Under what conditions does CEO attention influence corporate social performance? : the moderating effect of span of control, executive committee gender diversity and industry dynamism

(1)

Under what conditions does CEO Attention Influence Corporate

Social Performance? The Moderating Effect of Span of Control,

Executive Committee Gender Diversity and Industry Dynamism

Master Thesis student: Anique Helen Charlotte Gregson Student Number: 11572043

Date of submission: 21st_{June, 2018} Words: 12,847 (excl. Appendix, bibliography)

Qualification: Msc. Business Administration, Strategy Track

Institution: The University of Amsterdam, Faculty Economics and Business Supervisor: P. Georgallis

(2)

STATEMENT OF ORIGINALITY

This document is written by Student Anique Helen Charlotte Gregson 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)

ABSTRACT

Firm behaviour is affected by the attention focus of the decision-maker, where attention is a scarce resource limited by attention capacity and time constraints.

Most prior literature has focused on upper echelons theory with respect to top management individual characteristics as an explanation for differences in firm behaviour, including Corporate Social Performance (CSP). However, there is limited research with regards to firm and industry aspects in attention-based view studies and in a CSP context. I integrate the attention-based view (ABV) and theorise about and test the conditions under which CEO attention focus influences CSP. I argue that a narrow span of control of the executive committee, a greater proportion of females in the executive committee and low-dynamism industry moderate the relationship between a CEO long-term and external focus and CSP. Empirical evidence found support for four of the eight hypotheses and found two significant but contradicting results, through the use of a linear hierarchical regression analysis of 130 of the largest, publicly listed U.S. firms across ten industries, over five-years (2012-2016). This study has been able to identify the conditions that influence CEO attention focus toward CSP: either short-term (myopic) vs. long-term focus or internal vs. external focus. The findings suggest that CEOs are more willing to consider strategic decisions toward CSR investments if they look beyond the short-term, and attend to long-term. Moreover, CEO long-term and external focus are strengthened by firm structure (a narrow span of control) and industry (high-industry dynamism). Separately, CEO long-term focus is strengthened by executive committee gender diversity (a greater proportion of females) after a two-year time-lag is introduced.

(4)

1. INTRODUCTION

In 2016, 87% of executives reported feeling great pressure to exhibit short-term financial performance (McKinsey&Company, 2017). Meanwhile, the rising power of stakeholders (such as activist groups) pressurises firms to invest responsibly (Waddock, Bodwell, & Graves, 2002). This demonstrates a conflict of interest and challenges for managers to satisfy shareholders in the short-term whilst investing in projects that accrue benefits in the long term, especially when the benefits are seen for society and not for the firm (Barnett & Salomon, 2012). Whilst some firms (i.e. Microsoft) have actively pursued business strategies and partnerships aimed at serving the world’s poorest people. Others continue to resist pressures to engage in social issues (Spar & Mure, 2003), taking the perspective that “the only social responsibility of business is to increase profits” (Friedman, 1970).

Although some scholars suggest that CEOs are important for setting the direction of the firm (Ocasio, 1997; Hambrick & Mason, 1984) other research argues that CEOs act as barriers to change by focusing on past or present activities from familiar sources and failure to recognise the need to adapt (Hambrick, Finkelstein, & Mooney, 2005; Tripsas & Gavetti, 2000). Furthermore, it is argued that CEOs are simply not relevant for Corporate Social Responsibility (CSR) and that CSR programs are often initiated and run by a variety of internal managers, stemming from the bottom-up (Rangan, Chase, & Karim, 2015; Türker, 2015). Concluding, views are mixed on the role of the CEO in CSR. I link CSR to the attention-based view (ABV) theory which states that the structuring and management of attention are central to control processes and firm behaviour (Ocasio, 1997, p. 189).

Attention is deemed a key barrier in organisational activity and becomes more significant the higher up in the firm, as senior managers such as the CEO are faced with a collection of issues and opportunities at the same time. In addition, they have limited

(6)

basis of what they find most relevant. It is apparent that managerial attention is an important resource inside a firm and its delegation with regard to activities can be used to explain why some firms respond to CSR whilst others do not (Ocasio, 1997).

ABV is based on three interrelated premises: (1) focus of attention, (2) situated attention and (3) structural distributed attention. 67% of empirical research from 1999 to date, is limited to the first dimension: the focus of attention (Ocasio, 2011; Maula, Keil, & Zahra, 2013). Recently, upper echelon theory (individual values, behaviours and experiences) has been extensively used in explaining heterogeneity in CSR (Waldman, Siegel, & Javidan, 2006; Manner, 2010; Post, Rahman, & Rubow, 2011; Griskevicius, Cantu, & Van Vugt, 2012; Chin, Hambrick, & Trevino, 2013). Research has expressed the importance of looking beyond individual upper echelons aspects that affect attention and firm behaviour, at distribution processes and channels of communication that influence information flow (Ocasio, 2011; Maula, Keil, & Zahra, 2013).

I use ABV theory to fill this gap by exploring CEO attention in all three ABV dimensions. I further explore firm and industry level aspects, as recommended for future research (Ocasio, 2011). The majority of empirical ABV studies have focused on innovation adoption or response strategies to technological disruption (Yadav, Prabhu, & Chandy, 2007; Eggers & Kaplan, 2009). I argue it is important to focus on firm’s responses and adoption toward CSR due to the increasing conflicting pressures. Moreover, most studies have been conducted in a single sector; I broaden this by considering ten industries across the U.S.

Ocasio (2011) encourages future research to look beyond individual attention in explaining distributed attention, where organisational level factors are important for explaining the link between attention focus, situation, and structural distribution (Ocasio, 2011). Moreover, the majority of research is conducted in one context, with no differences in situated attention (Correa, Ortiz-de-Mandojana, & Bansal, 2018). This paper seeks to

(7)

determine under what conditions CEO attention influences Corporate Social Performance, by exploring the moderating effect of firm characteristics (span of control and executive committee gender diversity) and industry characteristics (dynamism).

Firstly, the span of control (the size of the executive team), broad spans of control mean access to more knowledge and capabilities and may contribute to enhanced CSP of firms (Haleblian & Finkelstein, 1993). At the same time, it may lead to miscommunication and/ or more time spent internally, with less time to address external or future trends (Bandiera, Prat, Sadun, & Wulf, 2014). Secondly, a study on board of directors found that females are more stakeholder-orientated whereas men are considered more aggressive and performance orientated (Post, Rahman, & Rubow, 2011). However, it remains unclear whether executive committee gender diversity influences the attention of CEOs toward CSR, but is important to consider due to the more frequent meetings between CEO and executive committee members (weekly) than CEO and board of directors (quarterly) (Finkelstein, Hambrick, & Cannela, 2009). Lastly, the level of industry dynamism which refers to the level of unpredictable change versus stability influences the amount of information that needs to be processed by executives. The amount of competition and pace of change in an industry (Child, 1972) is ultimately likely to influence CEO attention and CSP.

