M
ASTERT
HESISMS
CB
USINESSE
CONOMICS: C
OMPETITIONL
AW& E
CONOMICST
HE
R
ELATIONSHIP BETWEEN
C
OMPETITION AND
C
ORPORATE
S
OCIAL
R
ESPONSIBILITY
: A
N
E
CONOMIC
R
EVIEW
Thierry F. Wetzel
11834544
thierry.wetzel@student.uva.nl
Supervision:
Leonard Treuren
Professor dr. Maarten Pieter Schinkel
Statement of Originality
This document is written by Student Thierry F. Wetzel who declares to take full
responsibil-ity for the contents of this document. I declare that the text and the work presented in this
document are original and that no sources other than those mentioned in the text and its
ref-erences have been used in creating it. The Faculty of Economics and Business is responsible
Abstract
Rising public awareness of global phenomena such as climate change and the corresponding
influence of special interests and activist groups have made corporate social responsibility
(CSR) concerns increasingly pervasive. “Ethical” behaviour in business has often been
at-tributed to firms’ and managers’ individual actions as well as to pressures originating from
internal stakeholders. At present, however, it is clear that also external pressures such as
product market competition act as a determinant of socially responsible behaviour. This
thesis delineates what is currently known about the relationship between competition and
corporate social responsibility. Existing literature indicates that increased levels of
compe-tition are positively correlated to corporate social responsibility, yet much of the evidence
is based on ambiguous measures of competition and CSR. This thesis discusses issues and
implications of current research and gives an overview of recent results in this strain of
liter-ature.
Contents
Abstract ii 1 Introduction 1 2 Views on CSR 4 2.1 Selective CSR Background . . . 5 2.2 Definition of CSR . . . 82.3 Objective Function Firms Maximise . . . 10
2.4 Is CSR Profitable? . . . 11
3 Measures of Competition and CSR 15 3.1 Measures Used . . . 15
3.2 Measurement & Data Concerns . . . 22
4 Competition and CSR 30 4.1 Qualitative & Theoretical Work . . . 32
4.2 Empirical Work . . . 37
4.3 What do we Know? . . . 57
5 Discussion 59 5.1 Gaps & Future Research . . . 59
5.2 How to go About? . . . 60
1
Introduction
In 1999, the European Commission (EC) notably condoned an agreement on the basis of the
public interest defence in cartel offences1. The agreement in question had the objective to
cease the production of washing machines below a certain energy efficiency level. In this
way, the reduction in energy consumption of washing machines would not just result in large
benefits for the environment, but also allow consumers to recoup the higher initial costs of
more efficient washing machines through reduced costs of usage within a reasonable time
frame. The EC has acknowledged the environmental benefits and cost savings for potential
customers and correspondingly allowed the agreement to take place.2
The rationale behind the public interest defence in cartel offences is about shielding
com-panies from competition; this should enable them to deploy resources to corporate social
responsibility (CSR) causes, which are, arguably, most beneficial in industries where the
quality of services to the public may decrease in response to competitive pressures. As a
result there has been a social push towards less competitive industries in order to benefit
society. This view, however, strongly neglects the positive impacts competition can have
on innovation, pricing, development and product differentiation. Without a strong
justifica-tion, allowing green cartels could provide leeway to companies that seek to disguise socially
undesirable, if not malfeasant, behavior behind a fac¸ade of positive deeds.
For this reason, gaining a comprehensive understanding of the effects that competition
has on CSR in different industrial context stands as an indispensable condition to effectively
determine when a public interest defence is likely to achieve a welfare-enhancing outcome.
1Under the Treaty of the Functioning of the European Union (TFEU) Article 101(1) prohibits “all
agree-ments between undertakings, decisions by associations of undertakings and concerted practices which may affect trade between Member States and which have as their object or effect the prevention, restriction or dis-tortion of competition within the internal market. . .”. There is, however, the exception laid out in Art. 101(3), which states that agreements “which contributes to improving the production or distribution of goods or to promoting technical or economic progress, while allowing consumers a fair share of the resulting benefit” can be exempted. This exception is known as the public interest defence in cartel offences.
From an empirical standpoint, only few studies have focused on how competition
re-lates to CSR. The reasons for this relationship remaining underexplored may partially be
attributed to endemic problems of endogeneity that add to methodological and data issues3
that arise when testing the effect of competition on firms' CSR practices. The extant
liter-ature has primarily been focused on CSR and its effect on corporate financial performance
(Vishwanathan, 2017; Margolis et al., 2009; Orlitzky et al., 2008; Margolis and Walsh, 2003),
whilst others have investigated how CSR is affected due to pressures imposed by a wide
vari-ety of stakeholder groups such as, but not limited to, governments, consumers and employees
(Aguilera et al., 2007; Barnett, 2007; Schuler and Cording, 2006; McWilliams and Siegel,
2001; Donaldson and Preston, 1995). Nevertheless, the driving forces of firm’s engagement
in CSR are not just dependent on internal factors such as pressures from consumers and
em-ployees, which implies that external market conditions will, to some extent at least, influence
CSR conduct. Subsequently, one can only understand how CSR conduct is shaped once the
relationship between competition and CSR has been scrutinised. Yet, the question of how
competition relates to CSR remains underdeveloped (Flammer, 2015; Hawn and Kang, 2013;
Van de Ven and Jeurissen, 2005).
At present4, only a relatively small amount of empirical work has tested the impact
com-petition exhibits on CSR conduct. In most cases this growing body of research has employed
the Herfindahl-Hirschman Index (HHI), as a measure for competition, and the Kinder
Lyden-berg and Domini (KLD) database to quantify CSR conduct on multiple dimensions relating
to the firms' diverse stakeholders, e.g. employees, the environment or investors. As
elab-orated in section 3 these measures can produce misleading results: as the results can only
be as robust as the underlying data and most of this information is self-reported by firms,
inconsistencies in the inputs' accuracy may lead to systematic differences in outcome. The
3CSR is a relatively new concept and the reliability of quantifying it is questionable as elaborated in section
3.
4The first paper found to build an econometric model of competition and CSR was in 2010. This will be
differences between the studies' findings are in turn likely to be driven by the variety of
mea-sures and methodologies employed, thereby obfuscates the effect that competitive pressure
have on CSR conduct. Some works have used fitted HHI values that account for the firm
being active in multiple industries in addition to using a different CSR measure derived from
the Asset4 database. As discussed in section 3, many of the concerns that the HHI
val-ues raise are rarely addressed. Especially in high-tech industries and platform competition,
market power may originate from small technological advancements that are not accurately
proxied by using concentration measures such as the HHI based on market shares, number
of employees and/or size of a company.
In light of these shortcomings and their implications for research, this study sets out to
demystify the relationship between competition and CSR, by building on work from
man-agerial and economic literature that aimed at identifying what aspects of CSR are profitable,
shedding light on the relationship between competition and CSR. Accordingly, the research
addresses the following questions: What does existing literature say about the relationship?
