that women are confronted with this gap, causes them to have a stronger preference for social responsibility which leads to higher motivation and commitment. I will control for this variable by using the percentage of women employees from the Asset4 database.

Thirdly, country of origin will be controlled for. When obtaining the data for the moderator CSR firm-fit, it was noticed that firms in countries with a lot of natural disasters engage in external CSR aimed at help for these natural disasters, and thus not necessarily have a firm-fit. The World Risk Report (2016) scores all countries on the level of risk to experience a natural disaster and on how big the impact would be, using percentages divided into four categories. Using the 2016 report, all countries were recoded into a dummy variable

RiskCountry appointing “0” to high-risk countries and “1” to low-risk countries. The

separation is based on the threshold of 5,5% risk index, separating the two high-risk categories from the two low-risk categories.

Lastly, CSR reporting will be taken into account as a control variable. CSR firm-fit has been analyzed using CSR or sustainability reports if available, hereafter simplified as CSR reports. However not all firms published CSR reports. In the case where for a certain firm or one year within a firm, a CSR report was not available, annual reports were used. Since CSR reports can be more comprehensive, this could have influenced the CSR firm-fit.

(22)

21 Therefore, I will control for this variable to see if the difference in reporting affected the results. To add this variable to the analysis, the variable CSR reports was also dummy coded. Appointing “0” if separate CSR reports were available to obtain the firm-fit data, appointing “1” if the data was obtained from annual reports.

In the following chapter the method of analysis will be presented, followed by the descriptive information on the data and the results of the analyses.

4. Methods and Results

4.1 Methods

IBM’s SPSS was used for cleaning the data and summarizing the descriptives. Firstly, the data was cleaned for analysis. To test for errors, a frequencies test was executed, but no errors were found. The assumptions for normality, linearity, and homoscedasticity were also met. Secondly, a missing value analysis using univariate statistics was done to detect any missing values (see appendix 5). The analysis showed missing values for dependent variable

Retention Rate and Satisfaction score. It also showed missing values for the control variable

(CV) percentage of women. If less than 5% of items from a variable are missing, they can be replaced by the mean without significantly influencing the data (Field, 2013, p. 216). For the control variable percentage of women only 4% of the values were missing (Appendix 5). Therefore, these missing values were allowed to be replaced by the mean.

The dependent variables Retention Rate and satisfaction score showed a higher percentage of missing values (respectively 10,8% and 22,4%), therefore a missing value analysis was done to decide if the values could be replaced by the mean. The analysis showed non-significant results, indicating that the values are missing at random and should not simply be replaced by the mean. Therefore, linear interpolation was used on both dependent variables. Per firm, the missing values were interpolated if they had surrounding values of the

(23)

22 same firm a year earlier or later. Values missing in the first or last years were not replaced using extrapolation to ensure reliability. Observations with both dependent variables missing were removed, resulting in a dataset of 242 observations. Observations with only one

dependent variable missing (Satisfaction Score N=25, Retention Rate N = 12), were excluded pairwise.

Lastly, an analysis was done to detect outliers. Using the Z-score for all variables to detect univariate outliers. Only for the dependent variable Retention Rate outliers were found (Z-score >3.29, N=3). A multivariate analysis of outliers using Mahalanobis Distance was done as well to detect unusual scores by combining all independent variables. This analysis detected for both Satisfaction Score and Retention Rate, four additional observations that were classified as outliers (p <0.001), bringing the total to seven outliers. Five of the seven outliers were of the same firm “Lotte Chemical”. All seven observations were removed, resulting in a final dataset of 235 observations including 49 different firms.

Since panel data is collected, the degree of non-independence of the firms was analyzed to decide on what statistical model should be used. First, a Durbin-Watson test was done to test for autocorrelation, followed by an intra-class correlation to identify the amount of variance in employee motivation that is attributed to the clustering of the firms. Results showed that the observations are characterized by autocorrelation.1_{Since there is evidence of}

autocorrelation, the difference between and within firms need to be taken into account. Therefore, I chose panel data analysis as a good method for this model. Stata statistical software was used to run the model for both hypotheses.

