Content uploaded by Andreas Rasche
Author content
All content in this area was uploaded by Andreas Rasche on May 15, 2020
Content may be subject to copyright.
1
Which Firms Leave Multi-Stakeholder
Initiatives? An Analysis of Delistings from the
United Nations Global Compact
Andreas Rasche (ar.msc@cbs.dk) (Corresponding Author)
Copenhagen Business School
CBS Sustainability
Dalgas Have 15, 2000 Frederiksberg
Denmark
Phone: +45 2844 3955
and
Stockholm School of Economics
Mistra Centre for Sustainable Markets (MISUM)
Box 6501, 113 83 Stockholm
Sweden
Wencke Gwozdz (wencke.gwozdz@fb09.uni-giessen.de)
Justus-Liebig-Universität Gießen
Zeughaus, Raum 327
Senckenbergstraße 3
35390 Gießen, Germany
Mathias Lund Larsen (lundlarsenmathias@gmail.com)
International Institute of Green Finance
Central University of Finance and Economics
39 South College Road, Haidian District
Beijing, P.R. China100081
Jeremy Moon (jmo.msc@cbs.dk)
Copenhagen Business School
CBS Sustainability
Dalgas Have 15, 2000 Frederiksberg
Denmark
Cite as: Rasche, A., Gwozdz, W., Lund Larsen, M., & Moon, J. (2020). Which firms
leave multi-stakeholder initiatives? An analysis of delistings from the UN Global Compact.
Regulation & Governance (forthcoming).
!
2
Which Firms Leave Multi-Stakeholder
Initiatives? An Analysis of Delistings from the
United Nations Global Compact
Abstract:
This study analyzes which firms leave multi-stakeholder initiatives for corporate social
responsibility. Based on an analysis of all active and delisted business participants from
the United Nations Global Compact between 2000 and 2015 (n=15,853), we find that
small and medium-sized enterprises are more likely to leave than larger and publicly-
traded firms; that early adopters are less likely to leave than late adopters; and that the
presence of a local network in a country reduces the likelihood of leaving. Based on these
findings, we discuss theoretical implications related to MSIs’ output legitimacy, the
nature of organizational platforms supporting norm entrepreneurs within MSIs, and the
occurrence of legitimacy spillover effects in local networks.
Keywords: multi-stakeholder initiatives, corporate social responsibility, UN Global
Compact, delisting, legitimacy
3
INTRODUCTION
Multi-stakeholder initiatives (MSIs) are increasingly seen as a means of addressing social
and environmental problems in areas such as human and labor rights, deforestation, and
climate change. MSIs reflect voluntary predefined rules that seek to guide, assess, verify
and communicate firms’ social and environmental performance (de Bakker et al. 2019;
Jastram & Klingenberg 2018; Fransen & Kolk 2007). Although these initiatives differ in
terms of their aims and underlying mechanisms, they all rest on a collaborative approach
towards (global) governance (Rasche 2012). Within MSIs the authority for designing and
enforcing relevant rules is shared between different interest groups, which, as a whole,
cross the state/non-state and profit/non-profit-boundaries (Fransen 2012). While
certification MSIs audit participating firms against a set of predefined rules (e.g., the
Forest Stewardship Council), principle-based initiatives offer a set of foundational values
that firms can sign up to without any monitoring or certification (e.g., the Equator
Principles).
When looking at the MSI literature, existing studies have helped us to understand
why firms join MSIs (Bartley 2007), how these initiatives become institutionalized
(Gulbrandsen 2014), what influences their perceived legitimacy (Mena & Palazzo 2012),
and whether or not MSIs have an impact on the issues they are claiming to address
(Berliner & Prakash 2015). Yet, surprisingly little analysis has been devoted to what sort
of firms leave MSIs in general and principle-based MSIs in particular (for an exception
see Knudsen 2011). To address this shortcoming in the literature, this article studies the
characteristics of those firms that leave principle-based MSIs, either because they failed
to meet the minimum requirements of participation or because they voluntarily withdrew.
4
Our study rests on an analysis of all active and delisted business participants from
one principle-based MSI, the United Nations Global Compact (UNGC), between 2000
and 2015 (n=15,853). We excluded some participants from this universe: (1) all firms
that joined the initiative between 1 January 2013 and 31 December 2015, as firms that
joined during these years cannot be delisted due to the UNGC’s policy (UN Global
Compact 2017) and (2) firms which cannot count as delisted because they do not exist
anymore, have undergone a merger or acquisition or were consolidated under the parent
company. This yielded a final sample size of 11,499. Firms leave the UNGC (and hence
are considered delisted) because they are either forced to do so after failure to submit a
mandatory annual Communication on Progress (COP) report or because they decide to
voluntarily withdraw from the initiative. We selected the UNGC for three reasons. First,
it is the world’s largest MSI in terms of the overall number of business participants that
joined the initiative (more than 15,800 as of December 2015). Second, it is also the
initiative with the highest number of delisted firms (more than 7,000 as of December
2015). Finally, the UNGC is a global initiative with participants from more than 163
countries and not restricted to any particular sector or firm size. While our results cannot
be generalized to all types of MSIs (e.g., certification initiatives that delist because of
failed audits), the analysis enhances our knowledge about what sort of firms become
uninterested in principle-based initiatives that have a similar institutional design like the
UNGC (e.g., the Principles for Responsible Investment or the Equator Principles).
We find three main effects: (1) corporate size and ownership status influence
whether participants leave (i.e. smaller and privately held firms are more likely to be
delisted); (2) the time of joining affects delistings (i.e. early adopters face a lower
likelihood of being delisted); and (3) the presence of local network structures in the
5
context of a MSI impacts delistings (i.e. the existence of such networks ties participants
closer to the initiative and hence limits the likelihood of firms leaving, while the number
of participants in a network does not seem to have an effect). Based on these findings, we
make three main theoretical contributions. These contributions address shortcomings in
the current literature on principle-based MSIs and thereby show why, and in which ways,
our analysis is important for future theory development.