What are the issues with existing empirical evidence, and if yes what are the implications?
What future steps have to be taken in order to improve our understanding of the relationship?
Widespread CSR concerns have become increasingly pervasive due to the mounting public
awareness of global phenomena such as climate change and the resulting influence of
spe-cial interest agencies and activist groups, which include the World Wildlife Fund (WWF),
Greenpeace and the Sierra Club. Understanding the antecedents of CSR stands as a
precon-dition for designing strategies and policies that ultimately improve total welfare. Among its
antecedents, competition is expected to play a crucial role in shaping firms' CSR behavior
and the use of CSR as a competitive strategy. The nexus between competition and CSR is
thus not only theoretically relevant for economists and policymakers, but also for managers
who seek to understand the implications of CSR as a strategy in response to external market
This thesis contributes to the existing literature body by identifying the concerns that arise
when addressing the question of how competition and CSR relate to one another. Moreover,
the study investigates how these concerns have been addressed in current empirical work
and what the consequences of neglecting these concerns are. The main contribution is
sum-marising (1) what we know about the relationship, based on current empirical work, (2) what
aspects remain underdeveloped, and (3) how future research should address the question of
how competition relates to CSR.
The next section provides a brief overview of what CSR is and the conflicting views.
It will also briefly touch upon what firms maximise and whether profit maximisation is
achieved by employing CSR strategies. The core contribution of this thesis is to critically
assess the current empirical work conducted on the subject and provide an overview of what
evidence exists on the relationship of competition and CSR, which will be addressed in
Sec-tion 4. For that to be possible, however, it is necessary to provide an overview of how the
two main variables are measured. This will be addressed in Section 3. The final sections are
devoted to suggest avenues for future research and address what is needed to understand and
improve our understanding of said relationship.
2
Views on CSR
Providing a history of how CSR has been perceived and defined is not within the interest
of this research, it is, however, useful to provide a brief overview. This will illustrate to the
reader how difficult it is to measure and quantify a concept for which there are several
un-derlying paradigms and dimensions. For a comprehensive review see Carroll (2008, 1999);
2.1 Selective CSR Background
Prior to 1900 corporate contributions were discredited by many and such contributions were
legally bound to bring about benefits to the corporation (Muirhead, 1999). From the 1930s
the view has shifted and corporations started feeling pressure to effectuate social
contri-butions (Eberstadt, 1973). In the period leading to the 1950s there was a shift towards a
philanthropic responsibility of corporations and donations became the major aspect of
cor-porate social responsibility (Murphy, 1978). In 1991 Carroll5proposed the pyramid of CSR,
suggesting that it has as its basis the economic responsibilities, followed by legal, ethical
and philanthropic responsibilities. “Carroll's CSR domains and pyramid framework remain
a leading paradigm of CSR in the social issues in [the] management field” (Schwartz, 2017,
p. 504). Thus, CSR, currently, is seen as some combination of these dimensions: economic,
legal, ethical and altruistic responsibilities (Schwartz, 2017; Lantos, 2001; Carroll, 2008;
Carroll et al., 1991).
There are several views on underlying paradigms, motives and effectiveness of CSR
practices. Friedman proposed the shareholder view. He ends his essay by quoting a sentence
from his book: “[Social Responsibility is a] fundamentally subversive doctrine in a free
society, . . . , there is one and only one social responsibility of business — to use it resources
and engage in activities designed to increase its profits so long as it stays within the rules of
the game, which is to say, engages in open and free competition without deception or fraud”
(Friedman, 2007, p. 178). Thus, he indicates that the sole responsibility of the agent is to
the principal, which usually seeks profits. In this view, there is no responsibility beyond the
one to the principal, or shareholders. He claims that obligations of managers is to maximise
5The concept of CSR has been taken one step further to corporate social performance (CSP). Carroll (1979)
defines corporate social performance as a broader version of CSR i.e. a model of corporate social performance must include (1) a definition of CSR, (2) the dimensions in which corporations have responsibilities and (3) are companies reactive or proactive? (Carroll, 1979, p. 499) For the purpose of this research it is sufficient to look at the concept of CSR as all empirical work on the relationship of competition and CSR has been conducted using the concept of CSR not CSP. In section 2.4, however, CSP will occur.
shareholder value, in line with classical economic agency theory. However, by explicitly
calling out CSR as a measure that redirects resources from profitable strategies he could not
have accounted for recent trends in consumer preferences. Research into the awareness and
willingness to pay for sustainability is showing that in modern times consumers care more
for companies that went green and/or are involved in social activities.6
Contemporary much focus is on a different paradigm, namely the stakeholder approach.
Freeman et al. (2004) puts forward that “certainly shareholders are an important constituent
and profits are a critical feature of this activity, but concern for profits is the result rather than
the driver in the process of value creation” (Freeman et al., 2004, p. 364). He claims that
economic value is created by the interactions of individuals that get together on a voluntary
basis to elevate “everyone's circumstances”. Economic value is never explicitly defined but
comes closest to what economists perceive as total welfare. Thus, the process of improving
“everyone's circumstances” i.e. stakeholders, profits will follow. Caring for employees and
communities can ease the ways for organisations to conduct their daily activities.
Corre-spondingly, this can result in cost savings, productivity gains as well as talent attraction and
retention (Greening and Turban, 2000; Turban and Greening, 1997).
Even though stakeholder theory and CSR are related, Vishwanathan (2017) stresses the
difference amongst the two. That is to say, CSR literature is concerned about
understand-ing the firms behaviour with regards to the social perspective, whereas stakeholder theory
concerns the managerial perspective to establish that the CSR decisions undertaken are also
profitable for the firm. This does not, however, exclude the responsibilities a firm has to
society and especially its stakeholders, leaving stakeholder theory an important aspect in
understanding into the effects of CSR strategies and possibly into CSR activities. As
com-6Research into the awareness and willingness to pay for sustainability is showing that consumers and
em-ployees find it difficult to know the full range of CSR conduct as data is difficult to access (Lyon and Maxwell, 2006). Vermeir and Verbeke (2006) and Manaktola and Jauhari (2007) show that as consumers become aware the willingness to pay increases and Schuler and Cording (2006) and Madsen and Rodgers (2015) on how awareness of stakeholders of substantive CSR practices can enhance productivity and reduce costs.
petitors can be seen as stakeholders, it is reasonable to build upon stakeholder theory to
understand how competition and CSR are related to one another.