To address these gaps, I studied the 130 largest U.S. firms, from S&P500 over a five-year period (2012 – 2016). Although large firms in the U.S. are notorious for aggressive strategies to meet short-term shareholder needs (Laverty, 1996), they also receive the greatest pressure from external stakeholders to act responsibly (Rehbein, Waddock, & Graves, 2004). This empirical setting is important, in order to understand how CEO’s attend to answers and issues with conflicting pressures and limited attention capacity. This sample uses a text analysis of shareholder letters to measure CEO attention and ASSET4 to measure the CSP. I

(8)

variables and applied a Linear Ordinary Least Squares (OLS) hierarchical regression analysis to test the hypotheses. This study provides several relevant contributions.

To conclude, these findings have important implications for studies on attention-based view and corporate governance literature. First, prior CSR studies emphasise a future-time perspective, but this study separates CEO attention into four foci: short-term, long-term, internal and external (Yadav, Prabhu, & Chandy, 2007). By defining differences in CEO attention and further illustrating the effect of these different attention foci, it will support managerial understanding of the effect their attention has on CSR activities.

Secondly, less utilised in CSR literature, ABV theory proves useful in understanding individual, firm and industry aspects. Moreover, this study pays greater attention to internal and external mechanisms, important in understanding decisions in CSR. By addressing the gap, I provide additional empirical evidence to the current ABV and CSR literature regarding firm and industry level aspects that enhance or inhibit CSP (Ocasio, 2011; Maula, Keil, & Zahra, 2013). On a managerial level, this research will help CEOs in understanding the optimal firm structure with regard to a strategic decision in CSR and the attention focus needed in uncertain or stable industries. As discussed in the concluding section of this thesis, this study contributes to the current empirical research and building on ABV theory, uncovers the interactive effect of firm and industry factors that shape CEO attention and CSP.

2. LITERATURE REVIEW

Over the last decade, due to a growing demand from society increasing attention has shifted towards firms Corporate Social Performance (Ioannou & Serafeim, 2012) and society requires firms operate beyond making a profit (Marcus & Fremeth, 2009). In this literature review, I review existing literature on the topics given. In the first section, I cover the emergence and definitions of CSR and CSP, alongside the view of scholars and managers

(9)

with regard to shareholder and stakeholder theory to highlight the differing views of managers. The second section explains the attention-based view, which will provide a basis for the hypotheses to understand why certain firms attend and adapt to CSR, whilst others do not.

2.1 Corporate Social Performance

One of the most widely cited definitions of Corporate Social Responsibility (CSR) is: “a concept whereby companies integrate social and environmental concerns in their business operations and in their interaction with their stakeholders on a voluntary basis” (Communities, 2001, p. 3). Wood refers to Corporate Social Performance (CSP) as the measure of a company’s performance pertinent to CSR (Wood, 1991).

Some firms are highly responsive to societal needs, whilst others are not (Spar & Mure, 2003), emphasizing the importance of understanding why firms respond differently to embracing CSR strategies (Campbell, 2007). On the one hand, Friedman (1970) argues that a corporate executive is an employee of the owners (shareholders) and they must conduct business accordingly. He argues that “The Social Responsibility of businesses is to increase its profits” (Friedman, 1970). Shareholder theory builds on Friedman (1970) article, and proposes that a corporation’s role is to legally maximise long-term shareholder wealth (Jensen, 2002; Freeman, Wicks, & Parmar, 2004) and that profit maximisation is the only objective and fiduciary responsibility1 of a firm (Waldman & Siegel, 2008; Marcoux, 2003). Friedman (1970) concludes his research by arguing CSR is not within the fiduciary duty of the firm and it is, therefore, an additional cost beyond the demands of the owners. CEOs and top managers often still have the view that profit maximisation is the firm’s sole fiduciary duty and that responding to CSR activities is costlier (Spar & Mure, 2003; Ambec & Lanoie, 2008). For instance, despite many companies advocating CSR, ultimately firms do not

(10)

increase stakeholder welfare at the expense of shareholder wealth (Karnani, 2010; Fehre, 2015).

On the other hand, managers increasingly recognise the impact that their actions have on society and scholars emphasise the importance of serving all stakeholder needs, beyond shareholders. Stakeholders include anybody who can influence or is influenced by the actions of a firm’s objectives (Freeman E., 1984). Stakeholder theory shifts a firm’s focus from shareholders to the needs of multiple stakeholders, resulting in a multi-fiduciary relationship (Agle, Mitchell, & Sonnenfeld, 1999). Progressively more firms respond to external stakeholder pressures, suggesting the importance of external players beyond shareholders in strategic decisions toward CSR.

These studies show scholars and managers conflicting views towards embracing CSR. It still remains a point of debate about why certain firms invest more in CSR than others, given the short-term pressure from shareholders (Correa, Ortiz-de-Mandojana, & Bansal, 2018). Ocasio (1997) argues that attention is one of the most important scarce resources influencing firm behaviour and strategic responses (Ocasio, 1997).

2.2. Attention-based view

Attention is defined as “the noticing, encoding, interpreting and focusing of time and effort by organizational decision makers” (Ocasio, 1997, p. 189). Top management attention is especially important to consider as, the higher up in the organisation the more information individuals receive and the more important directing attention becomes (Ocasio, 1997). Consequently, understanding variety in CEO attention is crucial for understanding differences in firm behaviour, their adaptive capabilities and strategic choices towards CSR. Ocasio developed three key interrelated dimensions of ABV to facilitate understanding of the variety in attention.

(11)

1. Focused Attention

The first principle, focused attention, facilitates action towards issues and activities being attended to and inhibits action towards those that are not (Ocasio, 1997, 2011). CEOs are considered the most critical players when it comes to attention management and cognition (Ocasio, 1997). CEOs influence and direct information others attend to and they influence how employees interpret information (Daft, Sormunen, & Parks, 1988). According to the theory of bounded rationality (March & Cyert, 1963), given CEOs’ time and capacity constraints, CEOs often focus their attention and interpret information from familiar sources, using established methods and routines, cognitive experience or based on past knowledge or competition (Levinthal & March, 1993; Barr, 1998).

CEO attention focus is highly complex in today’s business context; due to simultaneous, short-term pressure to maximise profits and long-term pressure to act socially responsible (Correa, Ortiz-de-Mandojana, & Bansal, 2018). CEO attention focus has been found to significantly affect firm response rates, strategic behaviour and ultimately performance (Yadav, Prabhu, & Chandy, 2007; Eggers & Kaplan, 2009; Kammerlander & Ganter, 2015; Correa, Ortiz-de-Mandojana, & Bansal, 2018). It is fundamental to understand the different attention foci that influence or inhibit CEO attention towards CSR investment, beyond a CEOs’ future focus (Correa, Ortiz-de-Mandojana, & Bansal, 2018). Extending prior research, this study addresses different elements of attention, important for CSR: CEO focus towards the short-term or long-term focus (Türker, 2015; Shipp, Edwards, & Lambert, 2009) and internal or external focus (Yadav, Prabhu, & Chandy, 2007). By dissecting attention, this will advance previous CSR studies solely looking at future time perspectives (Correa, Ortiz-de-Mandojana, & Bansal, 2018).

