1
Amsterdam Business School
Increasing the Corporate Social Performance by
Intensive Engagement with Stakeholders.
Date: January 29, 2014
By: Derk Jan van Bijsterveldt Student Number: 10506543
MSc in Business Studies
Master Thesis International Management
Supervisor: Alan Muller Second reader: Arno Kourula
2
Table of Content
Abstract ... 3 Introduction ... 4 Conceptual Model ... 9 Theoretical Review ... 10 Stakeholder Theory ... 10 Stakeholder Engagement ... 11Corporate Social Performance ... 16
Stakeholder Engagement and Corporate Social Performance ... 19
Multinationality... 21
Multinationality, Stakeholder Engagement and CSP ... 23
Methodology ... 26
Sample ... 26
Corporate Social Performance ... 27
Stakeholder Engagement ... 30 Multinationality... 33 Control variables ... 35 Industries ... 35 Regression Analysis ... 36 Results ... 38 Correlations ... 38
Stakeholder Engagement 2011 – CSP Total Strengths 2011 ... 40
Stakeholder Engagement 2011 – CSP Total Concerns 2011 ... 41
Stakeholder Engagement 2011 – CSP Total Strengths 2012 ... 42
Stakeholder Engagement 2011 – CSP Total Concerns 2012 ... 44
Reversed Causality: CSP – Stakeholder Engagement ... 46
Answering Hypotheses ... 47
Discussion ... 50
Conclusion ... 50
Limitations ... 53
Implications and Future Research ... 55
3
Abstract
For many managers it is important to perform both financially and socially well. As is the
case for financial performance, to have a good corporate social performance(CSP) a
thought-out strategy is needed to effectively and efficiently reach this goal. A relevant issue in this
matter is the relationship between stakeholder engagement(SE) and CSP. This paper
examines this relationship by analyzing 135 S&P500 companies. Results show a positive
relationship between the SE and positive CSP. More interesting, a positive relationship is
established between SE and negative CSP. Research is also done regarding the effect that
multinationality has on the proposed relationship. The result showed no significant effect
4
Introduction
Stakeholder theory and corporate social responsibility are two constructs that cannot be
totally separated. In literature and business the two theories are linked and both are highly
present in the other. Freeman (1994) states that stakeholder theory entails two important
questions. First, what is the purpose of the firm? Second, what responsibility does
management have to its stakeholders? As these questions are essential for stakeholder theory,
the answers to these questions denote largely what the position of the company is regarding
its corporate social responsibility. The discussion about what the responsibility of the
company is has been going on for several decades. Although it is not exactly clear when it
started, a good case can be made for the early 50s (Carrol, 1999). Since then, the literature
has grown exponentially and with it the variety of stances taken in this subject. For instance,
Friedman (1970) wrote a letter to the New York Times in which he decries the idea that
businessmen have social responsibilities in the following words:
“ The businessmen believe that they are defending free enterprise when they declaim that business is not concerned "merely" with profit but also with promoting desirable "social" ends; that business has a "social conscience" and takes seriously its responsibilities for providing employment, eliminating discrimination, avoiding pollution and whatever else may
be the catchwords of the contemporary crop of reformers. In fact they are--or would be if they or anyone else took them seriously--preaching pure and unadulterated socialism. Businessmen who talk this way are unwitting puppets of the intellectual forces that have been
undermining the basis of a free society these past decades……The key point that, in his capacity as a corporate executive, the manager is the agent of the individuals who own the corporation or establish the eleemosynary institution, and his primary responsibility is to them.”
As is shown above, a radical stance is taken; he basically says that the only responsibility for
5
Although Friedman (1970) still has a large group of followers, during the years
another perspective has seen a rise in support. This group points to the advantages that
corporate social responsibility(CSR) can have for the company, thus having a more strategic
approach regarding CSR. According to Porter and Kramer (2006), companies should use
CSR to establish a competitive advantage. They state that for several years, companies did
not apply their CSR properly for two reasons. First, managers pit business against society,
when clearly the two are interdependent. Second, they pressure companies to think of CSR in
generic ways instead of thinking which CSR strategy is most appropriate for them. Although
this stance shows more in favor of CSR, it is still very much connected to what it can bring to
the company; it thus has some connection to Friedman’s (1970) position that management should create shareholder value.
Another point of view that is present according to Van Marrewijk (2003) is that a
company is part of a larger system in which it operates. It is communion that stops freedom
when it interferes with the freedom of others. This entails that a company cannot just harm
other persons, or the living space or environment in which the other inhabitants of the system
live. Being an entity of a larger whole obliges the company to adapt to its environment and to
be accountable for its impact on others. In comparison to the previous two points of view, this
focusses far more on the moral obligation from a system approach, instead of an economic
reasoning from the perspective of the firm.
As is shown above, people differ in their opinion whether CSR is an instrument, a
goal or if it has no significance altogether. For those who consider CSR to be an important
asset to the company and want to perform well socially, it is important to know how the
corporate social performance (CSP) can be increased. In this case, the CSP is the outcome of
the CSR policy, as for all that one does, the outcome may not always be the same as the
6
One aspect that could influence CSP is stakeholder engagement (SE). SE is defined
by ISEA (1999) as “the process of seeking stakeholder views on their relationship with an
organization in a way that may realistically be expected to elicit them”. Enabling and facilitating SE may have many benefits. It can create strong social ties, as an embedded status
within the network of the organization makes that the company is ‘in sync’ with its stakeholder expectation. It makes the company able to mobilize multiple stakeholders in a
coalition to build a sustainable and responsible business (Maak, 2007). According to
Greenwood (2007) SE is much associated with corporate responsibility, but SE in itself is a
morally neutral activity and may not necessarily enhance the CSP of a company. Although
this could be the case, according to Valor (2005) corporate social responsibility and
stakeholders complement and reinforce each other. In addition to this, Carrol (1998) states
that corporate citizenship, which is viewed as closely related to CSR, entails for a great part
the involvement of relationships with all important stakeholders. Also, Noland and Phillips
(2010) state that SE is an important part of CSP and applying the right SE techniques can
raise the CSP. Some disagreement on this matter is thus present, and research regarding this
relationship would be desirable. I argue that for the sustainability of the SE process, a
strategic motivation is necessary in addition to a moral judgment, because it would create a
win-win situation, which will prolong and mature the stakeholders engagement process. If
more clarity is established with regards to the issue of SE, and the relationship between CSP
and SE, managers who want to raise the CSP of the company will be better able to
comprehend the means by which to do so.