1_{Results of the Durbin- Watson test showed .975 for Retention Rate and .771 for Satisfaction Score. Both}

variables were not between 1.5 and 2.5, thereby showing a sign of autocorrelation. Results of the intra-class correlation (ICC) showed an ICC score of .719 (71,9%) for Retention Rate and an ICC score of .588 (58,8%) for Satisfaction Score. Both showing a value higher than 0.1, confirming that the variance is attributed by the clustering of the firms.

(24)

23

Given there are two dependent variables, the following models were created to test hypothesis 1:

H1a: RetentionRate i = 𝛽0 + 𝛽1𝐶𝑆𝑅Scorei + 𝛽2Revenuei + 𝛽3Profiti + 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦i+

𝛽5CSRreporti + 𝛽6PercWomeni+ 𝛽7RiskCountryi+ ɛi

H1b: SatisfactionScorei= 𝛽0 + 𝛽1𝐶𝑆𝑅Scorei + 𝛽2Revenuei + 𝛽3Profiti + 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦i+

𝛽5CSRreporti + 𝛽6PercWomeni+ 𝛽7RiskCountryi+ ɛi

Where, CSRreport is the CSR reporting, equal to 1 if a CSR report was issued and 0 if an annual report had to be used. Industry is divided into 10 dummy variables. PercWomen is the percentage of women employees in a firm. Finally, Revenue, Profit, RiskCountry and

CSRscore stand for revenue, profit, Risk of the Country and CSR score.

A Hausman test was conducted to test which panel analysis, a fixed or random effects model, should be used for analysis. Results showed a random effects model should be used for both Retention Rate and Satisfaction Score (χ2(6) = p > 0.05).

The second hypothesis aimed to see the moderating effect of CSR firm-fit on the relationship between external CSR engagement and employee motivation. Therefor,e the following models were created to test hypothesis 2:

H2a: RetentionRatei = 𝛽0 + 𝛽1𝐶𝑆𝑅Scorei + 𝛽2Revenuei + 𝛽3Profiti + 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦i+

𝛽5CSRreporti + 𝛽6PercWomeni+ 𝛽7RiskCountryi+ 𝛽8CSRFirmFiti +

(𝛽1CSRScorei * 𝛽8CSRFirmFiti) + ɛi

H2b: SatisfactionScorei= 𝛽0 + 𝛽1𝐶𝑆𝑅Scorei + 𝛽2Revenuei + 𝛽3Profiti + 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦i+

𝛽5CSRreporti + 𝛽6PercWomeni+ 𝛽7RiskCountryi+ 𝛽8CSRFirmFit +

(25)

24 Before performing both analyses to test the hypotheses, first all descriptive statistics were obtained to check if all assumptions are met. This will be discussed in the next section.

4.2 Descriptive Statistics

Table 1 provides all summary statistics for all variables and Pearson correlations and

p-values from the two-tailed test, using bi-variate correlation.

N Mean Std. Deviation 1 2 3 4 5 6 7 8 9 1. CSR score 235 81.5229 16.91510 1 . . . . 2. Satisfaction score 211 75.1550 11.23352 -.020 1 . . . . 3. Retention Rate 223 87.6425 7.29838 -.065 .117 1 . . . . 4. CSR firm-fit 235 1.21 0.694 .427* .035 -.163* 1 . . . . . 5. CSR reporting 235 0.30 0.460 -.212** .001 .022 -.265** 1 . . . . 6. Perc. of Women 235 38.1862 16.26086 -.122 -.059 -.308** .004 .146* 1 . . . 7. Revenue1 ₂₃₅ _17054.92 _115918.55 _-.016 _.049 _.154* _-.027 _-.082 _-.219** ₁ _. _. 8. Profit1 ₂₃₅ _11484.10 _64899.70 _-.011 _.042 _.141* _-.001 _-.064 _-.141* _.790** ₁ _. 9. RiskCountry 235 0.78 .413 -.261** -.013 -.396** .009 .032 .015 -.246** -.259** 1 10.Ind. Basic Materials 235 0.10 0.298 .182 ** _-.016 _.130 _.025 _.095 _-.400** _.336** _.241** _.069 11. Ind. Consumer goods 235 0.1277 0.33442 .232** _.110 _-.065 _.327** _-.113 _-.135* _-.051 _-.056 _.201** 12. Ind. Consumer Serv. 235 0.04 0.202 -.049 -.213** _-.224** _-.003 _.274** _.152* _-.029 _-.035 _.111 13. Ind. Financials 235 0.32 0.467 -.257 ** _.062 _-.219** _-.140* _.166* _.690** _-.085 _-.057 _-.038 14. Ind. Health Care 235 0.02 0.145 .077 -.004 .028 -.130 * _-.097 _.029 _-.021 _-.023 _.078 15. Ind. Industrials 235 0.11 0.320 .049 -.151 * _-.017 _-.108 _-.092 _-.153* _-.048 _-.050 _.028