(12)

echelons, but limited research considers the other two dimensions discussed below: situated attention and structural distribution of attention (Ocasio, 2011; Maula, Keil, & Zahra, 2013). 2. Situated attention

The second principle, situated attention, and what individuals do, depends not only on the attention focus of the individual but on the context, environmental stimuli and capacity to respond to unexpected situations (Ocasio, 1997). The situation influences an individual’s perception of the need to respond/change (i.e. the urgency) (Barr, 1998). Much less empirical attention has been paid to situated attention (Ocasio, 2011). Moreover, the majority of attention-based studies have been conducted in the U.S. and have looked at specific industries (Yadav, Prabhu, & Chandy, 2007; Eggers & Kaplan, 2009; Kim, Fairclough, & Dibrell, 2017; Correa, Ortiz-de-Mandojana, & Bansal, 2018). Limited research has tested the effects of contextual factors on managerial attention, beyond one industry (Li M. , 2015). This study includes the largest publicly listed firms in the U.S., as they hold the greatest pressure from external parties (i.e. from the media or social movement organisations) to act more responsibly (Hoffman & Ocasio, 2001; Rehbein, Waddock, & Graves, 2004).

The attention focus is not only determined by the individual characteristics of the decision makers (attention focus), but are shaped by the situations and environmental context (Ocasio, 1997). Thus, understanding the effect of aspects in the external environment will contribute to greater understanding of the role of industry. One of the few empirical studies by Nadkarni and Barr (2008) examined the determinants of attentional perspective finding that high-velocity environments influence attention toward competitive and market forces (Nadkarni & Barr, 2008). Nadkarni and Barr (2008) suggest exploring other boundaries beyond technological disruption and innovation adoption (Nadkarni & Barr, 2008; Eggers & Kaplan, 2013). I explore the effect of industry dynamism on CEO attention and CSP.

(13)

3.Structural distribution attention

This dimension refers to the existing rules of the game, routines, procedures, resources and social relationships within a firm that determine the importance of relevant issues and answers to attend to (Ocasio, 1997). The variation in attention also stems from the procedures and communication channels amongst group members who divert the attention of decision makers (Ocasio, 1997). There has been limited empirical research in the realm of structural distribution determinants that influence CEO attention (Cho and Hambrick, 2006; Ocasio, 2011; Maula, Keil, & Zahra, 2013). Specifically, the existing rules of the game may become counterproductive for CSR strategies due to its uncertain nature and CEO’s may be inclined to filter information and focus on current routines which limit top management’s attention to CSR (Levinthal and March 1993). Therefore, it is important to explore communication channels and firm-level aspects that affect CEO attention toward CSP. I explore the moderating effect of two internal-firm level mechanisms: span of control and executive committee gender diversity on CEO attention towards or away from CSR investments.

3. THEORETICAL FRAMEWORK

In this section I synthesise prior studies and develop arguments to link into my study. In section 3.1 and 3.2, the first perspective of ABV is considered regarding CEO attention focus and hypotheses are developed. In section 3.3, 3.4 and 3.5 this study considers the influence of moderators and hypotheses are developed for span of control, executive team gender diversity and industry dynamism.

3.1 CEO Attention: Short-term vs. Long-term focus

A “short-term focus” can be defined as firms who overvalue the short term (i.e. profits) at the expense of firm’s decision to invest in the long-term. “Long-term focus” refers to the amount of attention devoted to events or activities that are yet to occur and alerts managers towards

(14)

focus is important to consider with regard to CSR decision-making due to contradicting temporal pressure from shareholders and stakeholders.

Firstly, CEOs that have a greater focus on the long-term tend to have a greater awareness of events in the future and are more likely to anticipate future societal needs (Prahalad & Hamel, 1994; Yadav, Prabhu, & Chandy, 2007; Bluedorn & Martin, 2008; Wang & Bansal, 2012). Although firms focusing on the short term have quick returns and can adapt fast, managers have a tendency to act in a ‘myopic’ way and invest less in long-term projects due to the high costs, reduced current earnings and uncertainty of the outcome (Hannan & Freeman, 1984; Laverty, 1996; Devers, Cannella, Reilly, & Yoder, 2007; Griskevicius, Cantu, & Van Vugt, 2012). This is particularly visible in the case of public companies who are under pressure by their shareholders to deliver short-term results (McWilliams & Siegel, 2001; McKinsey&Company, 2017). Thus, suggesting that a CEO short-term focus, is not inclined to invest in new resources or explore new opportunities, necessary for CSR

strategies (Maletič, Maletič, Dahlgaard, Dahlgaard-Park, & Gomišček, 2014). I conclude, a CEO long-term focus is likely to positively affect CSP due to the long-term nature of CSR and the more inclined CEOs are to detect, anticipate and invest in future and uncertain needs of society.

Secondly, although investing in CSR is uncertain, a CEO long-term focus enables the CEO to look beyond the short-term costs and uncertainty of CSR. They are more likely to explore the potential long-term benefits CSR provide such as employee retention, brand reputation, positive externalities and favourable customer loyalty (Margolis, Elfenbei, & Walsh, 2009).

Thirdly, in today’s context, managers face significant pressure to deliver short-term profit maximisation whilst responding to long-term uncertainties (i.e. technological disruptions, competition and stakeholder’s request for social responsibility) (Spar & Mure,

(15)

2003; Eggers & Kaplan, 2009). These pressures are likely to cause challenges and a conflict of interest for a CEOs limited attention capacity. A CEO long-term focus will reduce the possibility that the CEO is ‘myopic’ or focused on past/ short-term aspects (Laverty, 1996) and enhance their focus on long-term uncertainties and for instance, encourage exploration into future societal needs. Although attending to information in the short-term is important, it is less likely to serve as a platform for meeting all stakeholders needs due to the costs, uncertainty and long-term nature associated with CSR (Slawinski & Bansal, 2015). I argue an increased focus towards the long-term decreases the likelihood that firms are focused on issues in the past or present, allowing room for looking into the future challenges society faces, such as climate change (IPCC, 2014).

Hypothesis 1 (H1): CEO Attention focus to the long-term will positively influence Corporate Social Performance

3.2 CEO Attention: Internal focus vs. External focus

Attentional differences between the internal and the external environments influence a firm’s interpretation (Donaldson & Preston, 1995), managerial and organisational cognition, and ultimately strategic decisions (Ocasio, 1997). An “external focus” is the amount of attention devoted to aspects outside the firm (Yadav, Prabhu, & Chandy, 2007). An “internal focus” is attention devoted to aspects from within the firm and internal channels (Weick & Daft, 1984). Empirical studies find that the emphasis managers place on the internal or external environment influences the organizational actions and performance (Garg, Walters, & Priem, 2003). CSR studies have addressed CEO future time perspectives (Correa,

Ortiz-de-Mandojana, & Bansal, 2018), however, prior CSR studies have not explored CEO internal and external focus. This is important for CSR studies due to the increasing importance of all

(16)

First of all, CEOs with an internal attention focus may be inclined to partake in greenwashing (to communicate their attention to external aspects regarding CSR), but not actually implement activities or perform better with regard to CSR known as “green talk” (Wan & Walker, 2011). Such false attention, is likely to lose support and credibility from employees. Whereas, CEOs with an external focus are more likely to attend to the external environment and implement action toward CSR due to their emphasis and understanding of market conditions, customers, competitors and the demands from the external environment (Yadav, Prabhu, & Chandy, 2007). I argue a CEO external focus is likely to deliver favourable CSP.