First, analyzing delistings allows us to extend our knowledge on the output
legitimacy of relevant initiatives. Output legitimacy refers to whether MSIs are effective
in addressing the issues they claim to address, while input legitimacy mainly deals with
the question whether those which are being governed through an initiative are also
involved in decision-making (Scharpf 2009). So far, the literature emphasizes (a) that
principle-based MSIs suffer from lower levels of output legitimacy as they do not monitor
participants (Mena & Palazzo 2012; Sethi & Schepers 2014) and (b) that mandatory
reporting requirements can at least partly offset these low levels of output legitimacy
(Hamid & Johner 2010). Our results extend this discussion by highlighting that enhancing
output legitimacy via reporting requirements is unlikely to work in the case of those MSIs
that allow firms of different sizes to join. Delisting firms for non-reporting only enhances
output legitimacy if participants perceive it as a credible threat. However, since the
credibility of such a threat depends on the expected reputational losses in the case of
delisting (among other things) and since such reputational losses are asymmetrically
distributed among larger and smaller firms, future theory development needs to consider
the size of participating firms much more when thinking about how to strengthen output
legitimacy.
6
Second, an analysis of delistings can help to address the lack of longitudinal
analysis of principle-based MSIs. Although initiatives like the UN Global Compact
(Rasche et al. 2013) and the Principles for Responsible Investment (Gond & Piani 2013)
have been studied frequently, research usually does not differentiate the behavior of early
and late adopters. Our results respond to this shortcoming. Our results confirm that
Finnemore and Sikkink’s (1998) norm cascade model holds in the context of principle-
based MSIs – i.e. early adopters show higher levels of commitment than late adopters.
Our analysis further shows that these dynamics even hold in situations in which the
organizational platforms that enable the work of norm entrepreneurs are weakly defined.
Finally, an analysis of delistings is important because it allows better insights into
the effects of MSIs’ local networks – i.e., local clusters that are set up to engage
participants in specific national contexts (Whelan 2010). While research has looked at
factors explaining cross-country differences in local network participation (Abdelzaher
et al. 2019), what remains unclear to date is whether such networks influence participant
behavior (and if so in which ways). Based on our results, we suggest that the presence of
a local network in a country decreases the likelihood that participants leave, because at
least some firms can reap positive legitimacy spillover effects by remaining part of the
network. Leaving the UNGC (and hence the local network) would imply that firms lose
the perceived association with a similar class of local organizations that are organized
under the legitimate UN umbrella. We thus complement Haack et al.’s (2014) claim that
MSIs need to be understood as arenas in which legitimacy spillover effects exist.
Our analysis proceeds by briefly reviewing relevant parts of the literature on MSIs
in general and the UNGC in particular. We then develop hypotheses, which predict what
makes delisting more/less likely in the context of the UNGC. Next, we explain what data
7
we used, how we measured relevant variables, what statistical analyses we conducted,
and how we checked the robustness of our results. The subsequent section presents the
results of our analysis. We then outline theoretical implications of our results and show
how far our findings extend existing scholarly insights. We close by outlining policy
recommendations for MSIs in general and the UNGC in particular.
MULTI-STAKEHOLDER INITIATIVES
What Are Multi-stakeholder Initiatives?
MSIs reflect voluntary initiatives, which involve public and private as well as profit and
non-profit actors (Fransen & Kolk 2007; Mena & Palazzo 2012). Based on multi-
stakeholder processes, MSIs define, implement and enforce rules that direct firms’ social
and environmental behavior in selected areas. Even though participation in MSIs is
voluntary, firms are expected to comply with the underlying rules once they have signed
up. The literature distinguishes three types of MSIs in the field of CSR (see e.g. Gilbert
et al. 2011): (1) principle-based MSIs – i.e. initiatives that define broad principles of
engagement without any monitoring and certification (e.g., the UN Global Compact); (2)
certification MSIs – i.e. initiatives that outline criteria for certifying factories along global
supply chains (e.g., Social Accountability 8000); and (3) reporting MSIs – i.e. initiatives
that outline frameworks for disclosing non-financial information (e.g., the Global
Reporting Initiative). Our study focuses on what sort of firms leave principle-based MSIs
using the example of the UNGC.
The interdisciplinary academic discourse on MSIs has focused on different topics
(see also de Bakker et al. 2019 for a review). First, a number of studies have looked at the
“production” of MSIs. These studies have highlighted the political nature of standard-
8
setting processes (Moog et al. 2015), the lack of inclusiveness of standard making and
governance (Boström 2006), and the role of input legitimacy (Glasbergen 2013). Second,
studies have looked at the “institutionalization” of MSIs, focusing, for instance, on how
different actors influence patterns of diffusion (Gulbrandsen 2014), the effects of
competition between MSIs on market creation processes (Reinecke et al. 2012), and
isomorphism among standard adopters (Manning & von Hagen 2010). This stream of
literature has also discussed possible benefits of adoption that drive institutionalization
processes (Amer 2018). Finally, some studies have looked at the (lack of) impact of MSIs.
Scholars have examined whether MSIs have any effect on the practices of participants
(Clark & Kozar 2011), while other studies have more directly researched MSIs’ impact
on the problems they are claiming to address (Oosterveer et al. 2014).
Depending on the type of MSI, participants can leave such initiatives in different
ways: (1) Principle-based MSIs usually demand a mandatory annual disclosure of
implementation progress from participants; non-reporting participants are delisted (e.g.,
the Principles for Responsible Investment require an annual disclosure of participants’
progress vis-à-vis their principles); (2) certification MSIs delist participants that fail to
reach certain performance standards during audits; and (3) all types of MSIs allow that
firms decide to voluntarily withdraw (e.g., Volkswagen left the UNGC after discussions
around the manipulation of diesel engines started in September 2015).