16. Ind. Oil &

Gas 235 0.06 0.245 .031 .123 .213 ** _-.204** _-.134* _-.184** _-.005 _-.034 _-.074 17. Ind. Technology 235 0.06 0.245 -.226 ** _.035 _.015 _-.079 _.018 _-.222** _-.038 _-.045 _.137* 18. Ind. Telecom 235 0.09 0.280 .065 -.088 .071 .260 ** _-.201** _.088 _-.023 _.043 _-.209** 19. Ind. Utilities 235 0.06 0.245 .006 .072 .257 ** _-.003 _.018 _-.324** _-.010 _.028 _-.285**

Table 1: Descriptive statistics and correlation matrix.

* Correlation is significant at the 0.05 level (2-tailed). ** Correlation is significant at the 0.01 level (2-tailed). 1_{In Dollars times 10.000}

(26)

25 Notable is the lack of significant correlation between CSR score and Satisfaction Score or

Retention Score. However, by doing a multiple regression analysis this will be further

investigated. The moderator CSR firm-fit does show a significant negative correlation with

Retention Score (r = -.163, p < 0.05).

Looking at the control variables, it is noticeable to that profit shows a highly positive significant correlation with revenue (r = .790, p < 0.01). Correlations with a value above 0.7 can indicate multicollinearity (Field, 2013, p. 325). To make sure the regression analysis will be valid I therefore had to remove profit as a control variable. Although some other predictors are also significantly related, they do not show a correlation of above 0.7, giving a first

indication that there is no multicollinearity in the other variables. However, to make sure there is no subtle form of multicollinearity a VIF analysis was done for both satisfaction score and Retention Rate, using linear regression. Levels of VIF are acceptable when values are lower than 10 and tolerance levels are higher than 0.2. A VIF analysis including “Profit” showed a VIF level of 6.253 with a tolerance level of 0.168. After removing Profit levels of VIF ranged between 1.091 and 2.761 with tolerance levels between .362 and .917. The results of the VIF analysis show that the assumption of multicollinearity is met and no other

variables should be excluded.

In the next section I will discuss the results of the analysis done to check for the hypotheses.

4.3 Hypothesis Analysis and Results

Firstly, the results of the first hypothesis are presented for both dependent variables (Table 2). Secondly, the results for the second hypothesis are presented for both dependent variables (Table 3). Lastly, I will discuss the overall meaning for the effect of CSR firm-fit on

(27)

26 the relationship between CSR-performance and employee motivation as a combination of both variables. The results are followed up by a robustness section.

First the effect of the control variables on Retention Rate was analyzed in model 1 (Table 2). Together the control variables explain 33% of the variance on Retention Rate between the observations. The results also show that the control variables explain little of the variance within firms (R2 = 0.014)2, which is expected since most of the control variables are equal over years within the firm (industry and country at risk). However, they explain almost half of the variance between firms (R2 _{= 0.438). This signals that factors such as originating}

from a country at risk (β = -6.580, p < 0.01) and the availability of CSR reporting (β = 0.049, p < 0.05) has a negative influence on Retention Rate. In other words, it increases the turnover rate of employees.

To test the first hypothesis, the predictor CSR Score is added in model 2. The results show that CSR significantly predicts the variance in Retention Rate (β = -.105, p < 0.01). Surprisingly, this effect is negative, indicating that a higher CSR score, accounts for a higher employee turnover. A result opposite from H1a, and therefore H1a is rejected. When

analyzing the R-square results, it can be concluded that CSR score within firms explains 6,5% extra of the variance in Retention Rate (R2 = 0.065). Therefore, it can be concluded that CSR score negatively influences the variance in employee turnover, only on a small scale within firms. In the robustness section, results of a fixed effects analysis will control if the assumptions made based on results of the R-Square within firms are correct.