Additionally, a CEO external focus leads to greater focus on opportunities and threats outside the firm (Yadav, Prabhu, & Chandy, 2007). External focus is highly important in CSR. First, due to societal demands coming from outside the firm and second, an external focus may be advantageous against competition and lead to faster interpretations and responses to opportunities and threats. On the other hand, a greater focus on internal operations is likely to reduce a firm’s awareness and response to external stakeholder needs. I do not dismiss the importance of internal focus alongside the need to motivate employees, and ensure operational efficiency. However, specifically with regard to CSR slow responses to the external environment may harm the firm’s reputation, stock valuation, adaptability, environmental and financial performance (Spar & Mure, 2003).

Hypothesis 2 (H2): CEO external attention focus will positively influence Corporate Social Performance

A potential resolution to the question of the conditions that affect this relationship is exploring moderators, recommended for future research (Kim, Fairclough, & Dibrell, 2017; Correa, Ortiz-de-Mandojana, & Bansal, 2018). Where prior research emphasises the first dimension of ABV: attention focus, this study builds on limited research considering all three

(17)

dimensions: attention focus, situated attention and structural distribution of attention (Maula, Keil, & Zahra, 2013). I expect CEO attention focus and CSP to be affected by certain internal firm aspects: span of control and executive committee gender diversity and external industry aspects: industry dynamism.

3.3 Span of Control

A firm’s structure is said to influence the behaviour of its employees and how decisions are made (Simon, 1945; Hall, 1999). In the CSR arena, limited research has identified aspects of organisational structure (Russo & Perrini, 2009; Marcel, 2009; Bear, Rahman, & Post, 2010; Kim, Fairclough, & Dibrell, 2017).

Responding to limited research on structural distribution attention (Ocasio, 2011; Maula, 2013), I empirically explore the executive committee structure which is a reflection of the structure of the entire firm (Guadalupe & Wulf, 2014) and these individuals are considered the most important for decision making (Mintzberg, 1979; Ocasio, 1997; Bandiera, Prat, Sadun, & Wulf, 2014). Considering the effect of firm structure on CEO attention and CSP will provide important contributions to the design of firm structures.

A span of control is defined as the number of individuals a manager is responsible for (Rajan & Wulf, 2006). Rajan and Wulf (2006) find that span of control has grown from 5 to 10 people in the U.S between 1986 and 2006 (Rajan & Wulf, 2006), further questioning the shift and impact of span of control. A broad span of control refers to a large number of individuals under direct supervision, whereas a narrow span of control refers to a small number of individuals under direct supervision (Rajan & Wulf, 2006). The majority of studies have looked at the structure of the board of directors and CSR investment (Ayuso & Argandona, 2009; Post, Rahman, & Rubow, 2011). Extending prior research, I consider executive committee span of control due to more regular communication between CEO and

(18)

executive committee compared to the board of directors (Finkelstein, Hambrick, & Cannela, 2009).

Bandiera, Prat, Sadun & Wulf (2014) looked at a small sample of 65 firms and found that span of control directly influences how individuals allocate their time. The scholars recommended further research to explore the effect of span of control on management behaviour and performance (Bandiera, Prat, Sadun, & Wulf, 2014). Extending this research, I explore the moderating effect of span of control on CEO attention and CSR Performance.

On the one hand, the larger the team, the more demographic diversity, often resulting in more coordination and communication problems that smaller groups do not have (Haleblian & Finkelstein, 1993). The team may be too diverse to reach a consensus and may actually increase the level of conflict (O'Reilly, Caldwell, & Barnett, 1989). On the other hand, scholars found that larger teams provide more knowledge and capabilities which lead to faster growth and more resources to formulate solutions (Haleblian & Finkelstein, 1993). I address these mixed results to study the effect of a span of control on CEO attention and CSP.

Firstly, the broader the span, the more likely the CEOs time and information capacity are limited to focusing on the short-term aspects that need attending to. This is partly because the CEO must attend to more employees during meetings, limiting his time to be able to attend to aspects beyond the short-term. Moreover, a broader span of control of executive committee suggests greater CEO time constraints and therefore CEO’s are likely to attend to information from familiar sources such as, competition or past knowledge (Barr, 1998). This is perhaps damaging for CSR due to the need for exploration, which requires looking beyond familiar sources and current practices (Maletič, Maletič, Dahlgaard, Dahlgaard-Park, & Gomišček, 2014). On the contrary, a narrow span of control may enable the CEO to spend more time focused on activities beyond the short-term, as he is able to respond to all

(19)

constituents faster, reducing time constraints and increasing his ability to attend to other information. I argue that, due to the long-term nature of CSR, a narrow span of control reduces their time constraints and will facilitate CEO attention towards aspects beyond the short-term and broaden attention to long-term activities, such as CSR.

Secondly, a broad span of control of executive committee, may encourage CEOs to delegate CSR to other constituents, as the broader the span, the less attention capacity the CEO has to look beyond the short-term and internal environment. Moreover, due to the voluntary nature of CSR, CEOs may consider CSR strategies as less urgent to attend to and delegating may be favourable.

Hypothesis 3 (H3): A broader span of control will negatively moderate the relationship between long-term focus and Corporate Social Performance.

In a larger executive committee, with a broader span of control, the more CEO attention is spent internally as the CEO must attend and justify his actions to more executives (Bandiera, Prat, Sadun, & Wulf, 2014). As a result, the CEO potentially has less time to spend on external opportunities and/or threats. This could be especially detrimental in the context of CSR, due to the increasing pressure firms receive from external stakeholders to invest responsibly (Agle, et al., 2008). Conversely, in a smaller team, CEOs may have more attention capacity and time to spend on things beyond current internal issues. As a result, the CEO’s attention can be placed towards external factors and respond to the needs of external stakeholders.

Hypothesis 4 (H4): A broader span of control will negatively moderate the relationship between external focus and Corporate Social Performance.

(20)

3.4 Executive Committee gender diversity

The majority of CEO positions in the S&P500 remain dominated by men. In my study (3%) of CEO’s are female.2 Therefore, the executive committee who support the CEO with strategic decision-making is considered a suitable, representable moderating variable (Rosenstein, 1988; Bandiera, Prat, Sadun, & Wulf, 2014).