In this matter, CSR is becoming more and more a way of doing business. Logically,
the geographical diversification can have a great effect on the process and outcomes of doing
business. According to Yang and Driffield (2012), multinational companies have
7
This thus entails that companies naturally want to get bigger, and expand to other countries.
But what does this mean for the CSP, and in particular to the effect that SE has on CSP.
Much research is being done regarding the effects of multinationality on the company’s
financial performance (Buckley and Casson 1976; Rugman 1986; Dunning 1988; Tallman
and Li 1996; in Yang and Driffield, 2012) and on the effects it may have on the internal
organization (Lu and Beamish, 2004; Hennart, 2007; Bartlett and Ghoshal, 1989). Yet, no
research has been done with regards to what happens to the relationship between SE and CSP
when a company internationalizes more. When a company internationalizes, logically it will
find itself dealing with more parties than it is used to. Dealing with more partners means that
the amount and variety of preferences and demands concerning how to behave as a company
will increase. When the relationship between SE and CSP is seen in the light of this
observation regarding multinationality it may have its implication. First, it could be that it
will be harder for the firm to interact with all the stakeholders. Second, as a result of the
internationalization process, some stakeholders could become less important, thus shifting the
focus and action of the company. Lastly, dealing with an increased amount and variety of
stakeholders may lead to opposing preferences, which may make it harder to formulate CSR
policy from SE. Multinationality may thus influence the relationship between SE and CSP in
many ways. It is important for managers to know if it will do so and how, because it could
mean that policies that were perfectly workable at a domestic level may no longer be
functional as the company starts to operate internationally.
In this study, the focus will be on the relationship between stakeholder engagement
and corporate social performance. When this relationship is established, it will be important
to determine what happens to this relationship when multinationality is added into the
equation. In accordance with the reviewed literature, as will be shown below, a positive
8
on this relationship. The analysis will be done in the following order. First, a theoretical
review will be performed; this will entail a clarification of what is important in the field of
stakeholder engagement, corporate social performance and multinationality. Alongside this,
there will be a review of what has been written about the relationships between the different
variables. Hereafter, the analysis will be performed using 135 companies from the S&P 500
list of 2012. In these sections the used methods will be explained and the results will be
shown. When results are collected and possible relationships established, the discussion will
be presented. In the discussion, the practical implications for managers and the implications
for the literature will be shown and possible limitations of this study will be put forward. It
will end with some suggestions of future research that may be of importance to the
9 Conceptual Model
H1
H2
This thesis will have a strong focus on corporate social performance (CSP), and what
influence the degree of SE will have on this. Expected is that intensive interaction with
stakeholders will raise the CSP, thus denoting a positive relationship. A second analysis will
be performed to show the effect that multinationality has on this relationship. As a result of
what is shown in existing literature, it is hypothesized that multinationality will have a
negative moderating effect on the relationship between SE and CSP.
Stakeholder
Engagement
Corporate
Social
Performance
Multinationality
10
Theoretical Review
Stakeholder Theory
Freeman (1984) defined stakeholders as “any group or individual who can affect or is affected by the achievement of the organization’s objectives”. Savage et al (1991) defined stakeholders as groups or individuals who “have an interest in the actions of an organization and … the ability to influence it”. According to Gao and Zhang (2006) these definitions reflect a two-way relationship; there is interdependence between the organization and its
stakeholders. During the 1980s, fundamental changes took place in business organizations.
External stakeholders demanded greater social responsibility of companies and stakeholders
obtained influence with respect to different organizational activities. These changes were
reflected in a 1987 cover story in business week (In Savage et al, 1991):
“Outside directors are asserting themselves. Other stakeholders – from employees…to communities- want a voice. The internal balance of power is beginning to shift… the days
when CEOs could neglect their big institutional owners and other corporate stakeholders are coming to an end…[N]ow managers will have to listen to – and learn from- other groups who are demanding a voice in the running of the company”
As shown above, and also expressed through the extensive attention given to within
the academic literature (Valor, 2005, Freeman, 1984, Rowley, 1997, Pedersen, 2006,
Donaldson and Preston, 1995), since the 80s increasing importance has been accorded to the
influence of stakeholders. According to Donaldson and Preston (1995) stakeholder theory
knows three different streams of literature, which are all interrelated as well as they are
distinct. The three different streams they denote are descriptive/empirical, instrumental and
11
stakeholder theory presents a model describing what the corporation is. “It describes the corporation as a constellation of cooperative and competitive interests possessing intrinsic
value”. The second stream, instrumental, refers to the establishment of a framework, which examines any potential connections between the practice of stakeholder management and the
achievement of various corporate performance goals. The final stream, normative, is the fundamental basis and involves acceptance of two ideas: (1) “stakeholders are persons or groups with legitimate interests in procedural and/or substantive aspects of corporate activity.
Their interests in the corporation identify stakeholders, whether the corporation has any
corresponding functional interest in them. (2) The interests of all stakeholders are of intrinsic
value”. Beside these different aspects, Donaldson and Preston (1995), supported by findings of others (Berman et al, 1999, Hillman and Keim, 2001, and Ogden and Watson, 1999), state
that the view of stakeholder management and favorable performance of the company going
hand in hand, has become more commonplace within both the professional and academic
management literature.
Stakeholder Engagement
A component of stakeholder theory is stakeholder engagement, which is defined by ISEA (1999) as “the process of seeking stakeholder views on their relationship with an organization in a way that may realistically be expected to elicit them”. According to Greenwood (2007) SE can be understood “as practices that the organization undertakes to
involve stakeholders in a positive manner in organizational activities”. There are various levels and ways of engaging stakeholders in the process of business (Gao and Zhang, 2001,
12
Pedersen (2006) talks about a continuum between low and high levels of engagement
on different dimensions: inclusion, openness, tolerance, empowerment and transparency. For
the first dimension, which is inclusion, a low level of engagement entails having a dialogue
with only a few privileged stakeholders, and a high level involves all relevant stakeholders.