The UN Global Compact
Launched in 2000, the UNGC offers businesses the opportunity to voluntarily align their
business practices with ten principles in four issue areas (i.e. human rights, labor rights,
9
environmental protection, anti-corruption). The ten principles are not tied towards a
particular sector, region or type of corporation; rather, they are supposed to be universally
valid (Kell 2013). While the UNGC primarily aims at enlisting businesses in support of
the ten principles, non-business actors (e.g., NGOs) can also join the initiative. The
UNGC has created rather low entry barriers. Corporations wishing to join have to send a
Letter of Commitment to the UN Secretary-General. Firms also need to provide some
basic information via an online application form (e.g., size of business, country). To
support the contextualization of the ten principles and to embed national communities for
interaction, the UNGC has supported the creation of local networks. Such networks
reflect “clusters of participants who come together voluntarily to advance the Global
Compact and its Principles at the local level […] by providing on-the-ground support and
capacity-building tied to distinct cultural, economic and linguistic needs” (Whelan 2010,
p. 318). So far, such networks have been established in over 80 countries. UNGC
participants can, but do not have to, join these networks.
The UNGC does not monitor whether participants live up to their commitment.
Rather, all business participants have to submit an annual Communication on Progress
(COP) report, a light form of self-regulation, in order to remain listed as “active”. The
COP is considered to be a public document (available via the UNGC’s website) and its
main purpose is to inform all stakeholders about a company’s efforts in support of the ten
principles (Hamid & Johner 2010). The content of COP reports is not verified by the
UNGC. COP reports are not standardized and only need to meet certain minimum
requirements. They need to include: (a) a statement by the chief executive expressing
continued support, (b) a description of practical actions in support of the four issue areas
(i.e. human rights, labor, environment, and anti-corruption), and (c) a measurement of
10
outcomes (UN Global Compact 2017). If a participant does not address one or more of
the four issue areas, it needs to explain the omission.
Whenever a firm submits a COP, which does not meet the basic requirements, it is
given a one-off, 12-month “grace period.” During this period, the participant is offered
support and guidance to come up with an adequate COP. Participants, which do not
submit a COP on time, are not immediately delisted. Rather, they are first designated
“non-communicating.” If a non-communicating participant does not submit a valid COP
report within another 12 months of becoming non-communicating, it is finally delisted
from the UNGC (UN Global Compact 2017). Hence, it can take up to 36 months to delist
a participant (initial 12 months of non-disclosure, plus 12 months grace period, plus 12
months non-communicating status). Other principle-based MSIs have adopted similar
mechanisms and grace periods (e.g., the Equator Principles). All delisted companies are
displayed on the UNGC website. Our analysis assumes that the decision to leave the
UNGC is a deliberate one, either because the firm does not want to (or cannot) produce
the COP. Prior to being delisted, participants receive a number of warnings from the
UNGC.
Despite the public nature of delistings and their high relevance for the UNGC,
surprisingly little research has focused on what sorts of firms leave the initiative and why
they do so. Knudsen’s (2011) early analysis of delistings provided some valuable insights.
She found that participants operating in countries with stronger domestic governance
institutions were less likely to delist. Further, she found that participants from Eastern
Europe and Africa have a higher likelihood of being delisted, while firms from the oil and
gas sector were less likely to be delisted. Although these results give important initial
directions, they also need to be treated with care, as the underlying dataset included only
11
227 firms that were delisted during the first six months of 2008. As we are working with
a dataset of the full population (i.e. all active and delisted participants from 2000-2015),
we complement the findings of Knudsen’s (2011) study.
HYPOTHESES DEVELOPMENT
Developing hypotheses, which predict what makes delisting more/less likely in the
context of the UNGC, emphasizes two phenomena in particular. First, MSIs can be seen
as voluntary clubs that act as signaling mechanisms. Prakash and Potoski (2007, p. 778)
argue that “club standards are signals to members’ stakeholders regarding what the
voluntary club wants members to accomplish.” Some of our hypotheses highlight the
UNGC’s role as such a signaling mechanism, for instance to investors (Amer 2018). But,
we also suggest that the UNGC’s role as a signaling mechanism is not equally important
to all types of companies, which, in turn, influences delistings.
Second, some of our hypotheses highlight that firms can reap reputational gains by
joining MSIs. Whether or not a participant can obtain reputational gains depends on a
number of factors that are likely to impact upon delistings (Berliner & Prakash 2014;
Prakash & Potoski 2007), for instance whether a firm’s stakeholders actually demand
MSI participation. If participants can secure reputational gains, this is likely to positively
impact their perceived legitimacy in the eyes of relevant stakeholders. The role of
legitimacy is important in the context of the UNGC, because 77% of all participants
joined the initiative in the hope that this would increase trust in the company and boost
social acceptance (UN Global Compact, 2011). Similarly, Arevalo et al.’s (2013, p. 11)
study of the UNGC found that firms were motivated “by the perceived opportunity to
achieve image gains.”
12
Corporate Size
We can expect that the size of corporate participants effects their ability and willingness
to remain part of the UNGC. Larger firms are more often exposed to reputational risks
and get more critical attention from stakeholders than smaller firms do (Kostova & Zaheer
1999; Prakash & Potoksi 2007). The more visible a company is to relevant stakeholders,
the higher its potential reputational risk (e.g., in case of NGO activism; see also Brown
& Knudsen 2012). Larger firms thus have more incentives to comply with the UNGC’s
COP policy, as they are more likely to face adverse reputational effects when being
delisted, especially since delistings are publicly communicated. Also, larger firms are
better able to use the “UN label”, for instance when interacting with NGOs (who target
larger firms more often than they do smaller ones; Sigwatch 2016).
Even though firms can get access to social legitimacy through other means (e.g., by
participating in other MSIs or by being listed on indices like the FTSE4Good), the UNGC
is an attractive initiative due to its association with the UN system. Despite constant
critique of its decisions and structures, the UN is a highly trusted organization because it
its viewed as a universal and inclusive organization (Torgler 2008). The Edelman (2017,
p. Q11-620) Trust Barometer finds that the UN enjoys much higher levels of trust around
the world than other international organizations (e.g., the International Monetary Fund)
and supranational entities (e.g., the European Union).