Firstly, a board of director’s composition and structure has been found to influence firm’s strategic decision making (Baysinger & Hoskisson, 1990; Kor, 2006; Chin, Hambrick, & Trevino, 2013). Predominantly, research has focused on the board of directors and found that the greater proportion of female directors on the board positively influence implementation of environmental actions (Bear, Rahman, & Post, 2010). Building on research regarding board diversity, I will look specifically at the executive committee gender diversity. The executive committee is likely to have a significant effect on the relationship between CEO attention focus and CSP, as they are the most influential individuals who directly report to the CEO (Ocasio, 1997). Moreover, they hold more frequent meetings with the CEO than the board of directors (Finkelstein, Hambrick, & Cannela, 2009). As a result, the more female members, the more likely they will emphasize the importance of stakeholder-orientated aspects which may affect CEO attention and CSP. This argument is strengthened by males whose behaviour tends to be strongly influenced by women’s preferences (Van Vugt, Griskevicius, & Schultz, 2014).

Secondly, previous studies have shown that men and women differ in their behaviour based on their values, nature and capabilities (Bear, Rahman, & Post, 2010; Griskevicius, Cantu, & Van Vugt, 2012). These differences attribute to the way men and women perceive situations and behave in their jobs. Men are seen as more risk-taking and aggressive than women (Daly & Wilson, 1988), advocating that women may be more orientated towards the

2_{Whilst attaining information regarding the CEO turnover from the annual reports, I collected}

(21)

long-term than men. Specifically, men directors are often more willing to minimise CSR activities for financial gain, supported by research that finds men more willing to waste the environment for short-term financial gains (Griskevicius, Cantu, & Van Vugt, 2012). Contrastingly, female directors are considered more stakeholder orientated (Van Zijl, Lückerath-Rovers, & De Bos, 2011). Building on these findings, the greater the proportion of females in the executive committee, the more likely they will affect CEO attention and CSP. Hypothesis 5 (H5): A higher proportion of female gender diversity on the executive committee will positively moderate long-term focus and Corporate Social Performance.

Hypothesis 6 (H6): A higher proportion of female gender diversity on the executive committee will positively moderate external focus and Corporate Social Performance.

3.5. Industry Dynamism

A recent study stated the importance of taking industry into consideration when conducting future research on CSR (Correa, Ortiz-de-Mandojana, & Bansal, 2018). The focus of prior studies has predominantly been on aspects within the firm (Maula, Keil, & Zahra, 2013). Yet, individual’s attention is also influenced by situations and players outside the firm’s boundaries (explained in ABV: situated attention) (Ocasio, 1997). Dynamism refers to the degree of uncertainty, instability and volatility in industry conditions (Dess & Beard, 1984). In a dynamic industry, change is common and requires significant cognitive demands (Wiersema & Bantel, 1993), rapid response strategies (Miller & Friesen, 1982; Keats & Hitt, 1988; D'Aveni & MacMillan, 1990) and innovation to survive (Brown & Eisenhardt, 1997; Christensen & Suarez, 1998). The results in management literature are mixed. On the one hand, research found uncertainty reduces environmental investments (Dowell &

(22)

Muthulingham, 2016) whilst, on the other hand, industry dynamism and munificence increase environmental investment (Rasheed & Goll, 2004; Chen, Zeng, Lin, & Ma, 2017).

Firstly, scholars claim that dynamism reduces managerial inertia (Laverty, 1996), but, with high uncertainty and unpredictability, CEOs may be more inclined to focus their time and attention towards innovation and new product introductions; considered key for survival in turbulent industries (Christensen & Suarez, 1998; Brown & Eisenhardt, 1997; Chen, Chen, & Nadkarni, 2015) and are less likely to adopt environmental initiatives (Dowell & Muthulingham, 2016). For instance, firms that confront uncertainty through innovation often outperform those that resist uncertainty or are slow to respond (Garg, Walters, & Priem, 2003). Moreover, in a stable environment, customer preferences are relatively steady and the increased costs of innovation are not necessary (Moorman & Miner, 1998). I argue that CEO’s with a long-term focus in high-dynamism industries are more inclined to focus on innovation due to limited attention capacity, time constraints and pressure to stay competitive. Whereas, a CEO with a long-term focus in stable environments, can invest in strategies beyond innovation, such as CSR.

Secondly, in a stable, low-dynamism industry, managers tend to place greater weight on routine-activities rather than innovation (Rothenberg and Zyglidopoulos, 2007). In stable industries, CEOs with a long-term focus will seek to adopt initiatives that hold the lowest risk. Whilst, innovation is seen as a high-risk, long-term strategy; CSR can be a tool for reducing risk. For instance, CSR can reduce risks by anticipating future legislation and/ or stakeholder needs (Fontrodona & MacGregor, 2008). I argue innovation is less necessary for firms in more stable environments; and firms in stable industries with CEO long-term attention often embrace more routine, low risk-taking strategies, such as CSR.

Hypothesis 7 (H7): A low-dynamism industry will positively moderate the relationship between CEO attention to the long-term focus and Corporate Social Performance.

(23)

A CEO with an external focus predominantly focuses on external stakeholders; and is more inclined to explore external opportunities, threats and market trends. In a stable industry, high-risk taking initiatives such as innovation and new product introductions are less urgent (Moorman & Miner, 1998). Giving a CEO with external focus more time and attention capacity to focus on the external environment beyond innovation and survival, on activities such as CSR (Fontrodona & MacGregor, 2008). Whilst, a CEO in a high-dynamism industry with an external focus is more likely to be more concerned with risk-taking strategies to ensure survival and competitiveness in the short-term.

Hypothesis 8 (H8): A low-dynamism industry will positively moderate CEO external focus

and Corporate Social Performance

4. CONCEPTUAL MODEL

H1 (+)

H2 (+) CEO Attention:

Long-Term focus

H3: Broad span of control (-)

H5: Exec committee diversity: Female (+) H7: Low dynamism Industry (+)

Corporate Social Performance CEO Attention:

External focus

H4: Broad span of control (-)

H6: Exec committee diversity: Female (+) H8: Low dynamism Industry (+)

(24)

5. RESEARCH DESIGN

5.1 Sample and data collection

I tested these predictions using a quantitative study, on the largest 130 publicly-listed firms operating in the United States (U.S.) during 2012-2016 from S&P500. I argue that strategic decisions to implement CSR do not have an immediate effect on CSP. I lagged all predictor variables in 2011 to ensure proper causal inferences (year t-1) (Bear, Rahman, & Post, 2010; Correa, Ortiz-de-Mandojana, & Bansal, 2018).

Several criteria guided the choice for this setting. First, scholars believe that larger firms have more resource capabilities, making it a feasible context for firms to invest in CSR (Wu, 2006). Second, I conducted this research across ten industries, which has been asserted important for future research (Correa, Ortiz-de-Mandojana, & Bansal, 2018) and will provide more generalizable results. Third, beyond industry, the context of national culture in my study remains constant as, culture has been proven to influence individual’s perceptions and interpretations of environmental uncertainty (Hofstede, 2001; Waldman, et al., 2006; Ioannou & Serafeim, 2012). Fourth, this study was conducted using data on the most recent years (2012-2016).3 Lastly, the U.S. is considered a risk-taking and capitalist nation, partially explained through short-term pressure firms receive to deliver high shareholder value (Laverty, 1996; Williams, 2000) whilst they receive increasing pressure from stakeholders to invest responsibly in the long-term (Spar & Mure, 2003), making the U.S. an appropriate context to test conflicting pressures on CEO attention focus toward CSP.