Moreover, the different concepts of CSR can be subdivided into three subgroups: ethical,
altruistic and strategic7 CSR (Lantos, 2001; Bansal and Roth, 2000). The former represents
that firms behave morally acceptable and e.g. do not injure employees; this form of CSR
is broadly accepted and does not require any future analyses as it is encompassed in the
economic and legal responsibilities (Lantos, 2001). In contrast, altruistic CSR, which comes
closest to the CSR Friedman opposes in his 1970 essay, has a voluntary aspect to it and is
conducted with no concern for profit. It is about solving social issues that are not caused
by the firm and should thus be handled through a taxation system rather than companies
direct contributions (Friedman, 2007). The last, strategic CSR refers to the win-win situation
in which a company can do good by generating revenue and is thus the most noteworthy
and economically rational (Lantos, 2001). Naturally, from profit alone it is not possible to
deduce whether firms engaged for altruistic or profit reasons. They could for example have
“accidentally” made profit out of a purely altruistic CSR strategy. This makes it near to
impossible to distinct between the two empirically based on profitability of such measures.
A final noteworthy distinction is the one between symbolic and substantive CSR.
Compa-nies that are engaging in “greenwashing” i.e. using CSR to deceive customers (Delmas and
Cuerel Burbano, 2011) are generally engaging in “symbolic” CSR, which does not translate
into any benefits for stakeholders or the environment. In very competitive environments it
is even found to decrease financial performance (Kim et al., 2018). Substantive CSR refers
to CSR conduct that is aimed at being effective and is found, if stakeholders are aware of it,
to reduce costs and increase productivity (Schuler and Cording, 2006). Thus, it seems that
CSR, if conducted on a substantive basis, is positively correlated to profitability (see Section
7Du et al. (2011) shows that consumers that are involved in CSR initiatives favour this company even if
they have built a trust relationship with an industry leader, being the first approach to provide evidence for the use of strategic CSR as a means to differentiate.
2.4) regardless whether it was conducted for strategic or altruistic reasons.
2.2 Definition of CSR
One of the most commonly used definitions is the one employed by McWilliams and Siegel
(2001) and McWilliams et al. (2006)8: “[CSR relates to] actions that appear to further some
social good, beyond the interests of the firm and that which is required by law”. Their paper
resulted in convergence of using the same definition and undoubtedly contributed to the
re-search since then by resolving the issue of authors relying on varying definitional constructs.
However, this definition limits the scope of CSR. If a firm aligns their business model with
social issues, then arguably, by this definition the firm is not engaging in CSR. Suppose an
entrepreneur decides to build a firm based on producing sustainable cloth, using little to no
child-labour9, small emissions in production and distribution, employing a disabled work
force and so on. If he does this, not because he is genuinely altruistic, but because he sees
a profitable business in doing so, then his actions are part of his economic responsibilities
towards his firm and stakeholders. His corporation would therefore, by definition, not be
considered as socially responsible. For the purpose of this thesis, which is to scrutinise what
we know about the link of competition and CSR, we would like to not limit the scope of
CSR to voluntary10 as CSR in response to competitive pressure is, arguably, not voluntary
8They attribute methodological issues as the driver for the varying effects found in whether CSR is profitable
or not. Namely: “[1] result of inconsistency in defining CSR, [2] inconsistency in defining firm performance, [3] inconsistency in samples, [4] imprecision and inconsistency in research design, [5] misspecification of models, [6] changes over time, [7] or some more fundamental variance in the samples that are being analysed.” (McWilliams et al., 2006, p. 12) For managers, they propose using hedonic ways (i.e. product characteristics) to estimate whether CSR can be beneficial. This is not the point of this research, but we wanted to stress that care should be taken when estimating such equations as proposed by McWilliams et al. (2006). It does not contain any demand factors and therefore does also not touch upon the difficulties of estimating a structural equation with both price and quantity involved.
9This point can be debated as many countries that are using child labour are engaging in it due to Ricardos
Trade theory. They have an abundance in (child) labour and are using this as their comparative advantage. Many families in third world countries rely on such child labour to feed their families. Thus, child labour could be argued to not be harmful, but this is not a debate for this research.
10What and what is not voluntary is again a concept that could be debated lengthy without an unanimous
and certainly not beyond the economic responsibilities if it leads to a better positioning of
the firm. Most aligned with the idea of this research are the following two definitions:
(1) “The social responsibility of business encompasses the economic, legal, ethical, and
dis-cretionary expectations that society has of organisations at a given point in time” (Carroll,
1979 p.500 as cited in Carroll (1999)).
This definition captures the dynamics of CSR and points towards why so much literature
has focused on defining the concept. Thus, Carroll's definition offers the possibility that the
concept of CSR changes through time. The definition that I would like to employ is a mix
between Caroll’s 1979 definition and the one used by Peter Drucker 1984:
(2) “The proper social responsibility of business is to tame the dragon, that is to turn a
social problem into economic opportunity and economic benefit, into productive capacity,
into human competence, into well-paid jobs, and into wealth” (Drucker, 1984, p. 62) as cited
in Carroll (1999).
This definition does not rely on the voluntary part as most definitions post 1984 do. By
incorporating a voluntary part into the definition of CSR, one cannot make the distinction of
strategic and altruistic CSR. This is because strategic CSR will be in response to something
such as competition, which, arguably, is not voluntary nor beyond economic requirements.
All definitions, also the modern ones, accept that the economic function of a firm is part of
its responsibility as society wishes firms to create jobs and wealth. Why would thus
philan-thropy and voluntarism be part of the definition of CSR, that would be the job of charities or
governments through a sound taxation system. If companies find a way to align their interests
something voluntary. However, as economic activity is voluntary to begin with, we do not see the reason to include a voluntary dimension, i.e. beyond economic and legal requirements, into the definition of CSR.
with CSR, naturally this is still voluntary, which in turn makes CSR the profit maximising
strategy, then this type of CSR would better be placed under economic responsibilities. As
society wants companies to generate wealth and jobs, which sometimes can certainly be
done precisely by donating or doing voluntary work, then the definition should not limit the
concept to a voluntary dimension. If companies find a way to conduct their business in a
sustainable and socially acceptable way, within the legal and economic realms, then why
would we classify them as not being involved in CSR conduct?
2.3 Objective Function Firms Maximise
Proposing a maximisation problem that can be generalised to all firms and managers will
not be possible unless we find a way to read owners' and managers' minds. Nevertheless, it
is a useful point of discussion. To figure out what firms actually maximise it is necessary
to look at two sides i.e. do principals seek profit maximisation and do agents perform such
maximisation if wanted? Suppose agents engage in CSR practices that hurts profits. Then
we have a typical agency theory problem and managers and firms maximise different
func-tions. If there is a consensus that CSR can increase financial performance then, even taking
Friedman's (1970) extreme perspective, CSR should be conducted by agents as it maximises
the principals profit. Naturally, if the principal is maximising utility or social aspects then,
even absent profitability, the agent should engage in such practices. Russo and Fouts (1997)
acknowledge that consumers may be willing to pay more in the recent future, or go for the
company that is going green. This, as aforementioned, has happened. Thus, stakeholders
interest can become “parallel one another” (Freeman et al., 2004; Russo and Fouts, 1997)
when it comes to investing in CSR conduct. According to the research conduct by the author
there is no research on what functions firms maximise.11
11Future research in that area could attempt at collecting value statements of corporations, codify them
and compare them to CSR effectiveness and profitability to see whether they are enforced. Combined with (anonymous) interviews of managers, one could give an indication of what precisely is maximised by managers. This would then also shed light on whether an agency problem prevails as companies engage in CSR conduct.