Larger firms are also more likely to possess the required resources that are
necessary to design a COP report (Brammer & Pavelin 2006). When signing up to the
UNGC corporations have to undergo some degree of change that requires financial as
13
well as non-financial resources (Rasche et al. 2013). Firms have to collect and analyze
relevant data and also write up the annual COP report. Companies with limited resources
will face difficulties in developing and submitting such a report and hence can be
expected to face a higher likelihood of being delisted. SMEs have on average fewer
resources available to manage their commitment to CSR (McWilliams & Siegel 2001),
and hence should face a higher likelihood of being delisted. The degree of required
organizational change also depends on the extant CSR engagement of a participant.
Companies, which were already engaged in CSR reporting prior to their commitment to
the UNGC, can submit these reports as a COP. This is due to the flexible COP policy of
the UNGC. As CSR reporting is particularly widespread among larger corporations, but
still not significantly developed among smaller firms (KPMG 2015), we can expect that
SMEs face a higher likelihood of being delisted, as smaller firms would have to create a
new reporting infrastructure from scratch. All of the above leads us to hypothesize:
Hypothesis 1a: SMEs are more likely to be delisted from the UNGC than larger
companies.
We can expect that the hypothesized effects of organization size on a firm’s ability
and willingness to submit a COP report are particularly strong for well-known global
firms given their assumed greater vulnerability to targeting by critical NGOs and the
mainstream media. Hence, we include an analysis of the delisting status of the Financial
Times (FT) Global 500 firms. These companies are ranked by market capitalization. FT
500 firms usually have significant resources devoted to CSR and have a well-developed
CSR infrastructure (e.g., dedicated departments; Baumann-Pauly et al. 2013). Also, FT
500 firms are disproportionally exposed to reputational risk and hence remain the main
14
target of NGO campaigning (Sigwatch 2016). Hence, we can posit that the magnitude of
the use of social legitimacy and its criticality are both high for these types of firms. We,
therefore, hypothesize:
Hypothesis 1b: FT 500 firms are less likely than other participants to be delisted
from the UNGC.
Ownership Status
We expect that participants’ ownership status affects the likelihood of being delisted. In
particular, we believe that publicly traded firms will behave differently than privately
held firms. Publicly traded firms have a high dependence on investors, who increasingly
recognize the importance of CSR issues for assessing the risks and opportunities related
to their portfolio (Eccles & Klimenko 2019). Publicly traded firms are particularly visible
and hence it is rather easy for investors to judge non-compliance. The extent to which an
organization complies with relevant external demands depends to some degree on
whether its actions or outputs are visible and can be assessed by others (Pfeffer &
Salancik 1978, p. 44). Such visibility matters in the context of the UNGC. Prior research
has shown that publicly traded companies that fail to submit a COP report are penalized
by financial markets with an average cumulative abnormal return of -1.6% over a period
of 5 trading days (Amer 2018). Investors seem to be aware of delistings and are prepared
to punish non-communicating companies (e.g., because of higher perceived risk levels).
This is not surprising, as the status of COP reports is available to investors via the
Bloomberg Professional service, which is one of the most widely used platforms for
financial professionals worldwide.
15
In the case of publicly traded companies, investors’ control of a critical resource
impacts participants’ actions (David et al. 2007). The larger institutional investors can
have a particular influence on companies’ attitude towards UNGC commitment. Over
2,300 institutional investors have signed up to the Principles for Responsible Investment
(PRI) thereby promising to integrate CSR-related concerns into their investment
decisions. The PRI emerged as a spin-off initiative from the UNGC and holds close ties
with the UN system. It can thus be expected that at least some PRI investors monitor the
COP status of publicly traded UNGC participants (Coulmont & Berthelot 2015; Gond &
Piani 2013). This exposes publicly traded firms to higher degrees of risk in the event of
being delisted. Further, prior research has found that publicly traded firms, which engage
in CSR activities, benefit from insurance-like effects if negative events occur (Shiu &
Yang 2017) and also decreased costs of equity capital (Dhaliwal et al. 2011). However,
such effects only exist if firms can credibly signal their CSR commitment to investors
(e.g., through COP reporting). Delisting from the UNGC would undercut such signaling
effects.
The role of investors needs to be considered together with the ability to regulate
corporate behavior (Drees & Heugens 2013). Both private and public regulators of
financial markets have emphasized the need to report on non-financial issues. In fact,
KPMG’s survey of the regulation of non-financial reporting emphasizes that many market
regulators by now expect publicly traded firms to annually report on their social and
environmental performance (relevant requirements exist, for example, in the EU, China
and Singapore; KPMG et al. 2016). Publicly traded firms often have to produce the
information, which is required for the COP report, because of existing financial
16
regulations. Hence, they do not need to invest additional resources when preparing the
mandatory COP report. All of these insights lead us to hypothesize:
Hypothesis 2: Publicly traded firms are less likely than other firms to be delisted
from the UNGC.
Early and Late Adopters
The existing literature suggests that early adopters are usually more committed to
promoting new norms because of their high degrees of ideational commitment
(Finnemore & Sikkink 1998) or their anticipated efficiency gains (Tolbert & Zucker
1983). By contrast, late adopters are primarily driven by isomorphism and conformity,
which can enable decoupling behavior (Bromley & Powell 2012). Although such
behavior of early and late adopters seems logical, we hypothesize that it is less relevant
in the case of the UNGC. There are two reasons for this. First, one assumption of the
literature is that early adopters base their actions on organizational platforms that help to
promote the underlying norms. While the UNGC reflects such a platform, it was also in
many ways a “historic experiment in learning and action” (Kell & Levin 2003, p. 176,
see also Ruggie 2002). Early adopters faced high levels of uncertainty around (a) what
exactly was expected from them and (b) which capabilities were needed to meet the
UNGC’s emerging requirements. The initiative changed many of its engagement
opportunities in the early days (e.g., the COP policy was adjusted multiple times and new
fora for learning and dialogue emerged; Rasche & Kell 2010). We can therefore expect
that early adopters lacked knowledge about what exactly the UNGC was, whether it was
of potential value to them, and whether their organization could meet the emerging
17
requirements specified by the organizational platform. Hence, the organizational
platform, which early adopters faced, was only weakly defined.