For the dependent variable Satisfaction Score, first the effect of the control variables was analyzed in model 3 (Table 2). Together the control variables explained only 12.9% of the variance on Satisfaction Score between the observations, and no significant effects of the

2_{A fact to keep in mind when comparing the R-squared values, is that the dependent variables are about human}

behavior. It is expected that R-square values to predict human behavior automatically have values lower than 50%, since humans can be hard to predict (Moksony, 1999).

(28)

27 control variables on Satisfaction Score were found. To test H1b, the predictor CSR Score was added in model 4. No significant effect was found. The results of the R-square test also do not show an increase in variance that can be explained. Most importantly, the model fit is not significant, indicating that the proposed relationship between CSR Score and Satisfaction

score is not proven and therefore H1b is rejected.

Hypothesis 1

DV: Retention Rate DV: Satisfaction Score

Model 1 Model 2 Model 3 Model 4

Control variables H1a Control variables H1b

Predictors β P-value β P-value β P-value β P-value

Control Variables

Perc of Women -0.022 0.799 0.013 0.885 -0.032 0.822 -0.028 0.846

Revenue1 _-_66.2 _0.920 _-97.6 _0.885 _6.07 _0.600 _6.01 _0.607

CSR reporting 2.095 0.049* 1.539 0.149 1.768 0.374 1.699 0.401 Country at Risk -6.580 0.004** -5.254 0.03* 1.209 0.764 1.367 0.743 Ind. Consumer goods -2.757 0.472 -3.529 0.370 5.378 0.422 5.289 0.436 Ind. Consumer Serv. -9.954 0.072+ _-12.387 _0.031* _-9.228 _0.334 _-9.523 _0.329

Ind. Financials -4.963 0.254 -7.934 0.081+ _3.856 _0.602 _3.497 _0.648

Ind. Health Care 0.638 0.923 -0.673 0.921 2.550 0.824 2.395 0.837 Ind. Industrials -3.739 0.328 -5.119 0.194 0.278 0.967 0.128 0.985 Ind. Oil & Gas 2.411 0.586 1.373 0.763 8.351 0.283 8.220 0.297 Ind. Technology -1.383 0.750 -4.371 0.340 3.246 0.675 2.853 0.724 Ind. Telecomm -2.801 0.547 -3.880 0.416 0.354 0.965 0.204 0.980 Ind. Utilities 1.766 0.702 1.463 0.757 5.125 0.521 5.080 0.529 Independent variables CSR Score -0.105 0.005** -0.013 0.846 Constant 95.842 0.000 103.857 0.000 72.442 0.000 73.472 0.000 R2 _within _0.014 _0.079 _0.003 _0.003 R2 _between _0.438 _0.390 _0.129 _0.130 R2 _overall _0.329 _0.301 _0.107 _0.107 Model Fit χ 2_{(13) = 29.92,} p<0.05 χ2_{(14) = 36.61,} p<0.001 χ2_{(13) = 6.14,} (p = 0.941) χ2_{(14) = 6.04,} (p = 0.965)

Table 2: Results of hypothesis 1: The effect of CSR engagement on Employee motivation

Retention Rate: N observed = 223, N groups = 49. Satisfaction Score: N observed = 211, N Groups = 49 1_{Divided by 10^10}

(29)

28 Hypothesis 2a predicted that CSR firm-fit has a positive influence on the relationship between CSR Score and Retention Rate. Therefore, the variable “firm-fit” was first added to model 5 as an independent variable and secondly as an interaction with CSR-Score in model 6 (Table 3). The results show that as an independent variable CSR firm-fit does not influence

Retention Rate. There is also no moderating effect found of CSR Firm-fit on the relationship

between CSR Score and Retention Rate. The results also show no significant changes in R-square overall, within or between firms. Therefore, hypothesis 2a was rejected. Despite the insignificant results of the hypotheses, the chi-squared test determined that the model on

Retention Rate is a good fit. This indicates that the variables are dependent on each other.