Is it possible that firms have different motives than profits? From a stakeholder theory
point of view it is precisely these “different” motives that result in profitability. For now it
seems that by maximising stakeholder value one can achieve profit maximisation. If correct,
it would make it irrelevant whether firms maximise profits or total welfare as one cannot
be done without the other. There seems to be an agreement that CSR is correlated with
profitability and some evidence of that is provided in section 2.4. Thus the question arises of
whether firms engage in (profitable) CSR in order to maximise their profits or whether they
genuinely want to do good — probably a bit of both.
What precisely is maximised, however, is hard to assess even if there is an indication that
CSR may result in profitability. It is does important to ask, regardless what firms maximise,
whether they get better at it. Provided the evidence on CSR engagement it seems that CSR
that is conducted slowly and gradually is more profitable, suggesting that there is a learning
curve of engaging in CSR conduct to achieve profitability. As there is a higher effectiveness
from “substantive” CSR on its effect on productivity and reduction in costs (Graafland and
Smid, 2016; Perrini et al., 2011; Schuler and Cording, 2006). This indicates that CSR, which
is of “symbolic” nature, neither maximises profits nor total welfare. Thus, CSR that is of
“symbolic” nature cannot be in the objective function of a rational firm regardless the motives
i.e. strategic or altruistic. This leaves the question of whether “substantive” CSR (henceforth
CSR unless otherwise indicated) is within the objective function of a firm. CSR that is
conducted in response to competitive pressure, provided that it will increase the consumer
mass as well as boost productivity and reduce costs, should thus be part of firm's maximising
strategy. Whether it is, however, is beyond the scope of this research to answer.
2.4 Is CSR Profitable?
There is the notion that by conducting CSR a firm can develop inimitable capabilities
2003). These include “skills and competences, knowledge and innovation, values,
legiti-macy, trust, and reputation in the stakeholder network” (Perrini et al., 2011, p. XX). These
in turn, can be leveraged as a competitive advantage through its effects on efficiency,
in-creased consumer attraction, and differentiation (Perrini et al., 2011; Greening and Turban,
2000; Turban and Greening, 1997). If CSR is profitable it is necessary to understand the
channels through which it impacts profitability in order to gain understanding of the role
competition plays in this relationship. This section will shed light on recent developments in
the CSR and profitability linkage and through which channels profitability is affected.
Due the effect of CSR as an inimitable capability it has a complex relationship to financial
performance, which is likely not bidirectional. Studies that aim at measuring the relationship
in an aggregate manner ignore such complexity (Perrini et al., 2011). Orlitzky et al. (2008,
2003) points to the difficulty, while analysing 30 years of CSR and profitability literature
in a meta study, of conducting meta studies in this context and shows that measurement
errors, sampling errors and stakeholder mismatching, in primary studies, can explain 15 to
100% of variation in findings (Orlitzky et al., 2003). Even though findings vary substantially
in their magnitude (Orlitzky et al., 2008) both studies indicate a clear positive correlation.
They do indicate that the CSR and CSP have a bidirectional effect i.e. the relationship is a
self enforcing mechanism. Furthermore, when CSR is conducted “slowly and consistently”,
with a focus on CSR dimensions that are related to the business and with higher initial
investments into core dimensions i.e. employees, consumers and shareholders, that then
corporate financial performance (CFP) will be enhanced (Tang et al., 2012).
Zhao and Murrell (2016) indicate that prior CFP is correlated with subsequent CSP,
whereas vice versa there is no relation. Thus an indication that without initial financial
health there is no CSR conduct supporting the argument of reduced competition as a means
to increase CSR, or at least kick start responsible behaviour. They further indicate that the
study. Garcia-Castro and Francoeur (2016) analyses, through costliness, complementarities
and contingencies, the relationship of CSR and CFP. Their findings suggest that “any attempt
to find a simple monotonic relationship between investments in stakeholders and firm
perfor-mance seems to be futile” further indicating the difficulty of finding a causal interpretation.
Nevertheless, they suggest that stakeholder investments are more fruitful when done across
all stakeholder groups.
These studies indicate the difficulty of finding a causal effect on the relationship of CSR
and CFP, but agree on the positive correlation between the two. Wood and Jones (1995) stress
that most studies vary in the way stakeholders are mismatched and that CSR, depending on
which stakeholders employed, will affect profitability differently. Naturally, competition will
thus affect each dimension differently. Moreover, they stress that “the empirical rigor and
mathematics of CSP theory are not just underdeveloped; they are missing” (Wood and Jones,
1995, p. 261).
Vishwanathan (2017) points out that there is still a high level of ambiguity about they
key relationship of CSP and CFP, not about the positive relationship, but rather the causal
interpretation. Indicating that both McWilliams et al. (2006) and Wood and Jones (1995)
critique have not been fully addressed. For example institutional theory would suggest that
firms engage in CSR to obtain legitimacy and that this legitimacy translates into reduced
risk. Others would argue, based on a more microeconomic approach, that investing into
CSR builds positive relationships which then mitigates losses when negative events occur
and hence reduce risk. Both conclude that it reduces risk, but have a very different theoretical
explanation for the effect (Vishwanathan, 2017).
Her meta analysis indicates that there are four pathways through which CSP impacts
CFP: (1) stakeholder endorsement, (2) reputitional improvement, (3) innovation, and (4) risk
mitigation. It is noteworthy to stress the fragility of reputational improvement. B´enabou
their discount factor. That is to say, if individuals engage in activities that receive to much
attention, pro-social behaviour will reduce due to the image concern of engaging in such
behaviour for the wrong reasons. This is likely to translate to corporations as well: if
cor-porations would engage in CSR and ensure that this CSR receives publicity, it is likely that
the public will perceive such behaviour as “greenwashing”. Thus there is a fine line
be-tween creating awareness that increases effectiveness of CSR practices and awareness that
will be deemed as “desperate”. Correspondingly, such companies would be perceived as
greenwashers by the public.
Whereas investing into core stakeholders such as employees and consumers can reduce
costs and increase productivity, there seem to be little short term benefits resulting from
firms engaging in environmental friendly conduct. In the context of competition, this would
suggest that CSR, on an environmental dimension, is not an effective strategy to deal with
immediate threats. Nevertheless, investing into environmental performance has its long term
benefits. In the presence of unexpected shocks, stock price movements are significantly less
negative in firms that have strong environmental control ratings (Shane and Spicer, 1983).