Second, early adopter behavior is also likely to be influenced by the availability of
alternative legitimacy sources. In the case of the UNGC, it is reasonable to assume that
early adopters had comparatively fewer alternative sources of legitimacy and hence were
more easily attracted by the initiative, even if the Compact did not fully fit their needs.
At the time of its launch (in 2000), few competing principle-based MSIs were around.
The World Business Council for Sustainable Development already existed, but it enjoyed
rather low levels of legitimacy due to its closed-club mentality. Many principle-based
MSIs with a sector or issue focus only emerged after the launch of the UNGC – for
instance, the Responsible Business Alliance was inaugurated in 2004. By contrast, late
adopters could chose (a) among a greater variety of principle-based MSIs and (b) often
had the opportunity to join principle-based MSIs that specialized on CSR issues in their
respective sector or addressed issues of high salience for their operations. We can thus
expect that late adopters had to decide more deliberately for the UNGC (and against other
MSIs), while at least some early adopters can be expected to have signed up to the UNGC
for a lack of anything better (Baccaro & Mele 2011). We therefore hypothesize that:
Hypothesis 3: Early adopters are more likely than late adopters to be delisted
from the UNGC.
UNGC Local Networks
Delisting behavior cannot be seen in isolation of the national context in which
participating firms are embedded. We suggest that the level of delistings is also influenced
18
by the degree to which the UNGC can influence participants “on the ground.” This
emphasizes the role of the UNGC’s local networks, which were set up as a way to guide
and impact participants in their respective national contexts (Abdelzaher et al. 2019).
Local networks only exist in selected countries. In those countries where such networks
exist, the global UNGC Office in New York can focus its influence attempts on only one
target (i.e. the local network focal point) and hence influence participants more easily
through this national intermediary. The main function of local networks is to influence
participants’ behavior and to make sure that firms submit their required COP (Whelan
2010). Many networks support their participants in creating COP reports. For instance,
the 2014 Local Networks Report states that “local networks received in-depth training
and guidance to support their participants with issues related to reporting and fulfilling
their COP […] requirements.” (UN Global Compact 2014, p. 14). We can therefore
expect that the UNGC can better influence participants to submit COP reports in countries
where local network structures exist. We thus hypothesize:
Hypothesis 4a: The delisting rate for firms originating from countries where local
networks exist is lower than for firms originating from countries without such
networks.
It is necessary to acknowledge differences between local networks. Prior research
has suggested that the level of engagement with a particular local network is dependent
upon the number of network participants (Rasche 2012). As UNGC participants are not
forced to join the local network in their respective country, we can expect networks with
a high rate of participation to also show higher levels of activity (e.g., because firms have
a basic interest in participation when they join such a network). Also, local networks with
19
more participants are likely to have access to more resources (e.g., because they utilize
in-kind donations by participants) which in turn can impact the level of activity. We can
therefore expect that network strength (i.e. the share of firms in a country deciding to
engage in a local network) influences whether participants see value in such networks
and, by extension, in the UNGC as a whole (see also Helmchen 2010). If participants
perceive local networks as active and valuable, they should be more interested in avoiding
delisting. We therefore also hypothesize:
Hypothesis 4b: The delisting rate for firms originating from countries where local
networks attract a high share of participants is lower than for firms originating
from countries where local networks attract a low share of participants.
METHODOLOGY
Dataset
Our analysis is based on the complete dataset of the UNGC’s active and delisted business
participants as of 31 December 2015 (n=15,853). The Global Compact Office in New
York provided the dataset. We excluded participants, which joined between 1 January
2013 and 31 December 2015, as due to the UNGC’s COP policy, firms that joined during
these years cannot be delisted (UN Global Compact 2017). Thus, a firm that joined on 1
January 2013 cannot be delisted before 1 January 2016 due to the current submission
deadlines and available grace periods. Our analysis therefore includes all participants that
joined the UNGC before 1 January 2013, and that were either labeled “active” or
“delisted” by 31 December 2015. From this universe, we excluded firms which left the
UNGC for the following reasons: merger or acquisition (n=60), organization does not
exist anymore (n=194), consolidation of commitment under the parent company (n=45),
20
and other reasons (n=17). This results in a final sample size of 11,499 firms (i.e. all firms
that joined before 1 January 2013 and that were mentioned as “active” or that were
“delisted” due to non-reporting or voluntary withdrawal).
Overall, the UNGC delisted 7,017 firms until December 2015. Of these, the vast
majority were expelled for failure to submit a COP (84.81%), while only a minority was
delisted because of requested withdrawal (11.33%) or other reasons (3.86%). Our
analysis does not further differentiate between delisting due to non-reporting and delisting
due to voluntary withdrawal. In the case of the UNGC, expulsion is not analytically
different from withdrawal. Firms, which leave because they fail to submit a COP report,
are not interested in submitting such a report and hence make the deliberate decision to
leave. We can assume such deliberate intention because the requirements for creating a
COP are minimal and firms receive multiple warnings prior to being expelled (see above).
Expulsion and requesting withdrawal would only be analytically different if participants
actually wanted to stay in the UNGC but were forced out of the initiative (e.g., because
the submitted document would not meet certain criteria).
Our dataset does not include non-business participants (e.g., NGOs). These
organizations have different reporting requirement (the so-called “Communication on
Engagement” policy). Furthermore, the delisting of non-business participants only started
in October 2016 and hence limited data are available at this stage.