Although no direct effect was found of CSR Score on Satisfaction Score, the second hypothesis was also tested for Satisfaction Score. Hypothesis 2b predicted that the CSR

firm-fit has a positive influence on the relationship between CSR Score and Satisfaction Score.

Therefore, the variable “firm-fit” was first added to model 7 as an independent variable and secondly as an interaction with CSR-Score in model 8 (Table 3). The results show that as an independent variable CSR firm-fit does not influence Satisfaction Score. Model 7 is also not significant and therefore not a good fit for the proposed relationship. However, in model 8 a positive significant effect is found for CSR firm-fit (β = 22.128, p < 0.01) and a negative significant effect for the moderating effect of CSR Firm-fit on the relationship between CSR

Score and Satisfaction Score (β = -0.231, p < 0.01). When looking at the R-Square results, a cautious assumption can be made that model 8 explains 5.6% variance of Satisfaction Score within firms compared to model 7, and thus the effect is mainly expected within firms (R2 ₌

0.079). These results could mean that within firms, CSR Score indeed positively effects employee Satisfaction, but only when there is a high CSR firm-fit. The moderating effect however shows that, when the CSR firm-fit increases over the years within a firm, this

(30)

29 reduces the effect of CSR-Score on Satisfaction Score3_{. H2b expected CSR firm-fit to have an}

increasing effect on the relationship, therefore H2b is rejected as well. More importantly the model fit results show that even though model 8 is a better fit, it is not significant, and thus not a good fit for the proposed hypotheses.

Hypothesis 2

DV: Retention Rate DV: Satisfaction Score

Model 5 Model 6 Model 7 Model 8

H1a with firm-fit H2a H1b with firm-fit H2b

Predictors β P-value β P-value β P-value β P-value

Control Variables

Perc. of Women 0.014 0.874 0.015 0.863 -0.018 0.898 0.007 0.963

Revenue1 _-1.16 _0.862 _-1.23 _0.856 _6.93 _0.557 _6.75 _0.571

CSR reporting 1.323 0.222 1.261 0.250 2.372 0.247 2.797 0.165 Country at Risk -5.418 0.024* -5.500 0.024* 1.894 0.653 3.118 0.467 Ind. Consumer goods -3.183 0.414 -3.275 0.407 4.203 0.541 4.432 0.524 Ind. Consumer Serv. -12.204 0.031* -12.269 0.033* -10.365 0.293 -10.887 0.275 Ind. Financials -7.979 0.077+ _-8.037 _0.078+ _3.314 _0.668 _2.283 _0.769

Ind. Health Care -1.295 0.848 -1.294 0.850 4.195 0.722 2.876 0.810 Ind. Industrials -5.288 0.175 -5.277 0.182 0.741 0.915 -0.726 0.918 Ind. Oil & Gas 0.865 0.848 0.822 0.858 9.883 0.218 9.951 0.221 Ind. Technology -4.308 0.342 -4.454 0.333 2.660 0.744 5.508 0.507 Ind. Telecomm. -3.612 0.444 -3.722 0.436 -0.868 0.916 -1.882 0.821 Ind. Utilities 1.404 0.764 1.384 0.770 5.424 0.507 5.000 0.546 Independent variables CSR Score -0.093 0.016* -0.113 0.089+ _-0.053 _0.462 _0.341 _0.023* CSR Firm-Fit -0.776 0.248 -1.676 0.525 2.424 0.066+ _{22.128 0.001**} Interaction Moderator CSR Score*CSR Firm-Fit 0.011 0.723 -0.231 0.003** Constant 103.967 0.000 104.739 0.000 72.870 0.000 58.301 0.000 R2 _within _0.080 _0.081 _0.023 _0.079* R2 _between _0.403 _0.402 _0.131 _0.120 R2 _overall _0.311 _0.310 _0.117 _0.118 Model Fit χ 2_{(15) = 38.43,} p<0.001 χ2_{(16) = 37.96,} p<0.01 χ2_{(15) = 9.30,} (p = 0.861) χ2_{(16) = 18.22,} (p = 0.311)