Furthermore, companies with stronger environmental performance i.e. less pollution have
been associated with reduced risk on stock prices (Spicer, 1978). One explanation for this is
that firms that are environmentally friendly have a long term objective reducing the risk of
short term default. This makes the stock more attractive, especially to institutional investors,
which in turn drives up its value (Graves and Waddock, 1994).
These points suggest that CSR measures, if correctly and specifically used, should be
employed by rational firms and can be used as a strategy in response to competition.
More-over, the relationship of CSR and financial performance is found to be bidirectional. This
may indicate that certain industries are affected differently by competition as in some cases it
will reduce CSR through its reducing effect on profits and in others it will be high in order to
on more factors than just industry and country fixed effects, which is what is mainly used to
control for endogeneity issues in empirical work of competition and CSR as elaborated in
section 4.
3
Measures of Competition and CSR
3.1 Measures Used
CSR. CSR conduct is generally measured by and provided by environmental, social and
gov-ernance (ESG) research. The most widely used providers are the MSCI ESG KLD STATS
(henceforth, KLD) measure and Asset4.12 These are used as a proxy of CSR conduct both
for responsibility concerns and strengths.
The KLD data base comprises of five universes labelled A through E:
(A) MSCI KLD 400 Social Index (KLD400) plus the MSCI USA Index, covering a
time-span from 1991 to 2014 and encompasses 650 firms. The index experienced slight
ad-justments throughout these periods, so the index's coverage slightly adjusted and
com-prises the said data only since STATS-2013 data set.
(B) the largest 1000 U.S. companies (by market capitalisation), which however, has been
discontinued as of the STATS-2013 data base.
(C) is comprised of the KLD400 and the largest 1000 U.S. companies over a time-span from
2001 to 2013 covering 1100 firms.
(D) contains the MSCI USA IMI Index covering 2400 firms from 2003 to 2014.
12There is also Sustainalytics, which yields a neglectable difference in results (Van den Heuvel, 2012); as no
empirical work testing the relationship of competition and CSR used Sustainalytics and that it yields similar results to Asset4 we will only look at KLD and Asset4
(E) is the “non-US universe” and only covers 2013 and 2014. It includes emerging markets,
the investable market index, Nordic IMI, Australia IMI, South Africa and Canada, and
encompasses 2600 firms.
The index's measures span across three pillars: environment, social and governance
where each pillar is further subdivided into themes such as climate change and human
cap-ital. For each theme the database contains key issues such as carbon emissions for climate
change. For each key issue the index assesses components of the management capabilities
of a company i.e. strategy & governance, initiatives and performance. Generally, a
com-pany gets scored on only four to seven of the most “material” ESG key issues for the firm's
primary industry. In addition, there are a set of key issues that are applicable to all
compa-nies such as carbon emissions, labor management, and health & safety. Each key issue is
rated on a scale from zero to ten. Also, the ratings are then converted into a simple binary
scoring model. If the given threshold for an indicator is met the variable takes the value of
one and zero otherwise.13 The scores are constructed using (1) macro data (at segment or
geographic level) from academic, government, and NGO datasets, (2) company disclosure
data i.e. 10-K, sustainability reports etc.) and (3) government databases.
For a detailed overview of how these measures are constructed see the KLD data base
executive summary methodology clickhereand for a detailed overview of how measures are
constructed seehere.14 To understand that these measures are constructed as a combination
of self-reported data and assessed data it is however useful to provide a brief explanation of
how some variables are constructed.
One of the environmental dimensions is waste management. Companies that report
13If a company has not been researched on an indicator it will naturally be indicated with NR (not
re-searched).
14Alternatively see MSCI ESG KLD STATS: 1991-2014 DATA SETS accessible through:
https://www.wiso.uni-hamburg.de/bibliothek/recherche/datenbanken/unternehmensdaten/msci-methodology-2014.pdf and see MSCI ESG Ratings Methodology 2015 accessible through:
strong programs (such as addressing electronic waste, having recycling programs etc.) and
track records of reducing waste/emissions score higher compared to companies that do not.
Additionally, there is the environmental opportunities section, which is likely to be one of the
more subjective dimensions. Corporations that refurbish buildings with “green”
characteris-tics15score higher. More, if they have a “green” strategy in place they yield a higher score
and consequently are better ranked. For an employee dimension there is e.g. gay and lesbian
policies. Corporations that have notably progressive policies in place, towards its gay and
lesbian employees, score higher. Another example is the community engagement indicator,
which identifies corporations that have notable community engagement programs in place
(concerning the local communities in which a firm is active in) and metrics include
commu-nity impact assessments. This measure also exists as a concern where the indicator assesses
the gravity with which a firm's controversies impact the community. This includes, but not
limited to, widespread or egregious community impacts, criticism by NGO's and other third
party observers as well as a history of development related legal cases.
These examples illustrate that even though rankings do not solely rely on self-reported
policies or programs, most measures do include such programs. Further, strengths and
con-cerns, as shown by the community impact, are not derived from the same methodology where
concerns generally comprise scandals or concerns of third party observers and strengths
en-compass self reported policy programs but also an assessment of its effectiveness. This, as
elaborated in the proceeding concern sections, may result in a bias of ranking as the scores
are at least partially based on self-reported data and/or constructing a net index i.e. strengths
minus concerns. It is also important to note that the index puts forward, in its disclaimer,
that “the information should not be relied on and is not a substitute for the skill, judgment
and experience of the user, its management, employees, advisors and/or clients when making
investment and other business decisions. All information is impersonal and not tailored to
15With “green” characteristics it includes, but not limited to, lower embodied energy, used recycled materials,
the needs of any person, entity or group of persons.” As the literature shows that CSR
im-pacts profitability differently, depending on the dimension used, it is also expected that the
impact of competition on CSR (if any) varies across dimensions and is likely to be
depen-dent upon industry and country contexts. Therefore, it seems that using an all generalised
index to measure may drive a systematic bias into the results when running regressions on
the relationship of competition and CSR as elaborated later on.