Variables
As the aim of this study is to find systematic patterns of firms being delisted from the
UNGC, our dependent variable is whether a firm is delisted or not from the UNGC. We
21
create a dummy variable where 1 indicates that a firm was delisted and 0 that a firm is
still active. Given our final sample size (n=11,499), we include firms that are listed as
active (n=4,798) and firms that were forced to leave the UNGC through failure to submit
a COP report or that requested withdrawal (n=6,701) into our dataset.
The independent variables are defined according to our generated hypotheses. For
H1a, we measure the size of a firm according to the official recommendation by the EU
(EU 2003/361/EC) where microenterprises are firms with less than 10 employees, SMEs
with less than 250 employees, and corporations are firms with 250 or more employees.
We recoded the number of employees into a dummy variable where 1 reflects micro-
enterprises and SMEs and 0 larger firms. Such a classification is in line with prior
empirical work on SMEs (Berliner & Prakash 2012, p. 158). For H1b, we used the
information of whether a firm is enlisted in the FT500 index. Again, we created a dummy
variable with 1 reflecting “FT500 enlisted” (UNGC participant is listed in the Financial
Times Global 500 as of December 2015).
For hypothesis H2, which assumes that publicly traded firms are less likely to be
delisted from the UNGC, we collected the publicly traded status of a firm and created
again a dummy variable where 1 means “publicly traded” and 0 “private company,
subsidiary, or state-owned company.”
To test hypothesis H3, which assumes that early adopters are more likely to be
delisted, we analyzed the time at which firms joined the UNGC. To distinguish early and
late adopters, we chose 2008 as a cut-off year, as the UNGC had gained a robust
membership base until then (6,207 participants until the end of 2008). The UNGC started
out with 50 firms in 2000 and grew especially in 2006, 2007 and 2008 (as participants
from China and India started to join). Again, we created a dummy variable where 1
22
reflects “early adopters” and 0 “late adopters.” We also acknowledged that a possible
early adopter effect has to be distinguished from an exposure effect.
1
Considering the
exposure effect is relevant in this case, because if any UNGC participant faces an identical
probability of leaving in any given year, early adopters will have departed in greater
numbers because they were exposed to the initiative for a longer time. We addressed this
problem through a survival analysis that is taking the time to being delisted into account.
For that, we used a time variable called duration calculated for delisted firms by
subtracting the join year from the delisting year and for active firms by subtracting the
join year from 2015. The failure event is the dummy delisted.
To test hypothesis H4a, which concerns the relationship of local networks to the
likelihood of being delisted, we collected data on whether local networks in the respective
countries exist. This data was retrieved from the UNGC website. If a local network
existed, we differentiated between a weak and a strong network to address hypothesis
H4b. We classified all local networks in which more than 70% of country participants are
organized as strong networks. This threshold was chosen, as most networks above this
threshold showed a high level of activity and participant engagement (UN Global
Compact 2015). The resulting variable describing the existence and strength of a local
network in a particular country was then coded as 0 = “no network or no information
available”; 1 = “weak network defined as less than 70% of country participants are
organized in the network”; and 2 = “strong network defined as at least 70% of country
participants are organized in the network.” We then split this variable into three dummy
variables one capturing all firms with no network, the second all firms with a weak
network, and the last one including all firms with a strong local network. We include
23
dummy variables for countries, industry sectors, and whether a country is an OECD
member as control variables.
Models
To test our hypotheses, we employ a multilevel mixed-effects generalized linear model
(MMGLM) where our dependent variable is whether a firm is delisted or active. There is
a self-selection bias with regard to who signs up to the UNGC – that is, it can be assumed
that only firms anticipating a benefit from joining will participate. Hence, the
participating firms in the UNGC (or any other MSI) will not be representative of all firms.
Still, we do not correct such a self-selection bias, as we take all firms that joined the
UNGC as the population of interest to investigate which characteristics are related to
being active or being delisted.
We then control for contextual effects such as shared environments by employing
a multilevel mixed-effects generalized linear model. The UNGC database has a nested
structure with two levels: country and firms. In such a sample, the individual firm
observations are generally not independent as firms within one country are subject to
national legislation and other national factors (Hox 2002). The estimated model looks as
follows:
!"# $ % &'# ( &)*+"# ( &),-"# ( ."#
(1)
&'# $ &'( &/0'( 1'#
(country level)
where Dij is the measure of whether firm i in country j is delisted or active. β0j is the
average outcome in country j which is equal to the sum of the population average (β0),
the country specific effect (υ0j) and the effect of being an OECD member (O0). Xij
24
captures all independent variables relevant for testing the hypotheses and Cij comprises
dummies for the industry sectors. εi is the individual error-term. The composite model
looks as follows:
234$%56( 57
8934(57
:;34( 5<=(>64 ( ?34
(2)
When it comes to hypothesis H3 (early versus late adopters), we employ a survival
analysis to disentangle a potential early versus late adopter effect from a mere exposure
effect. If the probability of leaving the UN Global Compact is the same in any given year,
early adopters are delisted in greater numbers only because of their longer exposure to
the UNGC. Employing a multilevel mixed-effects parametric survival model
(MESTREG) enables us to disentangle a mere exposure effect from joining the UNGC
before 2008 or thereafter while still accounting for the nested structure of the data with
the two levels country and firms.
@34ABC $ % @6
A
B
C
%DEFA56( 57
8934(57
:;34( 5<=(>64 ( ?34
) (3)
Where
@34ABC
is the hazard function,
@6
A
B
C is the baseline hazard function, and t is the
time to a failure event. The other denotation remain as described above.
fixed effects
random effects
25
FINDINGS
Descriptive Statistics
The overall descriptive statistics and the Pearson correlations are presented in Table 1.