Openness knows a low level of engagement when a dialogue is structured around a fixed set
of questions, as opposed to open questions. The third dimension is tolerance, whereby in a
low level of engagement, one position has priority over all the others. A high level of
tolerance entails that alternative and critical voices are respected. Empowerment is the degree
to which one stakeholder dominates the dialogue and decisions, as opposed to freedom and
equality in dialogue as well as in decisions. The final dimension in which the level of
engagement can differ is transparency. Within transparency, a low level of engagement
means there is no access to information about the process and outcomes of the stakeholder
dialogue. A high level of engagement concerning transparency entails full access to the
information. I will come back to these dimensions later, because the possibility to score high
on these levels of engagement may be more difficult if the degree of multinationality
increases.
Besides different dimensions and the related levels of engagement, Pedersen (2006)
talks about phases of stakeholder dialogue and related filters. Three filters precede the three stages. The first filter, the “selection filter” is concerned with the access to the dialogue. From all the possible stakeholders, a selection is being made regarding who can participate in the stakeholder dialogue, which is the first phase. The second filter, the “interpretation filter” precedes the decision phase. This filter is the transformation of multiple voices from the
dialogue into a limited number of decisions. Stakeholder dialogue is a difficult process and it
may not be possible to satisfy all stakeholders. The third and final filter Pedersen (2006)
13
decisions move out of the dialogue arena. There is a difference between preferred actions and
final outcomes of the SE process. The same is the case in this instance as it is for the levels of
engagement. The higher the level of multinationality, the harder it may be to satisfy all the
original stakeholders that were present before the selection filter. I will come back to this
later.
According to Noland and Phillips (2010) the topic of SE becomes more significant
and interesting. The difference between engagement and interaction needs a clear distinction.
Stakeholder interaction alone is no longer sufficient, interaction in this perspective can be
transacting without inquiring as to his or her wants, needs, wellbeing or capabilities. SE, on
the other hand implies, at minimum, recognition and respect of common humanity. Because
companies have an effect on persons and communities, companies must identify their
stakeholders and communicate with them.
In Table 1, different ways of conceptualizing SE are presented. All these possibilities
show some ranking of high or low SE, or attributes that are signs of higher SE. Gao and
Zhang (2001) divide stakeholder engagement into four clearly defined levels of engagement.
The first level is the ‘passive’ level of engagement, where stakeholders are merely given information. The number of stakeholders here is large and the information comes via public
media and through public reports. The second level they denote is ‘listening’, wherein, as the name suggests, stakeholders are consulted. This level includes a selected number of
stakeholders. The third level of engagement is the ‘two-way process’, which entails stakeholders engaging in dialogue with the company. Just as for the ‘listening’ level, there is
a limited number of key stakeholders. The fourth and maximum level of SE according to Gao
and Zhang (2001) is the “proactive” level. Within this level, stakeholders drive management. Examples of stakeholder approaches in this level are the creation of a stakeholder council,
14
allowing stakeholder representatives in management and asking stakeholder verification of
social report.
Levels of SE Critique
Gao and Zhang 2001.
They divide engagement into four levels: 1.passive (merely giving information), 2.listening (stakeholders are consulted), 3.two-way-process (stakeholders engage in dialogue with company), and 4.proactive (management is driven by stakeholder).
Clear distinctions between the different levels.
Question remains whether a level 0 could be present, which would mean that no information would be given.
Friedmand and Miles (2006)
Quality of SE(management) comprised by 8 categories: 1 and 2, manipulation and therapy. 3,4,5, informing, consultation and placation. 6, partnership, 7 delegated power, and 8 citizen control.
Is more applicable to established stakeholder engagement processes. Less concerned with the degree of SE.
Michelon (2011) Ordinal scale using the four criteria of GRI ranging from low to high engagement.
No specification what low engagement and high engagement entails. AA1000 According to AA1000 the proposed
engagement consists of five components: planning, accounting, auditing, reporting and embedding. Defines quality SE by some criteria that is must meet. Clearly defines scope, agreed decision-making process, transparent, etc.
These are useable guidelines for managers. Components do not show a clear distinction of low or high engagement.
Ayuso et al 2011 Two questions: one addressing the number of different external stakeholders the company regularly addresses through satisfaction surveys or perception studies (stakeholder scope) and one question about the number of mechanisms used by the company for engaging with these stakeholders (stakeholder process).
Score on stakeholder scope and stakeholder process could be dependent of many other factors like firm size, industry, country, policy, etc.
Manetti (2011) Uses GRI reports
Companies have a series of questions which they must answer using the GRI reports.
GRI reports are used on a broad scale and are
considered reliable because of independence of a possible GRI check.
All literature seems to denote some sort of ranking from high to low SE, although
some describe the different levels more specifically. In this study, the division as is shown by
Gao and Zhang will be used in the analysis, because with a raise in the level of SE, a clearly
defined extra attribute is added each time.
15
Behind the degree of SE from the company’s perspective may lay different motives.
A division of three kinds of motives is being made on the continuum between the normative
stream on the one side, and instrumental stream on the other side. As I will explain below, the
first two, being present in the philosophical reasoning of Jürgen Habermas represent moral
and strategic stakeholder engagement. The last motive denoted here, entails the ‘ethical strategist’. Moral and political philosopher Jürgen Habermas argued that engagement must be largely free of any strategic motivation, in order to ensure its moral legitimacy. This theory
lies within the normative stream of SE, as it denotes the importance and the justification of
SE. Scholars who comply with this reasoning, also called Habermasians, can be divided into
two camps: A school supporting moral stakeholder engagement on the one side, and a school
supporting strategic stakeholder engagement on the other. The first, moral engagement is
“marked by specific conditions of communication which ensure that the communication is uncorrupted by power differences and strategic motivation. The aim of this type of
engagement is agreement for the sake of agreement” (Noland and Phillips, 2010). The latter,
strategic stakeholder engagement is undertaken with strategic motivations. This does not
mean that these intentions are necessarily dishonest or malicious. Although this is part of the
normative stream, an instrumental aspect comes into play here, as it is interested in how SE
can enhance the achievement of a company’s goals. Whereas the followers of Habermas are
an important group within stakeholder theory, another school of thought consists of scholars
that can be called Ethical Strategists (Noland and Phillips, 2010). They state that honest, open
and fair engagement of stakeholders is necessary for a business to function properly. Whereas
the moral camp of Habermasians lies within the normative stream of SE literature, the Ethical
Strategists lies within the instrumental stream, because the SE should lead to the achievement
16
engagement and ethical strategists are both at the end of the continuum of why stakeholders
should be involved and that the strategic side of Habermasians is in the middle of these two.