The most noteworthy subsection is the KLD400 index, which contains data from 1991
to 2015 for US and 2013-2015 for non-US i.e. Canada AsiaPacific and European markets
(Source: KLD400 methodology). The research is conducted by the an independent unit of
MSCI and is the most widely used measure up to date. Moreover, it contains reference data
i.e. for example name of company. It includes observables such as emissions and scandals,
but also self-reported policy dummies such as employee benefit program and/or
environ-mental emission program. These scores are measured on a scale from AAA to C and are
normalised across industries and companies i.e. some industries may not contain a single
AAA rating. Apart from Standard & Poor (S&P) 500 the data base also contains the KLD
400 Social Index in which companies that are involved in alcohol, gambling, tobacco,
mili-tary weapons, civilian firearms, nuclear power, adult entertainment and genetically modified
organisms are excluded (Source: MSCI KLD 400 Social Index Methodology). Moreover,
companies need to receive a rating of above B in the MSCI ESG Rating to remain in the
KLD index. For a detailed explanation see KLD Methodology 2016 or clickhere.16
The Asset4 database, provided by Thomson Reuters (Reuters), is the predecessor of the
Reuters ESG Scores.17 The data base encompasses 7000 companies globally.18 Moreover,
16Or alternatively, visit the MSCI KLD 400 Social Index Methodology 2016 accessible here:
https://www.msci.com/eqb/methodologymeth docsMSCI KLD 400 Social Index Methodology May 2016.pdf
17Only one paper used the Asset4 and did not use the improved ESG Scores provided by Reuters, thus the
focus in the analysis lies on the implications of the Asset4 drawbacks albeit the ESG Score data base does correct for some issues. Most data, however, remains self reported by companies, thus, the improvement does not deal with the most apparent issue. As the KLD base is by far the most widely used, it would not add too much value of explaining the newer version of the Asset4.
the measures are constructed using only self reported company data. Only one study in the
linkage of competition and CSR used the Asset4, instead of KLD. As the data is based on self
reported company information only, it is not necessary to provide the same detailed overview
than for the KLD.
The index provides over 250 key indicators divided into ten main themes such as
emis-sions, environmental product innovation, human rights, shareholders and so on. These key
indicators are sub branches of the following five main pillars: (1) corporate governance, (2)
economic, (3) environmental and (5) social. The ratings work from A+ until D- and are made
up of a score from one to ten. It is important to note how Asset4 constructs the weights.
Cat-egories that contain several key issues such as management (it contains: composition,
diver-sity, independence, committees, compensation etc.) bear more weight than key issues such
as human rights that are made up of less or only one indicator. A very transparent illustration
of how it is calculated can be found in the Asset4 Reuters Methodology, or by clickinghere.19
Competition. How to measure industrial concentration has been debated lengthily over
the past decades, especially concerning what types of distribution should be used (for
ex-ample see Hart (1975) and Hart (1971)). A measure of industrial concentration should be
conducted using two observable, and thus measurable, phenomena i.e. the size and number
of firm distributions (Marfels, 1971). One measure that emerged is the entropy measure,
which allows e.g. concentration measured on a national level to be split into concentration
of several regions. This measure has rarely been employed in research and will thus receive
no further attention.
Furthermore, one can use characteristics of the size distribution of firms to draw
impli-cations on the level of concentration in a given industry (Curry and George, 1983). One
19Alternatively it can be found through Reuters ESG Asset4 methodology base accessible
here:
such measure is the use of the Lorenz Curve, intended to measure the distribution of wealth,
to measure the departure from a competitive industry in which each firm has the same size
(Gini, 1912; Lorenz, 1905). Indices that are based on the Lorenz Curve capture the
differ-ences in market shares fairly unambiguous but nevertheless ignore the number of firms (i.e.
the Lorenz Curve will be the same for 100 firms than for 5) (Davies, 1979).
The two most commonly used, by empirical literature on the relationship of competition
and CSR, and competition authorities, are the HHI, which is the sum of squared market
shares (Hirschman, 1964) and the ith-concentration ratio (CR
i) i.e. the sum of the ithlargest
firms (Marfels, 1971). See Belleflamme and Peitz (2015) and Tirole (1988) for a more
detailed description of the two. Using a linear Cournot model it is possible to relate the
Lerner index directly to the HHI (Belleflamme and Peitz, 2015), however, it is important
to note that most industries are not competing on quantities. The most useful property of
the HHI, which is not addressed using concentration ratios, is that larger firms receive more
weight. Different market structures, as elaborated in the concern section, can lead to the
same level of concentration when using concentration ratios. This problem is addressed in
the HHI by squaring market shares.
To overcome the concern of cross ownership20researchers have modified the HHI to
ac-count for such ownership structures. In merger cases it has been suggested that the modified
HHI (MHHI) captures cross ownership changes competition depending on how much of a
target is acquired (O’brien and Salop, 1999). Azar et al. (2016) have illustrated that this cross
ownership can be captured by a generalised HHI (GHHI). The GHHI does not change the
fact that the measure relies on the assumption of Cournot competition and that a manager
follows shareholders' interest. The only difference to the HHI is that the GHHI “allows for
20Suppose multiple companies have a vested interest in one another. E.g. Goldman Sachs will have a
(controlling) stake in multiple firms active in a given industry as well across industries. Much of the CSR conduct may then be attributed to spillover effects from one major firm that has a management in place that puts a heavy focus on CSR measures. Some firms' CSR conduct would then be driven by cross ownership, not competition per se. This will be elaborated further in the concern section.
simultaneous common ownership and cross ownership” (Azar et al., 2016, p. 12).
Another measure, which is found in any microeconomic or IO text book, is the Lerner
index. The general idea behind the index is that in perfect competition firms price at marginal
costs. This results in the index taking a value of zero. The index measures the markup
(i.e. price cost difference as a percentage of price) that can be charged by a given firm
and is therefore not dependent on size distributions and should be applicable to almost any
situation.21 Moreover, the index has a useful property: it can be expressed as the inverse
elasticity of demand. This is also known as the inverse elasticity rule according to which “the
markup is higher the less elastic is demand” (Belleflamme and Peitz, 2015, p. 27). If data is
available, which may proof to be a major issue in certain industries, then there are, broadly
speaking, two main ways to estimate price cost margins in the literature: (1) to estimate
each firm's price cost margin as something similar to revenues minus variable costs divided
by revenues (for example Scherer and Ross (1990)) and (2) the structural approaches such
as estimating elasticities (for examples see Graddy (1995) and Berry et al. (1995) (Boone,
2008).
Boone (2008) proposes a new measure of competition. The measure is constructed by
taking the difference of profits of the most efficient and least efficient firm relative to the
difference of the second most efficient firm's and the least efficient firm's profit; the ratio
of the two then represents the relative profit difference (RPD or Boone measure). In this
model an increase in competition can be modelled by either a fall in entry barriers or through
tougher competition of existing players. There are theoretical papers (Amir 2002, Bulow and
Klemperer 199 etc. as cited in Boone) that show how competition increases the price cost
margins instead of reducing them, which implies that, in traditional price cost estimations,
competition can in fact increase margins. The Boone measure's underpinning logic is that
21Certainly, there are several drawbacks to this as well. Most industries are not perfectly competitive to begin
with and in other industries mark ups are not indication of monopoly power as elaborated in the proceeding section.
when competition increases there is a larger shift of profits from less efficient firms to more
efficient firms: there is a punishment for being inefficient. Consequently, the measure is
monotone in competition under derived conditions (Boone, 2008). The measure has not
been used in any of the papers, but it may prove to be a useful tool for future research. It
is noteworthy to mention that the measure has been proposed as an additional tool, as it
requires the same data, not per se as a superior measure on all aspects. Albeit it is shown to
be theoretically more robust (see Boone (2008) for the derivations).