3_{The most recent years have been chosen due to the rise in theoretical and managerial discussions, emphasising}

the increasing importance of CSR (Bhattacharya & Sen, 2004). Also, due to the increasing pressure firms are receiving in today’s business context on the short and long-term (Correa, Ortiz-de-Mandojana, & Bansal, 2018).

(25)

5.2 Dependent variable

I collected data regarding my dependent variable, CSP from DataStream (ASSET4) over five-years from 2012 – 2016 (note, the year 2011 is removed due to the lag on predictor variables). The point of time of the effect of CEO attention on CSP is unknown. Therefore, the effect has to be tracked over a number of years, rather than just a one-year period. I constructed a CSP index for each firm for every year in my sample, named weighted average CSP (WACSP). These scores are equally weighted and aggregated in my study across the three pillars 4 to avoid subjectivity of categories. I did not, however, consider the economic scores in my analysis as these are deemed less relevant for my study (Waldman, Siegel, & Javidan, 2006; Waldman, et al., 2006; Ioannou & Serafeim, 2012).

5.3 Independent variable

Following the collection of CSP per firm, I used a text analysis of shareholder letters to operationalise CEO attention, a widely used source for detecting the meaning of words (Occasio, 1997; Cho & Hambrick, 2006; Yadav, Prabhu, & Chandy, 2007; Tausczik & Pennebaker, 2010). Numerous studies have shown that shareholder letters are a useful proxy for determining the attention allocation (Abrahamson & Hambrick, 1997; D'Aveni & MacMillan, 1990; Cho & Hambrick, 2006; Yadav, Prabhu, & Chandy, 2007; Fehre, 2015). The data was collected from shareholder letters over six-years from 2011-2016. Note, a lagged independent variable is included in the regressions (year t-1). This study focuses on The United States to ensure all shareholder letters are in the same language for interpretation and to keep environmental and national pressure constant (Bhattacharya & Sen, 2004).

Linguistic Inquiry and Word Count (LWIC) is a computerised content analysis program developed by Pennebaker, Francis and Booth (2001) (Tausczik & Pennebaker, 2010). The use of this computer-aided text analysis tool made this study feasible in the time

(26)

frame and removes reliability concerns related to manual coding (Short, Broberg, Cogliser, & Brigham, 2010; Duriau, Rhonda, & Pfarrer, 2007). Based on prior research on future attention toward innovation adoption (Yadav, Prabhu, & Chandy, 2007), I developed a custom dictionary5_{and cross-referenced synonym book by (Rodale, 1997). I constructed} dictionaries for coding which related specifically to the context of CSR investment, which can be found in (Appendix Table 1) (Yadav, Prabhu, & Chandy, 2007; Pennebaker, Chun, Ireland, Gonzales, & Booth, 2007; Short, Broberg, Cogliser, & Brigham, 2010;). In order to measure this, I construct two ratio variables to measure: the proportion of long-term focus relative to short-term focus and the proportion of external relative to internal focus, per firm. 5.4 Moderating variables

5.4.1 Span of Control

To test for H3 and H4, I needed to gauge the differences in CEO span of control. Span of control is operationalized through the number of executives under direct command of the CEO (Bandiera, Prat, Sadun, & Wulf, 2014; Li, Maggitti, Smith, Tesluk, & Katila, 2013). I gathered this information in the annual reports.6 Note that a lagged independent variable is included in the regressions (year t-1), with 2012-2016 fit for analysis.

5.4.2 Executive Committee: gender diversity

To test H5 and H6, I explored the effect of executive committee gender diversity on CEO attention and CSP, which looks beyond extensive research regarding board of director’s gender diversity and the effect on decision-making (Bear, Rahman, & Post, 2010). This variable is operationalized by the number of female executives divided by the total number of executives. The gender of the executive committee members is collected manually from annual reports from 2011 – 2016, and a time lag (year t-1) is included in the regressions.

5_{To verify the reliability of the custom dictionaries, I ask for multiple opinions from individuals}

fluent in English.

6_{I take care that firms use different terms for the same meaning in annual reports: “executive}

(27)

5.4.3 Industry conditions: Dynamism

To test the final moderator for H7 and H8, the firms are first classified into ten different industry classifications7 attained from DataStream (ASSET4) under “ICB Industry name”.

I then operationalised industry dynamism, by gathering data from Compustat on each firm’s R&D expenditure from 2011-2016; note, thereafter a time lag is included (year t-1). It has been argued that the nature of an established industry remains relatively stable over time (Catrogiovanni, 2002; Henderson, Hambrick, & Miller, 2006). I argue that although change amongst firms is common in high-dynamism industries, the level of industry dynamism doesn’t change significantly over time and, as a result, I keep industry dynamism as a constant per industry over the course of my study.

Thereafter, I calculated the mean (average) industry level R&D investment for all of the firms in the ten classified industries (Acs & Audretsch, 1990; Thornhill, 2006). Finally, the industries are categorised into low, medium and high industry dynamism, based on the level of R&D investment depicted in Appendix, Graph 1. To reduce bias, I took the frequency of cases per industry into consideration, to ensure a similar distribution of cases across the three levels of industry dynamism. To ensure that the categorical variables were interpretable in the regression, I used low-dynamism as the reference dummy variable (0) and created two dummy variables for medium and high industry dynamism relative to low-dynamism industries.

5.5 Control variables

This model includes several individual, firm and industry level variables and to improve the estimation, I needed to control for possible variations in CSP. Note, the control variables have been lagged (year t-1) leaving 2012 -2016.

(28)

Firstly, firm size is controlled for, because a larger organization often has more resources to engage in CSR and thus might have a better CSP than a smaller firm (Wu, 2006). In addition, larger firms, often receive greater pressure from external stakeholders (Solomon & Lewis, 2002). I measured firm size through the total number of firm assets (McGuire, Sundgren, & Schneeweis, 1988; Manner, 2010; Galbreath, 2010; Ioannou & Serafeim, 2012; Wu, 2006; Yadav, Prabhu, & Chandy, 2007) to account for differences in CSR investment. I attained this data from DataStream (ASSET4).

Secondly, firm performance as performance may have an effect on a CEO’s attention focus and strategic decision-making (Ling, Zhao, & Baron, 2007; Souder & Shaver, 2010). CEOs that perceive high performance, may focus on longer-term that have uncertain outcomes, such as CSR activities (Rothenberg & Zyglidopoulos, 2006); whereas bad performance encourages short-term focus to improve the current performance. I measured firm performance through return on assets (ROA) attained in DataStream (ASSET4) (Correa, Ortiz-de-Mandojana, & Bansal, 2018; Thorne, Mahoney, & Manetti, 2014; Ioannou & Serafeim, 2012; Manner, 2010; Margolis, Elfenbei, & Walsh, 2009 ).