Table 3: Results of hypothesis 2: The influence of CSR firm-fit on the relationship between CSR engagement and employee motivation

Retention Rate: N observed = 223, N groups = 49. Satisfaction Score: N observed = 211, N Groups = 49 1_{Divided by 10^10}

**p < 0.01; *p < 0.05; +_{p < 0.1}

3_{In the robustness section, results of a fixed effects analysis will control if the assumptions made based on}

(31)

30 To conclude, CSR Score does not have a positive effect on employee motivation. Results show that it does not have an effect on Satisfaction Score and that there might even be a negative effect on Retention Rate within firms over a five-year time period. Furthermore,

CSR firm-fit does not increase the effect of CSR Score on employee motivation. Disregarding

the model fit, CSR firm-fit might have a negative effect on the effect of CSR Score on

Satisfaction Score within firms over a five-year time period. This will be examined in the

next section using a fixed effects model. Therefore, all hypotheses are rejected.

4.4 Additional Robustness Tests

Five extra analyses were run to control for the measurement method and the control variables, since some unexpected results were found. Firstly, I examined if the amount of control variables and dummies could have had an influence. This was done running the same analyses again, but each time excluding one different control variable. Results remained similar as above and therefore it was concluded that the dummies and amount of control variables did not influence the model.

Secondly, year was added as a control variable. Since no influence of time on the model was expected, and the use of too many dummy variables can influence the variance explained by the model, Year was not included in the original analyses. Results only show a small difference4_{for model 1, where the control variables were checked with the dependent}

variable Retention Rate. Since all other results are similar as the main model, it was concluded that time did not influence the results.

Thirdly additional to the random effect test, a fixed effect test was run to rule out the impact of choice of test. Results showed that R-square between score, decreased for all models below 0.05. However, this is an expected decrease since all control variables equal

4_{A significant negative effect was found for Year 2015(β = -2.20,}_{p < 0.05), concluding that when CSR score is}

(32)

31 within firms were removed (Industry and Country of Risk), resulting in less variance on differences between firms. The conclusions drawn on the variance within firms were checked by comparing the R-Squared within results. The results of the fixed effects analysis also showed an increase of R-Square within variance between model 7 and model 85. These results confirm the conclusion drawn that within firms, CSR Score indeed positively effects employee Satisfaction, but only when there is a high CSR firm-fit. All other results also remained similar to the random effect test. Therefore, it was concluded that the R-square within scores were interpreted correctly and the choice (based on the Hausmann test) for a random effects model was correct.

Certain industries, like consumer service have a significant influence on the

dependent variable Retention Rate. Therefore, fourthly, an extra analysis was done to control for the effect of industry as a moderator. Industry was added as a moderator to the main relationship between CSR score and employee motivation, and as an extra moderator in the complete model next to CSR firm-fit. In both models, no significant effect was found of industry as a moderator and all other results remained similar. Thus it was concluded that there is no moderating effect of industry.

Lastly, an extra analysis was done to see if there was a delay in time on the influence of CSR and CSR firm-fit on employee motivation. The effect of this is expected to be minimal due to the measuring method. CSR score and employee motivation scores are obtained in the database from annual reports obtained at the end of each year. The numbers on CSR score are based on activities over a whole year, as well as the turnover rate and satisfaction score are mostly obtained at the beginning of the following year rating the

previous year. However, to avoid discussion on the chance of delay in influence on employee

5_{The results for R-Square within when doing a fixed effects model analysis are 0.0274 for model 7 and 0.084}

for model 8. The results show similar explained variance within firms compared to the random effects model (0.5% or less increase per model).

(33)

32 motivation, an extra analysis was run. A time lag of one year was used in the dependent variable, excluding 50 observations. One difference that was found was the lack of a

significant influence of CSR score on Retention Rate. This indicates that the CSR score has a direct influence on Retention Rate, but that there is no effect of CSR score on Retention Rate from a previous year. All other results were similar to the main analysis, therefore it was concluded the main model was chosen correctly.