As Rawls (2000, in Godfrey 2005) denotes, different stakeholders and individuals are
influenced by different doctrines, which provide them with definitions of what a good society
should be. This means that their beliefs on this matter may well be different from someone
else’s. In accordance to this, Godfrey (2005) states that a firm’s “public” consists of multiple communities, each representing different ethical values and value systems. Few ethical values
will be common across all communities, meaning that these values and preferences of
communities may very well conflict with each other. These different ethical values may pose
challenges when internationalizing. Besides this notion of multinationality, Greenwood
(2007) states that stakeholder engagement is a morally neutral action, though it may be
related to corporate responsibility. But SE in itself may be morally neither good nor wrong.
In line with the authors mentioned above, the goal of this paper is not to put a moral mark on
SE. The focus will be what the effect of SE will be, in this case on CSP, without making a
moral judgment regarding the justification of engaging stakeholders in the process of
business. Besides this, I argue that for the sustainability of the SE process a strategic
motivation may be preferable as this may entail a win-win situation, which will prolong and
mature the stakeholders’ engagement process. I will thus follow an instrumental approach,
and write this paper while slightly taking the role of an ethical strategist.
Corporate Social Performance
Stakeholders are increasingly becoming concerned about the corporate social
performance (CSP) of the company’s operation. Consumers may prefer eco-labelled products
17
screens are used by investors and corporations develop socially responsible purchasing
practices to promote more sustainable supply chains (Chen & Delmas, 2010). The
above-mentioned examples are just a part of the increasing number of expressions of CSP concerns
of stakeholders. CSP has been a topic in businesses and literature since the mid-1970s
(Wood, 1991), but the intellectual roots can be found, in the 1950s, in Boulding’s (1956, in Wood, 2010) view of complex organizations as open systems, intricately connected to their
larger environment. This was contradictory to the general opinion in those days, whereas
organizations were viewed of as closed systems. In accordance with this new view of the
organization, CSP is concerned with the harms and benefits generated by the organization’s
operations by interacting with its environment. This is reflected within different dimensions
of the surrounding; for instance social, cultural, political, economic and natural dimensions
(Wood, 2010).
According to Turban and Greening (1996), CSP can be defined as “a construct that emphasizes a company’s responsibilities to multiple stakeholders, such as employees and the community at large, in addition to its traditional responsibilities to economic stakeholders”. In addition to this definition, Mitnick (2000) states that the distinction should be set out
clearly between social performance and social action, whereas the first entails outcomes
(impacts) and the second represents outputs (activities). The definition that is given to CSP
by Turban and Greening (1996) seems to represent the latter of this distinction, the activities
of the company, not as much as the outcome. According to Clarkson (1995) a definitive
conclusion of what CSP exactly entails is not established yet, and Wood (2010) adds to this
that the CSP domain has remained controversial and fluid. But Clarkson (1995) states that
there has been some agreement with the definition of Preston (1988), which emphasises the
outcomes of corporate behaviour instead of the action. In this paper, CSP will thus be
18
Mitnick (2000, in Mattingly and Berman) in consideration. In line with this consideration, the
view of Preston (1988, in Clarkson, 1995) will be the focus or definition of CSP in this case;
it states that CSP is ‘the impact of business behaviour on society’.
A good CSP has been argued to have many effects. An important notification here is
the word ‘argued’, because for many of the relationships concerning CSP, opinions are divided into camps of believers and non-believers. First, it has been argued to enhance the
company’s performance, for instance because of the company’s engagement in cause-related marketing, corporate philanthropy, and green marketing. Marketers point to advantages it can
have on the company’s reputation, customer loyalty and customer-company identification
(Luo and Bhattacharya, 2009). Second, besides the somewhat obvious view from a marketing
perspective, a large part of the CSP literature focuses on the relationship between CSP and
firm financial performance. Unbelievers in CSP state that a company should not engage in
this because it demands resources that the company could better allocate otherwise. On the other hand, many scholars believe the adage of “Doing well by doing good”, and thus see a positive relationship between CSP and firm financial performance (Orlitzky, 2011). In
accordance to this finding, Wood (2010) carefully states that good social performance results
in a better bottom line for the company, and vice versa, bad performance is likely to result in
financial harm, but she denotes that it is hard to measure what the exact impact is of CSP on
firm financial performance. For instance, there is no direct payment from charity or CSR
initiatives because these may be translated into financial return of investment through
reputation, or acquisition of NGO alliances or pacification of potential belligerent claimants.
Next to this notion, Wood (2010) states that CSP can have an impact on a varied set of
variables, which can be seen as essential to the company’s survival. Some of these she denotes are reputation, innovation and access to needed resources.
19 Stakeholder Engagement and Corporate Social Performance
According to Greenwood (2007) stakeholder engagement is much associated with
corporate responsibility. It seems logical that if an organization shows more commitment to
its stakeholders and lets them get involved in the process of business that it has a responsible
attitude towards their stakeholders. Greenwood (2007) states that this is a myth and a
simplistic assumption that SE would necessarily increase the corporate responsibility. More
engagement is not the same as being more accountable and responsible towards the
company’s stakeholders. She also states, justly, that SE knows many goals, and can be used
as a mechanism for different desired outcomes. This assumption makes Greenwood’s (2007)
central argument that SE is a morally neutral practice. A distinction she makes in her model
is between SE (high engagement is where the stakeholder-involved activities are numerous or
of high quality) and stakeholder agency (whereas low stakeholder agency means involving a
single of few stakeholders). Different combinations and ways of engagement can thus be
characterized by quantity of the stakeholder-involved activities and quality of these
stakeholder-involved activities.