Thirdly, all regression models include CEO attention focus, so I controlled for a change in CEO, which may influence the CEOs perspective and the direction they take the firm in. Hambrick and Fukutomi (1991) argued CEO tenure influences the strategic direction (Hambrick & Fukutomi, 1991). I controlled for a change in CEO through the use of a dummy variable (Hardy, 1993; Yadav, Prabhu, & Chandy, 2007; Fehre, 2015). This information is found in annual reports and is named CEO Turnover (CEOT).

Fourthly, I control for ‘time’ operationalised by years (Bhattacharya & Sen, 2004). Multiple years enabled me to fully explore the impact of the relationship between CEO attention and CSP. It aided me to account for things that are different year to year, that are not otherwise explained or accounted for in my other variables, such as for example the

(29)

financial crisis (Fehre, 2015). I created five categorical binary dummy variables for each year in my study, to observe the effect of years in my sample, with 2012 as the reference year (0) (Correa, Ortiz-de-Mandojana, & Bansal, 2018).

Lastly, in terms of measuring CEO attention, the number of words per shareholder letter is controlled for. The length of letters to shareholders vary across firms and may correlate with the measure of managerial attention (Yadav, Prabhu, & Chandy, 2007). The data regarding word count was simultaneously retrieved from the LIWC program during the text analysis of shareholder letters.

6. METHOD, MODELS & RESULTS

The following section provides the methods, models and results of this research. Firstly, I provided descriptive statistics and conducted normality tests for the variables in this study, presented in Table 2 and (Appendix Tables 3, 4 and 5). Secondly, I performed a correlation analysis found in Table 6 and the significant correlations shown. Lastly, based on the nature of my study and the econometric specifications explored, I conducted three multiple hierarchical regression models to test the hypothesis and can be found in Table 11.

6.1 Descriptives

A frequency test was run to account for missing data, concluding <13% for all variables. Due to the large sample size and data missing at random, it will not affect my results significantly (Little, 1998; Fidell & Tabacknick, 2012) and I removed 20 firms to leave a final data-set of 130 firms and 650 observations.

The majority of the variables are continuous variables and an overview of the mean, skewness and kurtosis can be found in Table 2 below and variables outside the acceptable range underwent a log transformation (Appendix, Table 3). The categorical variables underwent a frequency analysis (Appendix Table 4). 94.3 % of CEOs did not change

(30)

dynamism (34.9%); medium dynamism (29.3%) and high dynamism (35.8%). I additionally used frequency analysis for all my variables to check for the outliers in my data. After re-checking the data was entered correctly, and considering the nature of my variables, where large differences in investments can be expected, I did not remove outliers.

Table 2 Descriptive Statistics

6.2 Correlation

I used Spearman’s correlation coefficient for the categorical variables in my study which have been proven to be best suited for this type of variable (Hauke & Kossowski, 2011) and I used Pearson’s correlation coefficient for all continuous variables.

Descriptive Statistics

N Min Max Mean

Std.

Deviation Skewness Kurtosis

WACSP 650 52.67 287.44 212.30 60.26 -.73 .096 -.696 .191 Long-term Focus l 650 .08 20.57 1.32 1.45 5.56 a .096 54.53 a .191 External focus l 650 .05 12.67 1.05 1.53 4.30 a .096 22.34 a .191 Span of Control l 650 4.00 56.00 11.95 5.85 2.84 a .096 13.84 a .191 Exec Committee Diversity l 650 .00 .57 .15 .103 .246 .096 -.157 .191 Word Count l 650 10.0 91365.0 2614.30 4155.21 15.64 a .096 322.80 a .191 Firm Performance l 650 -4.61 3.77 1.77 .939 -1.38 .096 4.132 a .191

Firm Size l 650 1290883.00 2.57E+9 97554787.8 328507933.00 5.49 a .096 31.58 a .191 Notes: This table represents the descriptive statistics for the sample of this study. All the variables receive an l for those that have been lagged by one year. In addition, all variables other than the two categorical variables are reported here, however, those with an a will be log transformed, to correct for skewness and/ or kurtosis which do not meet the acceptable range between +2 and -2.

(31)

Table 6 Correlation Matrix

6.3 Regression Analysis

To ensure the correct model was being used and to derive non-biased results, I tested the data against the linear regression assumption requirements: (1) normality, (2) linearity, (3) no multicollinearity, (4) no auto-correlation and (5) homoscedasticity (Poole & O'Farrell, 1971). Firstly, Table 2: Descriptive Statistics provided the skewness and kurtosis values for continuous variables which are deemed acceptable and normally distributed when they lie between - 2 and 2 (George & Mallery, 2010). Thus, variables outside the acceptable range (a) have been transformed, (Appendix, Table 3). Secondly, a Kolmogorov-Smirnov and Shapiro-Correlations 1 2 3 4 5 6 7 8 9 10 11 1. Year 1 2. WACSP .202** 1 3. Long-Term Focus Log -.017 .016 1 4. External Focus Log -.034 .015 .093* 1 5. Span of control Log .034 .208** .051 .062 1 6. Executive Committee Diversity .135** .263** .003 .130** .092* 1 7. Word Count .031 .014 -.032 .013 .020 .007 1 8. Firm Performance -.007 .050 .200** .012 -.021 .061 -.018 1 9. Firm size .063 .274** -.285** -.091* .144** .054 .030 -.540** 1 10. Industry Dynamism .011 .109** .066 -.110** -.039 -.034 -.056 .518** .518** 1 11. CEO Turnover .057 .088* -.051 .082* .057 -.009 .057 -.016 -.016 -.026 1

Notes: In this table, I have used Pearson’s Correlation Coefficient for all continuous variables; for the categorical variables, 8 & 9: Change in CEO and Industry Dynamism I have used Spearman’s Correlation Coefficient due to the improved accuracy of measuring such variables (Hauke & Kossowski, 2011).

(32)

Wilk statistics test was conducted, accepting the null hypothesis (p<.05) confirming the variables are normally distributed and linear (Appendix, Table 5) (Marsaglia, Tsang, & Wang, 2003). Thirdly, none of the variables in Table 6 indicate a correlation of >0.8; indicating none of the variables are measuring the same concept. Furthermore, another accurate method to test for forms of multicollinearity, is through the variance inflation factor (VIF) (Kutner, Nachtsheim, Neter, & Li, 2005), the greatest VIF score is 1.736, well within acceptable range (<10) (Appendix Table 7). Although there is no multi-collinearity present, I have centred the mean for all continuous, independent variables, as this is highly recommended for models with interaction variables (Aiken, Reno, & West, 1991; Echambadi & Hess, 2007) in order to not bias estimates of regression coefficients (Hayes, 2017). Fourthly, the results of the Durbin Watson test confirmed no autocorrelation, as the average score is 0.662, inside the acceptable range between 0 and 4 (Breusch, 1978) (Appendix, Table 8). Lastly, to ensure the errors are unrelated to any predictor variables, I tested for homoscedasticity (Hayes & Cai, 2007). In the scatter plot (Appendix Graph 10), the clustered cases suggest the homoscedasticity assumption is violated. The results of the Breusch-Pagan test reject the null hypothesis of the constant variables (<.005), where Chi2(df) = 1915.14 (….df), p<0.001, see (Appendix, Table 9). I used ‘hetero robust’ in STATA to reduce the bias of deflated t-scores and increase accuracy (Hayes & Cai, 2007).