In the literature that aimed to analyse the link of competition the HHI index was the most
widely used. Some authors, as elaborated in section 4, used instruments such as a reduction
in trade barriers. These are good ways of measuring the impact additional competition has,
but cannot be summarised as a general measure. Also, variations using the fraction of
em-ployees (relative to all emem-ployees in an industry) and/or asset size (relative to all assets in an
industry) instead of market share and/or using the number of firms as an indication have been
employed. These are all variations of the HHI but are noteworthy as each of these proxies
has its own drawbacks as elaborated in the proceeding sections.
3.2 Measurement & Data Concerns
CSR. One has to make a distinction between social performance and social action: social
im-pacts are not necessarily the same as social activity (Mitnick, 2000). This questions whether
self reported data allows researchers, investors, policy makers and managers to assess the
CSR behaviour of companies or at least its effectiveness. Companies have a moral hazard
when reporting such programs and are likely to overstate their CSR efforts, questioning the
legitimacy of variables based on self reported data in the use of CSR research. Studies that
are based on measures such as the Asset4 are thus inherently measuring what companies
re-port and thus they are measuring the relationship of competition and rere-ported CSR, not CSR
level data, and therefore, controls for some of these concerns.
Moreover, there are concerns regarding the endogeneity of CSR itself. Due to external
pressures resulting from, not limited to, competition, profitability or stakeholders, managers
may increase their CSR (reporting) to satisfy the party exhibiting the pressure. If stakeholder
pressure drives firms' decision to engage in CSR, for example, then this has to be controlled
for when assessing the relationship of competition (or any other measure) and CSR, which
may prove to be harder than said. Also, by only taking into account self reported data, there
is a sample selection bias. It is suggested that managers who are seeking acknowledgment,
from external observers such as investors and society, are more inclined to enhance their CSR
(reporting) efforts out of fear of not being reported (Shahzad and Sharfman, 2017; Doh et al.,
2010). Shahzad and Sharfman (2017) show that such a bias exists and can be controlled for
by implementing proxies to capture stakeholder pressure.
Furthermore, if measures do not clearly indicate the outcome of certain policies in place,
the distinction between substantive and symbolic CSR becomes blurry. By engaging in
greenwashing, a company will have to invest significantly less resources while having a
probability of not getting caught. A manager or company is thus, arguably, incentivised to
do so under certain conditions. Also, this likely means that managers have a rough estimation
of the consequences of CSR conduct and whether it is profitable to engage in greenwashing.
Naturally, this might backfire to managers as it has been shown that greenwashing can reduce
profitability, that is, if caught. The complexity of distinguishing between greenwashing and
CSR calls for a better reporting of the effects of CSR programs in order to understand the
forces that lead to companies acting in socially responsible ways not just for the sake of
it, but actually to elevate total welfare, regardless whether the motives were profitability,
gaining a competitive advantage or genuinely doing good.
The KLD measure is regarded as the superior measure of CSR (for example see
does it need to be to allow for useful insights. The KLD measure is largely based on other
research than only self reported company data, which already is a significant improvement
with regards to other measures. Nevertheless, self reported data such as sustainability reports
still feed into the ratings. Chatterji et al. (2009) tested how well the KLD data base does in
predicting historical and future environmental performance and find that KLD environmental
ratings do a “reasonable job of aggregating past environmental performance”(p. 50).
Also, they find that total environmental strengths score did not do well in predicting,
whereas net environmental concerns (i.e. strength minus weakness) can help at
predict-ing future performance. The methodology employed to measure strengths and weaknesses,
however, questions the construction of such a net index and whether it makes
methodologi-cal sense to subtract the two scores from one another. As strengths and concerns do not have
convergence validity, it is not possible to construct a grounded measure of net CSR by
sub-tracting concerns from strengths. (Flammer, 2015; Mattingly and Berman, 2006;
Johnson-Cramer, 2004)
There has thus to be a focus by modern researches into CSR into the core issue: how to
measure CSR? Without a clear consensus of of measurement that makes clear distinctions
between whether or not data was self reported, or whether a policy implemented was of
substantive or symbolic nature, it becomes near to impossible to comprehend under what
circumstances CSR conduct becomes a strategy and/or in what scenarios it is profitable. As
there is a different effect on profitability when engaging in substantive or symbolic CSR, it
is expected that this difference drive much of the results. This will also make it harder to
assess a causal relationship between competition and CSR.
In addition there is a problem with the aggregation of CSR scores i.e. each dimension has
a different linkage to profitability and is thus also expected to be differently affected by
com-petitive pressure. This leads to yet another concern with the measurement of CSR for
the same methodology in aggregation or tested the relationship of competition on the same
individual measures. Moreover, without a clear distinction of which measures are increasing
profitability or which measures are increased by profitability, aggregation of such data makes
a causal interperation even harder, especially when looking at the competition and CSR
re-lationship. Suppose for example environmental initiatives are less important to deal with
immediate competitive threats as they have little short term effects on core stakeholders and
productivity. Now imagine that at the same time focusing CSR on core stakeholders
trans-lates into increased productivity, motivation and quality of work that in turn reduces cost
and/or boosts profits. The two dimensions of CSR, if even affected by competition, work in
different directions. If one would now aggregate environmental scores and employee benefit
programs the effects of competition will likely cancel each other out. Thus aggregating may
cause large problems (that are not observed) by combining dimensions that have opposing
views one will underestimate the positive effect CSR can have, or overestimate the effect
less efficient CSR use can exhibit.
Proposing different types of CSR, such as altruistic, strategic, symbolic and substantive,
was a good advancement in understanding the concept. Which one of these actually prevail
under what contexts however, has not been unanimously agreed upon. As many dimensions
are differently affected, which is reasonable to believe at this point, then it is questionable
to aggregate a linkage of general CSR and competition. Needless to say the two data-bases,
i.e. KLD and Asset4, do allow for a coherent measurement across studies. However, most
authors are using their own ways of aggregating measures and selecting key indicators as
elaborated in section 4. Moreover, these data bases do not indicate the effectiveness of such
measures making it impossible to distinct between symbolic and substantive CSR by only
using these data bases.
Further, the time horizon for which these measures are available, especially for European
analyses, which may be an explanation for why most studies in the competition and CSR
relationship have used US data.