Wood and Jones (1995, in Wood, 2010) theorized, by examining 65 studies
concerning the relationship between CSP and financial performance, that within this
relationship, stakeholders play at least four different roles of importance. Stakeholders are the
source of expectations, thus constituting desirable and undesirable firm performance. They
experience the effects of corporate behavior, and evaluate these. And finally they act upon
their interests, expectations, experiences, and evaluations. Stakeholders thus notice much of
the CSP and enact upon these notifications. Taking the instrumental approach in
consideration it is thus of importance from a strategic stance that stakeholders are engaged in
order to perform in a way that positively influences them. By engaging with stakeholders, the
20
According to Valor (2005) corporate social responsibility and stakeholders
complement and reinforce each other. The needs and social demands of different stakeholders
vary and change over time. Consequently, good interaction and engagement with stakeholder
gives the company information about what the needs of the stakeholders are, and thus
increases the chance to perform socially well and improve the CSP.
Corporate citizenship entails for a great part the involvement of relationships with all
important stakeholders (Carroll, 1998). In accordance to this, both corporate citizenship and
CSR propose that companies should be controlled by society and not only by shareholders.
This is in conflict with the neoclassical view that the companies should only satisfy their
shareholders (Valor, 2005). Thus, if the responsibility of the company goes beyond
shareholders satisfaction according to CSR theory, it is undeniable that stakeholders
constitute a large part of the corporate social performance. The question remains, whether the
influence of SE has a neutral (according to Greenwood, 2007) or a positive influence
(according to Valor, 2005).
The process of SE does not stop at interaction, as is explained by Noland and Phillips
(2010), instead, it is a process that ends in outcomes. Taking this in consideration, the
outcome of this process of SE results partly in the outcomes of the company, which
determines the degree of CSP. From the perspective that the stakeholder’s social demands
and needs change over time, and are different per place, SE may enhance companies to
capture these changes. An important notion should be made here that CSP, as is stated earlier,
is regarded here as the outcome of the company’s performance, not just the outputs.
The conclusions of these authors lead to the assumption that SE will enhance the
company to positively influence the CSP outcomes. Thus, taking all of the above mentioned
21
consequence to this, the score on corporate social performance will increase. This leads me to
the following hypothesis:
H1: More stakeholder engagement leads to a higher score on CSP
Multinationality
There are indications that effects of stakeholder engagement may be complicated by
the exposure to a wide range of stakeholder views, which may be due to increased
internationalization. According to Yang and Driffield (2012), alongside many other authors,
multinational companies have opportunities to achieve greater returns from international
exploitation of intangible assets. Advantages resulting from being a multinational company
may be benefits of internalisation, including economies of scale and scope, and the ability to
relocate activities to reduce costs. Further, by being multinational a company has the
mechanisms at its disposal to allocate resources more efficient, through the creation of
intra-firm markets when intermediate markets are missing (Yang and Driffield, 2012). An effect of
these advantages is that costs can be lowered and productivity increased, leading to increased
financial performance of the company (Buckley and Casson 1976; Rugman 1986; Dunning
1988; Tallman and Li 1996; in Yang and Driffield, 2012). Of course, companies can also
experience disadvantages from multinationality, for example from increased coordination and
management costs and cultural diversity. In addition to this, according to Lu and Beamish
(2004) a multinational enterprise’s (MNE) performance will decrease when they reach a certain high degree of internationalization. They propose an S-curve, stating that with early
and mid-stage internationalizers the performance will increase, partly as an effect of the
22
their internationalization they experience a decline in their performance, as an effect of
cultural distance and coordination costs of very dispersed markets (Lu and Beamish, 2004).
Hennart (2007) states that the multinationality-performance literature has made two
predictions. First of all, the more internationally diversified an MNE is, the lower the risks.
Second, generally the more internationally diversified an MNE, the greater its profitability.
Another aspect of multinationality is the degree of local responsiveness or integration. MNEs
that choose to be locally responsive find it of importance to listen to local markets and along
with it, give subsidiaries in countries or areas thus some freedom to act on different and
changing demands within their markets. If, on the other side, a company chooses to have a
high degree of integration, this can positively influence cost reductions and there is much
control from HQ over subsidiaries. These choices are thus twofold, and lead to four types of
MNEs: (1) Centralized exporter, with a low degree of responsiveness and an high degree of
integration; (2) International projector, with both a low degree of responsiveness and
integration; (3) International coordinator, having a high degree on both areas and (4)
Multi-centred MNE, having a low degree of integration and a high degree of responsiveness
(Bartlett and Ghoshal, 1989 in Harzing, 2000).
All these aspects of multinationality have an effect on the company’s performance.
Although what the effects precisely are may be debatable, but that the aspects of
multinationality have any effect is universally accepted. Logically the different aspects and
effects of multinationality may also have its effect on smaller parts of the operation, in this
case, corporate social performance (CSP). Perrini et al (2007) studied the differences and
similarities of large companies and medium and small companies. They found that larger
companies find it more relevant to define and implement CSR strategies. Besides, small
companies make less use of CSR instruments than large companies do. Large companies are
23
and reporting strategies. Whether the degree of multinationality or other aspects of it
influence these differences is not clear. Further, the impact of CSP on a company’s
multinationality is studied (Bouquet and Deutsch, 2008), and also the relationship between
CSR practices and subsidiary autonomy is investigated (Muller, 2006), but these are not
comprehensive studies concerning the relationship between stakeholder engagement and
CSP, and the effect that multinationality has on this relationship.
Multinationality, Stakeholder Engagement and CSP
Multinationality or geographical diversification can have a great effect on the process
and outcomes of doing business. Multinational enterprises(MNEs) may have more difficulty
to identify the right issues. This stems from a few different attributes that come hand in hand
with doing business across national borders. First of all, the company has more actual and
potential stakeholders, just because it is present in more countries. Second, certain
transnational organizations must be taken into account which may not be the case when a
company only acts domestically. Also in this case, when operating in more countries,
logically more transnational organizations are stakeholder to the company. For example, if a
company operates in the different countries in Europe it has the European Union as a
stakeholder. When the company decides to go more abroad, for instance to the South
America, it will get an extra transnational institution to deal with, for instance the USAN, the
political and economic union of the South America’s. Third, even if a MNE engages in essentially the same basic activity in each of the various countries, it is very likely that each
country proposes a different set of stakeholders to the company (Nigh and Cochran, 1987).