5. Discussion and Conclusions

The purpose of this research was to expand the current literature and elucidate some of the previously contradictory results found on the effect of CSR on firm performance

contradicting (Aguinis & Glavas, 2012; Surroca et al., 2010). In this study I therefore focused on employee motivation and the effect the CSR firm-fit might have. Through various hypotheses the question tried to be answered in this study was; What is the effect of CSR

firm-fit on the relationship between CSR-performance and employee motivation?

The first hypothesis stated that CSR score positively effects employee motivation, wherein employee motivation is measured as Retention Rate and Satisfaction Score. In this

study no support was found for this proposed effect. However, a negative effect was found of

CSR Score on Retention Rate, indicating a higher CSR Score would indicate a higher turnover

rate. The second hypothesis stated that CSR firm-fit has a positive effect on the relationship between CSR score and employee motivation. This hypothesis was also not supported by the results found. However, a negative effect was found of CSR firm-fit on the relationship between CSR Score and Satisfaction Score. This result is both unexpected according to the literature consulted, but also since no relationship was found in the first hypothesis. The lack of results and the contradicting results in the current study could indicate that other

(34)

33

relationships exist or could be due to limitations in the research design chosen. Both will be discussed in the following sections.

5.1 Theoretical Implications

The first theoretical contribution this study aimed for, was to give more clarity on the relationship between CSR and employee motivation and commitment. Even though no support was found for the first hypothesis, some interesting contradicting results were found. Results showed that a higher CSR performance increased the turnover rate. Based on the social identity theory it is assumed that employees will positively identify themselves with a firm that engages in CSR. Consequently, when an employee identifies with a firm, they depend more on the firm. Which causes the employee to perform better, because they are dependent on the firm to do well, and therefore more motivated to do well themselves

(Brammer et al., 2007; Burbano, 2016; Farooq et al., 2014; Hansen et al., 2011; Larkin et al., 2012). However, in this study I did not find the expected results. One explanation could be that there is no direct effect of CSR performance on employee motivation. CSR might not have the magnitude of influence that was expected on social identity, and therefore

motivation. To ratio, other factors such as pay, training and future perspectives might play a bigger role in the variance of employee motivation (Ramlall, 2004). However, since several scholars have found confirming results, it could also be caused by the constructs that were chosen to measure employee motivation. Burbano (2016) for example, chose to measure employee motivation in the willingness to be paid less. This will be discussed further in the section on suggestions for future research.

Against expectation, the results also showed that a higher CSR score increases the turnover rate when measured over time within firms. A possible explanation could be that firms with higher turnover rates try improving other aspects first, and only try improving CSR performance as a last resort. The solution might then come too late to decrease the

(35)

34 turnover rate. This could indicate that there is a reversed causality; firms with higher turnover rate might increase their CSR performance to solve their “problem”. This should be further investigated in future research. Another possibility is a lack of communication. When

improving CSR with high costs but a lack of explanation, an employee could think that a firm is taking a risk and “wasting” its money. Resulting in the employee changing jobs to have more security. Both possible explanations need further investigation and are discussed in the last section.

The second theoretical contribution this study aimed for was the influence of CSR firm-fit. Some research has been done on the CSR firm-fit, however the relationship with employee motivation had not yet been explored (Liu et al., 2010). When employees see that a firm is more aligned and shows appropriate behavior, such as having a high CSR firm-fit, it was expected that they will have a stronger organizational identification (Becker-Olsen et al., 2006; Hoeffler & Keller, 2002; Öberseder et al., 2014; Wagner et al., 2009). When the organizational identification is stronger due to the high CSR firm-fit, the employee depends more on the firm. Which causes the employee to perform better, because they are dependent on the firm to do well, and therefore more motivated to do well themselves. Therefore, in H2 it was expected that when CSR firm-fit is high, the relationship between CSR on employee motivation is stronger and vice versa. In the results no support was found for H2 on employee turnover. A significant negative result was found for satisfaction score when measuring over time within firm, however the model fit was not significant. A possible explanation why CSR firm-fit could have a negative effect on employee motivation could be caused by the measure of the construct. CSR firm-fit was analyzed based on CSR and annual reports. However, the possibility of greenwashing was not taken into account in this analysis. CSR reports might contain CSR activities that are not representable for the reality, which is an insight that employees might have. Consequently, employees might be less motivated when a firm is