Based on these assumptions being met (after homoscedasticity was adjusted for), I used SPSS, to conduct a multiple hierarchical fixed effect regression analysis (Aiken, Reno, & West, 1991; Torres-Reyna, 2007). A hierarchical regression model was chosen to analyse the difference in predictive capacity amongst variables. This study corrects for multiple measurements through making time (years), a fixed effect in my model. The regression models have been developed based on this formula:

(33)

Corporate Social Performance = β0 +β1 H1+β2 H2 +β3 H3+β4 H4 +β5 H5+β6 H6+β7 H7+β8 H8+ β9

H1 * H3 +β10 H1*H5 +β11 H1*H7+β12 H2*H4+β13 H2*H6+β14 H2*H8 + Control Variables + error

The results of the hierarchical regression Models 1-3 and the coefficient beta and significance levels are shown respectively in Table 11. The standardised beta is preferable and beneficial for my study as, I am comparing the effect across different scales and standardised beta allows for relative comparison (Hamilton & Hunter, 2006). In addition, the R squared is shown, which features the strength of the relationship between the variables.

(34)

Table 11: Multiple Hierarchical Regression Model Model 1 Control Variables Model 2 Direct effect Model 3 Interaction effect

Predictors Std. β P-value Std. β P-value Std. β P-value

Control Variables Firm Performance .268 *** (.000) .138 *** (.003) .125 *** (.008) Firm size .406 *** (.000) .437 *** (.000) .433 *** (.000) Word Count -.007 (.830) -.007 (.838) -.014 (.668) CEO Turnover .060 * (.095) .060 ** (.023) .063 *** (.017) Year: 2013 -.003 (.947) -.019 (.663) -.020 (.648) Year: 2014 .048 (.286) .0172 (.693) .018 (.659) Year: 2015 .136 *** (.003) .105 *** (.013) .109 *** (.009) Year: 2016 .176 *** (.000) .137 *** (.001) .135 *** (.001) Independent variables Long-term focus .070 ** (.048) -.085 (.138) External focus .037 (.326) -.067 (.213) Span of Control .108 *** (.001) .101 *** (.001)

Executive Committee diversity .179 *** (.000) .162 *** (.000)

Industry Dynamism Medium -.213*** (.000) -.206 *** (.000)

Industry Dynamism High .042 (.287) .048 (.212)

Moderating Variables

Long-term focus * Span of control -.033 (.255)

External focus *Span of control -.102 *** (.007)

Long-term focus * executive committee diversity

.018 (.605)

External focus * executive committee diversity

.006 (.857)

Long-term focus * High industry dynamism

.175 *** (.000)

External focus * High industry dynamism

.156 *** (.001)

Long-term focus * medium industry dynamism

.089 * (.082)

External focus * medium industry dynamism

.023 (.628)

Constant (WACSP) 200.914 (.000) 212.270 (.000) 211.029 (.000)

R2 .169 .273 .306

Adjusted R2 .159 .257 .281

Notes: The standardized beta is shown in the table, including the significance level of standard error (p-value) shown in parentheses after adjusting for heteroscedasticity. Amending for the heteroscedasticity, only changes the

significance (p-value) slightly and is unlikely to impact the interpretation, but the original results are in (Appendix, Table 10). The standardized beta is represented, as the variables in my study are measured on vastly different scales. It allows for a more accurate relative comparison of the predictor variables as they are measured on different scales. ***significance level of <.01; **significance level of <.05 *significance level of <.1

(35)

6.4 Regression Models

In Model 1, I regressed weighted average CSP (WACSP) on the control variables. In Model 2, I added the effect of the independent variables on WACSP, including control variables. In Model 3 I added the effect of interaction variables on WACSP including control variables.

It can be found that 16.9% of the dependent variable WACSP can be predicted by the control variables in Model 1. As expected, firm performance (z = .268, p = .000) and firm size (z = .406, p = .000) for both p<.001 and are positive and strongly statistically significant towards CSP. In addition, CEO turnover has a positively significant impact on CSP (z = .063, p = .017), p<.05. Time affects CSP, where year 2015 coefficient is positive and strongly significant (z = .109, p=0.009) and year 2016 coefficient is positive and strongly significant (z= .135, p=0.0021) compared to 2012 for both p<.001. However, surprisingly, word count did not significantly influence CEO attention towards CSP (Z = -.014, P = .668).

In Model 2, the independent variables were added increasing the predictive capacity by 10.3% to 27.2%. H1 proposes that a CEO long-term focus positively influences CSP. Model 2 shows that CEO long-term focus coefficient is positive and significant (z = .070, p= .048), p<.05 in support of H1. I also added CEO external focus to Model 2, where I postulated CEO external focus to positively impact the CSP. Surprisingly, the coefficient capturing CEO external focus toward CSP was not significant (Z = -.067, P = .213) and H2 cannot be accepted. A notable reason behind the insignificant results is there may be an underlying dampening effect of greenwashing, whereby firms attempt to please external stakeholders by overemphasising their external focus in order to improve their reputation (Laufer, 2003). As Walker & Wan (2011) found, a discrepancy between ‘green talk’ and ‘green walk’ will negatively affect financial performance (Wan & Walker, 2011). An alternative explanation was that the time-lag was insufficient, which was tested (Robustness

(36)

Therefore, I argue that the results of this study may be explained by the empty discussion by CEOs over external CSR activities within annual reports without actively pursuing them.

I then added span of control to Model 2, where the coefficient is positive and significant (z = .108, p= .001); surprisingly, a broader (larger) span of control has a positive, direct effect on CSP. When executive committee gender diversity is added to Model 2, its coefficient is positive and significant (z = .179, p= .000). Moreover, two dummy variables were created to capture the level of industry dynamism (volatility). I added medium level industry dynamism in Model 2, its coefficient is negative and significant compared to low-level dynamism at (z = -.213, p= .000), whereas high industry dynamism compared to low dynamism surprisingly, does not have significant results (z= .042, p = .287).

Lastly, Model 3 includes the interaction effects, which increases predictive capacity by 3.3% to 30.6%. H3 proposes that a CEO long-term focus toward CSP is negatively affected by a broader span of control. Surprisingly, the coefficient capturing long-term focus and span of control was not significant (z = -.033, p = .255) and therefore H3 cannot be accepted. H4 proposes that a CEO external focus toward CSP is negatively affected by a broad span of control. As expected, external focus and span of control (broad) have a negative and significant coefficient (z = -0.102, p=0.007), and H4 is supported, see Graph 5.

Under what conditions does CEO attention influence corporate social performance? : the moderating effect of span of control, executive committee gender diversity and industry dynamism