According to Pedersen (2006) stakeholder influence or engagement goes through three filters. The first filter is the ‘selection filter’, as the name suggests, this is where the selection is being made regarding which stakeholder gets influence in the next phase, the ‘stakeholder
24
dialogue’ phase. When, in accordance to Nigh and Cochran (1987) the amount of countries a company operates in gets larger, the number of stakeholders increases,| this automatically
means that there is a larger pool to select from and it may be harder to distinguish which
stakeholders are of most important. The next filter is the interpretation filter; this concerns
‘the transformation of the multiple voices from the dialogue into a limited number of decisions’. If the multitude of voices is bigger, it is logically more difficult to translate these into decisions best suitable for most stakeholders. In accordance to this, Husted and Allen
(2006) state that diverse stakeholders and conflicting values between local and global social
needs, as a consequence of being more multinational, demands a more complex CSR
strategy. The final filter, called the ‘response filter’, is concerned with the activities that take place when the decisions move out of the dialogue area. It then is being influenced by local
interpretations, conflicting interests and organizational changes (Pedersen, 2006). Al of these
may influence and determine the outcomes of the SE and thus determine the corporate social
performance.
The notion that Godfrey (2005) made regarding different ethical values of stakeholders and stakeholder’s communities, also has its consequences when the degree of multinationality increases. When the degree of multinationality increases, logically more
stakeholders and stakeholder’s communities are involved by, or may involve, the process of doing business. When this amount thus increases, and alongside it the moral preferences will
become more varied, it will make it harder for a company to enact upon these more and more
varied opinions about what is the right way to do business.
As is described earlier, according to Lu and Beamish (2004), when MNEs continue to
increase their internationalization, they experience a decline in their performance, as an effect
of cultural distance and coordination costs of very dispersed markets. This may very well also
25
influence CSP, the presence of more stakeholders and a more diversified group of
stakeholders may have its negative effect on the corporate performance, just like the S-curve
theory of Lu and Beamish (2004).
When taken the above mentioned aspects of multinationality in consideration, the
degree of multinationality thus may, logically, influence the positive relationship between SE
and CSP. I argue, in accordance to these notions, that when the degree of multinationality
increases, and with it the number and diversity of stakeholders, it will be harder for
companies to retain the positive relationship between stakeholder engagement and corporate
social performance. This leads me to the following hypothesis:
H2: The degree of multinationality negatively moderates the positive relationship between stakeholder engagement and CSP.
26
Methodology
In this section, the methods of this study are being discussed. First, I will explain how
the selection of the sample took place, secondly, I will discuss several techniques to measure
the variables that are of importance in this paper. This section will end with the elaboration
on why certain analysis techniques are chosen.
Sample
To test the hypotheses, I developed a dataset based on 135 firms. In order to have a
useable sample large companies of which the most are active in several countries should be
included. Taken the variables into consideration a sample was collected from the S&P 500
list of 2012. This listing holds many multinationals and has a wide variety of industries. In
order to prevent that industry characteristic influence the outcome of the research I choose to
use a stratified sampling method. Nine of the most common industries are taken from the
S&P 500, being the following in random order: consumer discretionary, consumer staples,
energy, financials, health care, industrial, information technology, materials and utilities.
From each industry fifteen companies are randomly selected, which results in a sample of
135 companies. By applying this sampling method, certain industry related characteristics
27 Corporate Social Performance
Measurements of firm’s financial performance are readily available, think of return on assets or return on investment. The CSP counterparts measurements for performance in this
area are not this readily available. This notion can be ascribed in part to the more qualitative nature of CSP, whereas the focus is somewhat more on ‘soft’ measures instead of ‘harder’ measures. Several authors have described the challenges associated with measuring CSP
(Carroll 1999, Graves and Waddock 1994, Chen and Delmas 2010). Academic researchers
have measured CSP using many different approaches, for instance using survey
questionnaires, content analyses of annual reports, expert evaluations, and regulatory
compliance data. According to Graves and Waddock (1997), the difficulty with measuring
CSP lies in its multidimensional construct. Behaviors in this construct have a variety of
inputs, entail internal behaviors (treatment of employees) and external behaviors (community
behaviors). Besides this, it also occurs on a wide variety of industries, with different histories,
cultures and characteristics. Lastly, what makes it harder to measure is the absence of a
universally accepted definition of the construct CSP. In the past, reputation of the public was
used as an indicator of CSP (Aupperle 1991, in Graves and Waddock, 1997), though this may
be debatable taking into consideration that it actually only measures appearance, and not true
performance. At this moment, the most commonly used database for assessing CSP is the
KLD database (Chen and Delmas 2010, Turban and Greening 1997, Graves and Waddock,
1994, Johnson and Greening 1999, Griffin and Mahon, 1997), and it is widely used in
different prominent magazine such as Journal of Business Ethics, Business and Society, the
Academy of Management Journal, and many others (Chen and Delmas 2010). This widely
used and accepted way of measuring CSP by using KLD databases shows that it is an
accurate and sufficient measurement and will thus be used in this paper too. KLD rates
28
in this research the following dimensions will be of importance: Corporate governance,
community, diversity employee relations, environment, human rights and product. Although
using KLD to measure CSP is widely accepted, the way to measure it has differed in the past.
Whereas at first, researchers have measured it by detracting the strengths from the concerns
and thus obtaining one overall score for CSP (Turban and Greening 1996). Collapsing KLD
scores into a unidimensional index may mask the individual dimensions that are of relevance
to measuring CSP. It is thus not a proper way to compute all the different dimensions into one
single score (Johnson and Greening 1999). According to Mattingly and Berman (2008),
positive and negative social actions are both empirically and conceptually distinct construct.
For this reason, these constructs should not be combined. This means that detracting negative
from positive scores and thus creating one overall score for CSP is not accurate. In
accordance to these findings, and the findings of this research, which will be elaborated upon
later, the choice is made to keep the CSP strengths and concerns of the KLD database
separate, and thus perform analysis with these separately.
For each company the KLD-index divides scores in groups of strengths and concerns.
Taking into consideration that I use both KLD
scores from 2011 and 2012 this will both be
shown in this section. In the sample of 2011, the
number of KLD strengths a company has can
have a minimum of 0 and a maximum of 20 (M
=7.02, SD = 5,25). Figure 1 shows that the CSP
Total Strengths of the companies from the
sample are not normally distributed. The
Shapiro-Wilk test shows that this too (W(131) = .931, p < .05). A high frequency is noticeable on the
29
low scores, it could thus be stated that the distribution of the strengths in 2011 is skewed to
the right.