58.27% of the 11,499 firms are delisted. We take a first glance at Hypotheses H1 to H4
by inspecting the descriptive statistics with the final sample of 11,499 firms. In Table 2,
we present the descriptive statistics by listing status of a firm. Hypothesis H1a suggests
that SMEs are more likely to be delisted than larger firms. In total, 7,035 SMEs were
listed in the UNGC whereof 70.01% were delisted between 2000 and 2013. For larger
firms, only 40.77% were delisted over the same period of time. Hence, the share of SMEs
delisted is much larger than the one of participants with more than 250 employees –
supporting H1a. Hypothesis H1b suggests that FT500 firms are less likely to be delisted
from the UNGC than other participants. This hypothesis is supported. Out of all UNGC
participants, 187 firms were listed in the FT500. Only 4 of these 187 firms were delisted,
while the delisting rate of non-FT500 firms is at 59.19% (6,693 of 11,307 firms). H2
predicts that publicly traded firms are less likely than other firms to be delisted from the
UNGC. We find support for this hypothesis, as publicly traded firms have a lower share
of delisted firms (23.03%) than firms with other ownership structures (62.26%).
=====================
Put Tables 1 & 2 About Here
=====================
Hypothesis H3 suggests that late adopters are less likely to be delisted than early
adopters. We find that 67.87% of the early adopters are delisted, while later adopters face
26
a lower delisting rate (47.62%). Hence, this hypothesis seems to be supported at first
glance (however, see discussion below when taking the exposure effect into account).
Hypothesis H4a suggests that the delisting rate for firms originating from countries
where local networks exist is lower than for firms originating from countries without such
networks. This hypothesis was supported, as the delisting rate for firms originating from
countries without a local network is much higher (74.85%) than for firms originating from
countries with either strong or weak local networks (see below). Hypothesis H4b predicts
that firms originating from countries with strong local networks (i.e. networks where a
high number of country participants are organized) face a lower likelihood of being
delisted. This hypothesis is not supported. The delisting rate for firms from countries with
weak networks (51.52%) is even lower than for firms from countries with strong networks
(57.91%).
Regression Results
Table 3 presents the multilevel mixed-effects generalized linear model (MMGLM). We
estimated two models where we added the duration of membership in the second model
(Table 3, Column 2). The regression analyses produce similar results for all hypotheses
but H4 which concerns the influence of local networks. For the interpretation of the
estimates, we focus on the second model.
===========================
Put Table 3 & Figure 1 About Here
===========================
27
Also when controlling for all other variables that are hypothesized to be associated
with being delisted or not, the descriptive results are largely confirmed. For example,
SMEs have a higher likelihood of being delisted compared to larger firms (Odds Ratio
OR=3.504, p≤.001), meaning that the likelihood of being delisted is 3.504 times higher
compared to the likelihood of a larger firm being delisted. Hypotheses H1b and H2 are
also confirmed (Table 3). Firms with the following characteristics have a lower likelihood
of being delisted: being listed in the FT 500 and being publicly traded. Thus, hypotheses
H1a, H1b and H2 are supported.
At first glance, H3 (early adopters are more likely to be delisted) seems to be
supported by the regression results presented in Table 3 (OR=2.928, p≤.001). However,
once we separate a potential exposure effect from an early versus late adopter effect by
employing survival analyses regression, the results reverse and we find that early adopters
are less likely to be delisted than late adopters (Table 3, Column 3). This is visually
depicted in Figure 1 with the survival estimates, where we find the survival rates of early
adopters above the curve for late adopters. The divergence between the descriptive
statistics and the multilevel regression results (Table 3, Column 1 and 2) as well as the
survival analysis results can be explained by the exposure effect. The longer exposure of
15 years of potential membership of early adopters to the UNGC leads to a higher
likelihood of being delisted compared to the relatively short exposure of seven years of
potential membership for late adopters. Once the exposure effect is taken into account,
we find early adopters are less likely to be delisted than late adopters (OR=0.537, p≤.001).
Hence, H3 is not supported.
For the effect of local network strength on the likelihood of being delisted (H4b),
firms in countries with weak local networks are less likely to be delisted (OR=0.540,
28
p≤.001), but firms in countries with strong local networks are 2.870 times more likely to
be delisted than are firms in countries with weak networks. Countries explain 8.65% of
the variance of being delisted (ICC Intra Class Correlation coefficient). H4b is therefore
not supported.
DISCUSSION AND IMPLICATIONS
Theoretical Implications
Output Legitimacy of Principle-based MSIs. Principle-based MSIs are often criticized
for having low levels of output legitimacy, mostly because such initiatives do not have
monitoring mechanisms (Garsten & Jacobsson 2011; Sethi & Schepers 2014). Others
have argued that mandatory reporting requirements (such as the UNGC’s COP policy)
can at least partly offset such low levels of output legitimacy (Amer 2018; Hamid &
Johner 2010). This argument rests on the assumption that MSIs which can articulate a
credible threat that non-reporting participants will be removed from the initiative, can
increase their output legitimacy. Our results put such claims into perspective. The impact
of a delisting threat on MSIs’ output legitimacy is contingent upon participants’ size and
ownership status as well as their status as an early or late adopter. Reporting requirements
do not necessarily reflect credible threats and only support output legitimacy vis-à-vis
selected groups of participants. Our analysis shows that the credibility of the threat to
remove non-reporting participants depends most of all on participants’ expected
reputational losses. The ability of principle-based MSIs to increase their output
legitimacy via reporting requirements is therefore limited in the case of inclusive
29
initiatives (like the UNGC), because in such initiatives the reputational gains of
participation are asymmetrically distributed between larger and smaller firms.
Future theoretical discussions of principle-based MSIs’ output legitimacy (Kaan &
Liese 2012) have to follow a more fine-grained assessment, for instance by distinguishing
between inclusive and less inclusive initiatives. The output legitimacy of inclusive
initiatives is likely to suffer as long as mandatory reporting is the only available
enforcement mechanism because the reputational loss associated with non-reporting is
not equally relevant to all participants. By contrast, less inclusive principle-based MSIs
(e.g., initiatives that target specific types of participants, like the Equator Principles
targeting larger banks) should be able to strengthen their output legitimacy via the
introduction of reporting requirements, as long as the threat of removal remains credible
vis-à-vis the primary target group.