Lastly, there is a heavy selection bias of studies using the KLD 400 Index. Measuring
only corporations that receive a rating of “AA” or above, limited to the 400 highest rated
companies, does not give insights into corporations that are not socially responsible. In turn,
this magnifies the sample selection bias that already exists by only using corporations that
have CSR reporting in place. This limits the spectrum of CSR conduct that is measured and
will not give any insights into how competition is related to corporations that are not engage
in CSR. Understanding what drives CSR conduct in companies that are fairly engaged in
it, is a key piece in understanding the role competition plays especially when deciding on a
public interest defence in cartel offences. Moreover, ignoring some “dirty” industries such
as alcohol, tobacco, military weapons and genetically modified organisms does not allow to
gain insights into how CSR conduct relates to competition in industries with large
externali-ties on society now and future generations.
Competition. Whereas the Lerner Index ignores the number of firms (Davies, 1979), the
CRi22and the HHI use both differences in market shares and number of firms. Nevertheless,
these two measures place a too high value on the number of firms, leading to a certain
de-cline in competition as long as the market share of one additional firm exceeds 0.05 (Hart,
1975). Thus, they overemphasise the impact additional entry of a small firm may have on
concentration (Davies, 1979).
All studies that tested the relationship of competition and CSR do use the HHI or some
22The CR
ihas not been used by researchers in the empirical literature testing the relationship of competition
and CSR. The concerns arising with this measure are enormous and if authors used only a certain amount of firms to construct a measure of concentration, they still used the HHI approach, but only on a certain number of firms, not all. To illustrate why: suppose we are measuring the CR4. An industry that has one firm with 65%
with the remaining share divided amongst seven firms equally leaving 5% each. Now, suppose that there are
four firms with 20% market share and four firms with 5%. Both of these industries would yield the same CR4
form of it, as a proxy for concentration. Few works such as Flammer have used an instrument
such as a reduction in trade barriers.23
The first issue with the HHI is that slight measurement errors can change concentration
proxies drastically. Suppose that the largest players in the market have 35, 25 and 15%
market share and the remaining market shares are divided into five companies with a share of
5% each. This yields an HHI of 2175. If the market share is measured as 40% and 20% i.e. a
5% difference in the largest two companies the HHI jumps to 2325, and with 10% to 2575. If
measurement errors in the larger shares of companies vary across years and industries, then
concentration ratios may mask the effect competition exhibits on CSR as some industries
may in fact be more competitive than others albeit otherwise indicated. Also, by squaring
market shares the HHI may understate the value of small firms and overstate the value of
large firms (Roberts, 2014). This concern carries through to proxies such as using ratios of
employee numbers or ratio of asset size of corporations.
The next issue is cross-ownership of firms active in multiple industries, which in turn may
mask the level of competition. Suppose several firms have a stake in one another. According
to the basic HHI this would not be accounted for as it is simply the sum of squared market
shares. Suppose that four firms are active each having a share of 25% (HHI of 2500). If
now, however, each of the four owns a 50% in one of the others, for the sake of the argument
suppose that this cross ownership is between two firms each thus creating two groups of two
cross-owned firms, then concentration would be a lot higher than suggested by the HHI. Due
to the cross-ownership they do not just have the means to coordinate easier, they also have
a vested interest in keeping competition at a minimum. The standard HHI measures would
fail to capture such effects, whereas the MHHI does indeed account for such ownership
structures. For this reason has the GHHI and MHHI been proposed.
23Naturally, such instruments are also questionable in their exogeneity. But Flammer, for example, also
only looked at multilateral agreements, as they are harder to be influenced by a large corporation, to reduce endogeneity concerns of her measure of concentration. This will be elaborated later on.
More importantly, measuring concentration, which the HHI does, does not necessarily
imply competition even if it has been accurately measured. Market concentration does not
have to lead to competition in any situation. If for example capacity constraints are present,
and large initial investments into plants are required, then this creates entry barriers for new
entrants. Suppose that the capacity constraint firms can still operate profitable and have no
incentive to leave the market. In this scenario, a large number of firm does not necessarily
imply competition. The number of capacity constraint firms would provide the good at their
maximising production level, while other firms can start increasing prices as the other firms
will not have the ability to capture this. Moreover, in this situation barriers would be high
and potential entrants would have to conduct significant investments.
This brings the next critique on the HHI: entry barriers. An industry that is very
concen-trated by a two to three large firms, but faces zero entry barriers does not allow its players
to exhibit market power. Thus, the HHI would suggest that there is low competition, when
in fact their is extremely high potential competition, which in turn exhibits pressure on to
firms to not raise prices for example as this would attract immediate entry of potential
com-petitors. Further, the degree of concentration is impacted by many other factors apart market
shares and entry barriers such as locational distribution, psychology of corporate officers
and/or product substitutability (Roberts, 2014; Herfindahl, 1950). Such indices, relying on
size distributions, also fail at capturing firm specific demand elasticities which in turn
deter-mine the level of output and prices (Elzinga and Mills, 2011). For illustration purposes one
more example is provided: suppose there is an industry with two dominant players of each
40% market share. If one of the remaining firms, let us say for simplicity two more firms
with each 10% share, is developing a new platform, algorithm and/or patent, then it is likely
that this industry will be highly competitive in the near future. However, it will take time for
this firm to reach a large enough market share. Due to capacity constraints it may choose
seem to be dominated by two firms, based on market shares, whereas in reality the smallest
firm has significant power in the market, which in turn makes it a competitive industry. The
two large firms have significant buying power and the small firm has capacity constraints
resulting in all three firms preventing each other from abusing its dominance (that is, absent
collusion of course). In this scenario, competition is working perfectly fine and consumers
can benefit from the small firm licensing out its technology to others. Nevertheless, the HHI
would be 3400 indicating a high level of concentration.
Also assume that proxies are used measuring the number of employees and asset sizes
instead of market shares. By measuring the number of employees one will underestimate
the effect capital intense firms have on concentration. Suppose Amazon continues its
atom-isation of factories and carries such practices to all levels. Once Amazon would be fully
automised, it would in fact not be taken into account when measuring the number of
em-ployees. Needless to say, if Amazon manages to fully atomise itself it should be seen as
more dominant opposed to being neglected in such measures. The same argument goes for
relying on asset size as it would understate the impact labor intense corporations have on
competition.
Apart the criticism on the HHI, there are several concerns when using price cost margins
as a proxy of the level of competition. The Lerner index itself has been said to measures
market imperfection not the degree of monopoly power as it solely measures the departure
from a perfectly competitive environment in which no profits can be earned (Scitovsky 1995
(CITE)). The index is a static measure and therefore fails at capturing the dynamics of
com-petition such as “technological advancement, innovation, and learning by doing” (Elzinga
and Mills, 2011, p. 559). If firms exhibit a significant strengths in learning by doing, the
in-dex fails at capturing potential competition and possibly overstates the level of competition
based on market shares, when it is in fact driven by innovation and companies getting better