Regarding the CSP Total Concerns of 2011, the following is the case. The minimum
score a company from the sample of 2011 could
get was 0, the maximum CSP Total Concerns
would be 15 (M = 3.64, SD = 2.62). In both the
CSP Total Strengths and the CSP Total
Concerns, the variance exceeds the mean. This
will be of importance when analysis of the data
is being done. The CSP Total Concerns of 2011
are distributed as figure 2 shows. As was also the
case with the CSP Total Strengths of 2011, also here a distribution can be noted that is
skewed to the right. The Shapiro-Wilk test shows that there is no normal distribution (W(131)
= .882, p < .05). Taking in consideration that both are not normally distributed, means that
this should be taken into account when
analysing the data.
In the sample of 2012, the number of
KLD strengths a company can have a minimum
of 0 and a maximum of 17 (M =7.02, SD =
5,25). Figure 3 shows that the CSP Total
Strengths of the companies from the sample are
not normally distributed. The Shapiro-Wilk test
Figure 2: Distribution of KLD concerns. 2011
30
shows that there is no normal distribution (W(131) = .954, p < .05). Although less skewed to
the right than a year earlier, high scores are noticeable between 0 and 10, and hereafter, the
frequencies of the scores are lower.
Concerning the CSP Total Concerns of 2012 the
following can be stated. The minimum a
company can score here is 0, the maximum a
company can score is 13 (M = 2.15, SD = 2.4).
Figure 4 shows the distribution for the CSP
Total Concerns of 2012. The Shapiro-Wilk test
shows that there is no normal distribution
(W(131) = .814, p < .05). What can be noted here, as
was also the case in the previous distributions, that the distribution is skewed to the right.
This being constantly the case with the distributions, it may imply that similar analysis
technique would be sufficient in these cases.
Stakeholder Engagement
A clear and universally accepted measurement of SE is not present in the literature. Even
more, a quantifiable measurement of SE seems to be absent. Ayuso et al (2011) used two
questions to establish the degree of SE; one addressing the number of different external
stakeholders regularly addressed through satisfaction surveys or perception studies
(stakeholder scope) and one question about the number of mechanisms used by the company
for engaging with these stakeholders (stakeholder processes). Taking the time and
possibilities of this study in consideration makes this method not appropriate. Manetti (2011)
31
uses GRI reports. Although these reports are regarded as being trustworthy, the problem is
their availability. Some company’s GRI reports are available, but not all. A second point of
attention here is that this not necessarily entails a quantifiable measurement, for GRI reports
in itself do not give available usable scores of SE. Several authors(Gao and Zhang, 2006;
Friedman and Mills, 2006; Michelon, 2001), do make a ordinal scale out of SE, most of them
ranging from low to high SE. In this study, the ordinal scale of Ghao and Zhang (2001) will
be used as a guiding scale. What the different levels exactly denote will be explained below,
but it should be stated that this scale is not yet a measurement in itself. It entails a possibility
to qualitatively sort companies into one of these four levels of SE.
The strategy used in this paper was to make use of annual reports, CSR reports, other
reports or other information from the website. From these sources, clues of evidence were
being found of the degree of SE. Some examples that linked different quotes to SE are the
following:
SE level Exemplary quotes / explanation
SE level: 1 a company would fall into this category when there was no mentioning of stakeholder engagement or anything of its kind. This was the lowest
stakeholder category because companies in the S&P 500 are prohibited to
provide information about the achievements and actions of the company.
SE level: 2 “I would like to hear from even more voices throughout the company and
from outside stakeholders”
“Closing, I assure our stakeholders that we are committed to balancing ... challenges, and we welcome your thoughts and engagement on these”
32
suppliers, nongovernmental and nonprofit organizations and professionals in
industry, government, labor and education in a variety of informal and formal
communications."
"We routinely meet with external stakeholders to help us identify issues most
material to our business operations and to maintain our license to operate."
SE level: 4 "Positive relationships with employees and business partners help us to improve efficiencies, cost and quality, and allow us to develop and to
innovate. Effective two-way communication with our customers, dealers and
other stakeholders helps us to understand and deliver the products that
customers want"
“We believe active stakeholder engagement is a critical driver of performance and our continued improvement and success.”
Table 2: Examples of SE levels
Stakeholder engagement thus entails the independent
variable in this research. SE is being measured on the
above-described scale of 1 to 4. The score entail the following, 1: the
company has a passive attitude and merely gives information 2:
The company actively listens to its stakeholders, they are
consulted, 3: there is a two-way process; stakeholders are engaged
in dialogue with the company, and 4: a proactive attitude is taken
by the company and management is driven by stakeholder’s input.
By using annual reports, CSR reports, other reports and/or information on the websites,
companies are categorized into one of these four categories. The data is being collected for Category Frequency 1 39 2 13 3 66 4 15 Total 133
33
the year 2011, which means that information of 2012 was not in the data collection. The
following descriptive statistics are of importance to the SE. Of the total sample of 135,
information regarding SE could be collected for 133 companies. A minimum score could be
obtained of 1, the maximum score was 4 (M = 2,43 ST = 1,03), in accordance to the
categories explained above. In Table 3, frequencies are presented. What shows from this table
is that most companies engage with stakeholders in a two-way process, this is nearly half of
all the companies. Hereafter, the most common degree of SE is merely giving information.
What can be stated reasoning from this table is that companies most of the time seek input
from stakeholders. If they decide to actively interact with stakeholders, they do so in active
way but their decisions are most of the time not driven by stakeholders.
Multinationality
Multinationality can be measured in many different ways. Some examples are ratio of
foreign assets to total sales (Daniels and Bracker, 1989, Ramaswamy, 1989, in Kotabe,
Srinivasan and Alaukh, 2002), subsidiaries in foreign countries (Tallman and Li, 1996) or
foreign to total employment ratio (Kim, Hwang and Burgess, 1989, in Rugman, Yip and
Jayaratne, 2006). Though many legitimate ways of measuring multinationality exist,
according to Rugman, Yip and Jayaratne (2006), the majority of the studies since 1970 have
used foreign to total sales as an indicator of multinationality. Because of this widely used
mechanism, and the convenience of this measurement taking the time and possibilities of this
34
Multinationality is measured here, as is explained here above, by the ratio of foreign
sales to total sales. This measurement results in a continuum scale of 0 to 1, whereas 1 entails
maximum multinationality, whereby all sales are made in foreign countries, and 0 entails a
company that only sales in its domestic
country, in this case representing minimal
multinationality. In this sample, scores could
be taken from 131 companies, and
multinationality can score at minimum 0 and
at maximum 0.85 (M = .349, SD = .260).
When taking figure 5 into consideration, one
thing is stands out, this is the overrepresentation of companies with a multinationality score
of 0. When doing a normality test, the Shapiro-Wilk test, shows that the distribution is not
normal (W(131) = .916, p < .05). When the companies with a score of 0 are removed from the
sample, a normal distribution is present. The
Shapiro-Wilk test shows this to (W(101) =
.977, p = .079). In figure 6 it is also shown,
that when 0 scores are removed from the
analysis, the scores are normally distributed.
But in this paper, the option is being chosen to
include all 131 companies into the analysis.
Figure 3: Distribution of MN scores, 2011
35 Control variables
With the selection of the sample, the influence of the type of industry on CSP is being
is being minimalized, because the nine largest industries are equally present in the sample.
Other company characteristics that might influence CSP may be company size and profit
(Stanwick and Stanwick, 1998). These two are being taken into consideration. Company size
is measured by taking the total amount of assets of each company, profit is measured by
obtaining the gross profit of each company using WRDS as a ratio to sales. The first control
variable, size of the company, is being measured by Total Assets, this can take a minimum
score of 1311 and a maximum of 2129046 (M=81233, ST = 265239). This variable is not
normally distributed (W(134) = .282, p < .05). Because it is highly skewed, a log
transformation is needed.
The second control variable is the ratio proft/sales, this can take a minimum score of ,03 and
a maximum score of 1 (M = ,4108, ST = ,21737). Also this control variable is not distributed
equally (W(134) = .960, p < .05).
Industries
In the sample, nine different industries are taken into account. Table 4 shows the
different values for the dependent and independent variables per group of industries. The
values are shown for the mean and the standard deviation. What shows is that ‘Health Care’ scores the highest on SE. Health care also scores high on CSP strengths of 2011, but what is
odd, is that the difference with CSP strengths 2012 is big, whereas in 2012 the scores are
much lower for CSP strengths. Concerning CSP Total Strengths of 2011, the industry that scores the highest is ‘Consumer Staples’. The ‘Financials’ industry scores the lowest on SE, with almost half the level of SE as the ‘Health Care’ industry average. It should be mentioned
36
that the sample of each industry is not of a sufficient size (N is 14/15) to make complete
accurate judgments, but it could be an opening for further research.
Table 4: Variables per industry
Regression Analysis
In this analysis a binomial regression analysis will be used to test the relationship
between the variables. Taking into consideration that the data of the dependent variable is
counted, the choice will be best made between a ‘Poisson regression’ and a ‘Negative binomial regression’. The second, negative binomial regression analysis can be regarded as a
Industry N Mean Std. Deviation
Consumer Discretionary SEscore 15 2,27 1,223 TS2011 15 7,8667 5,95059 TS2012 15 4,7333 3,93640 Consumer Staples SEscore 15 2,87 ,915 TS2011 15 10,0667 4,41534 TS2012 15 6,8000 3,07525 Energy SEscore 15 2,33 ,976 TS2011 14 4,4286 3,87724 TS2012 15 5,0667 3,76955 Financials SEscore 15 1,73 ,961 TS2011 15 5,4667 5,12510 TS2012 15 4,6000 3,60159 Health Care SEscore 15 3,20 ,561 TS2011 14 9,0000 6,66795 TS2012 15 5,1333 4,08598 Industrials SEscore 15 2,20 ,941 TS2011 15 4,9333 4,36654 TS2012 15 5,2667 4,58984 Information Technology SEscore 15 1,93 ,961 TS2011 15 7,2667 6,04113 TS2012 15 5,1333 4,15532 Materials SEscore 14 2,71 ,994 TS2011 13 7,5385 5,14159 TS2012 14 7,2857 4,06540 Utilities SEscore 14 2,64 1,008 TS2011 15 6,6667 3,51866 TS2012 15 6,2667 3,28344
37
generalization of Poisson regression and it has an extra parameter to model over-dispersion.
Over-dispersion is the case when the conditional
variance exceeds the conditional mean. When this is
tested in SPSS, table 5 is the result. What can be seen in
the figure is that the variance does exceed the value of
the mean. This leads to the conclusion that a negative
binomial regression analysis would be most appropriate
here to measure the regression between the variables. In
accordance to all what is said above, the regression
analysis will be performed four times, these analyses will not differ from each other except
that the dependent variable will be different.
Hereafter, a test will be performed to check reversed causality. It could be the case
that SE and CSP only co-vary instead of any causality between the two. By testing reversed
causality any statements regarding the relationship between the two can be made with more
certainty. Mean Variance TS 2011 7.02 27.58 TC 2011 3.64 6.88 TS 2012 5.57 14.83 TC 2012 2.15 5.81
38
Results
To test the hypotheses, correlation and regression analyses are performed using SPSS.
The results of these tests will be discussed in the following sections. First, the correlation
between the variables will be shown, hereafter, the hypotheses will be tested in more depth
using regression analyses.
Correlations
In this section, the correlation between the variables will be tested. The variables for
which correlation will be tested are SE, CSP, MN and the control variables. A distinction will
be made between the two different dependent variables, namely CSP strengths and concerns
of 2011 and CSP strengths and concerns of 2012. The independent variable stays the same, as
is mentioned earlier, this is the SE of 2011. By testing it this way, a statement could possibly
be made if the effect of SE on CSP shows delay. Because the variables are not normally
distributed, and the independent variable is of an ordinal scale, a Spearman’s correlation will
be used. In Table 6, the correlation is shown.
Table 6: Correlation between variables ** Correlation is significant at the 0.01 level (2-tailed) * Correlation is significant at the 0.05 level (2-tailed) 1. 2. 3. 4. 5. 6. 7. 8. 1.SE . 2. CSP TS2011 .52** . 3. CSP TC2011 .102 .25** . 4. CSP TS2012 .56** .68** .40** . 5. CSP TC2012 .35** .51** .63** .54** . 6. MN -.03 .09 .06 .04 .06 .
7. Total Assets (log) .18* .56** .44** .57** .53** .03 .