Early Adopters and Organizational Platforms. So far, there is little longitudinal
analysis of principle-based MSIs (mostly due to a lack of relevant data; for an exception
see Haack et al. 2012). Our analysis shows that early adopters are less likely to be delisted
and hence show higher levels of commitment. This result confirms existing theoretical
work that predicts more commitment among early adopters, while late adopters are
primarily motivated by isomorphism and conformity (Finnemore & Sikkink 1998;
Tolbert & Zucker 1983). Our results therefore highlight the importance of seeing early
UNGC adopters as norm entrepreneurs – that is, firms who hold strong beliefs about the
underlying norms and highlight their importance and appropriateness by joining early.
What is interesting about our results is the role of organizational platforms that are
supposed to support the behavior of norm entrepreneurs.
30
Finnemore & Sikkink (1998, p. 899-900) saw such platforms as an important
prerequisite for norms to emerge and institutionalize. As discussed above, the UNGC
acted as such a platform; however the initiative provided a weakly defined and rather
unstable platform in the early days. Our results can therefore be interpreted in the
following way: norm entrepreneurs show high levels of commitment to a principle-based
MSI even in those cases where relevant organizational platforms are still weakly defined.
Why, then, are norm entrepreneurs attracted by platforms that seem to be under
development? One reason is their rather high level of ideational commitment (Payne
2001) to the norms that the MSI promotes. This commitment may compensate perceived
weaknesses of the underlying platform, or it may even lead to a situation in which norm
entrepreneurs value the rather non-specified character of a platform (because it enables
experimentation and reframing within existing standards of “appropriateness”; March &
Olsen 1998).
However, our results also point to the relevance of another factor. Organizational
platforms do not exist in isolation; they are often embedded into larger organizational
structures (Gest et al. 2013). Although the UNGC was still emerging as a platform for the
work of norm entrepreneurs, the initiative has benefitted from the relative stability
provided by the UN system (Kell 2013). Hence, the broader organizational context in
which the platform was embedded and supported by selected state actors, provided
enough continuity to enable the work of norm entrepreneurs and to prevent them from
leaving the initiative. This implies that early adopter commitment may be less shaped by
the comprehensiveness and sophistication of organizational platforms but rather by their
very existence and supporting organizational environment. As many principle-based
MSIs have emerged out of existing organizations (e.g. the Principles for Responsible
31
Banking emerged out of UNEP), there is a need to further study the relationship between
the platforms that these initiatives themselves provide and their embeddedness into
permanent organizational environments with well-defined agendas.
Local Networks and Legitimacy Spillovers. Given that numerous MSIs have
adopted a nested network structure (i.e. local networks that are embedded in a larger
global network; Overdevest & Zeitlin 2014), it is important to consider whether and how
these local networks influence the MSI as a whole. Our analysis shows that the mere
existence of local networks influences the delisting rate, while the strength of a network
(measured by the number of active participants) does not seem to affect delistings
significantly. One factor that can explain the relatively lower delisting rate for firms
originating from a country with a local network is the existence of positive legitimacy
spillover effects within MSIs (see e.g. Haack et al. 2014). Prior research has argued that
“similarity” acts as a heuristic that allows outsiders to construct legitimacy transfers based
on established categories (Kostova & Zaheer 1999). Being part of a local network acts as
such an established cognitive category in that such networks unite firms that are similar
in terms of their support for the UNGC. Outside audiences thus confer legitimacy to
UNGC participants based on whether they can typify them into a category of
organizations that share a common feature (Bitektine 2011). In other words, similarity in
organizational features (e.g., being part of a UNGC local network) can help outsiders to
confer legitimacy to an organization (e.g., a firm participating in the UNGC) that is
otherwise hard to judge.
Leaving the UNGC would imply losing access to the local network and hence the
perceived association with a similar class of organizations. Although participants in local
networks are likely to differ in terms of their size and sector, they are all similar in terms
32
of being perceived as working towards goals that are organized under the UN umbrella.
As the UN enjoys high levels of legitimacy, we can expect that outsiders will confer
legitimacy to organizations that they can typify into a recognized cognitive category (i.e.
“UNGC local networks members”). We believe that such positive legitimacy spillovers
are tied to local networks, as these networks are the “face” of the UNGC on the local level
and hence more known to local audiences (e.g., local government agencies and NGOs;
see Helmchen 2010 and Whelan 2010). These results are also relevant beyond the UNGC.
They show that MSIs, which set up local networks, may be better able to keep participants
tied towards the initiative.
Policy Recommendations
Principle-based MSIs. Our study shows that policy makers have to find ways to protect
the legitimacy of principle-based MSIs. High delisting rates, as in the case of the UNGC,
are likely to damage the legitimacy of an initiative. The legitimacy of an MSI is hard to
judge from the outside and by non-experts. Hence, evaluators will rely on heuristics.
Publicly available information on delistings acts as such a heuristic and thus influences
MSIs’ perceived legitimacy. In case of a high delisting rate, firms with a strong CSR
profile will find it harder to credibly signal their performance through an MSI. As a result,
companies with a stronger CSR profile may shy away from participation. For policy
makers this implies that (a) it is important to sufficiently differentiate MSI participants
within an initiative (e.g., through different participation tiers) and (b) they should consider
how to adjust entry barriers. We suggest increasing entry barriers to a level that ensures
that participants have a basic willingness to engage in the initiative and the necessary
33
resources to do so. This could be done in different ways, for instance by increasing
reporting requirements (see below) or by asking for a sign-up fee.
Our results show also that inclusive principle-based MSIs face a challenge as they
cannot easily establish a credible threat vis-à-vis SMEs. However, we should not
conclude from this that relevant initiatives should ignore smaller firms. Prior literature
has emphasized that SMEs are vital when it comes to addressing political CSR problems
(Wickert 2016), for instance because they are part of global supply chains and are
therefore connected to regulatory gaps (which MSIs typically aim to address). Principle-
based MSIs have to find ways to connect reporting requirements to threats that are
perceived as credible in the eyes of SMEs. Since SMEs often have higher levels of
dependencies towards their local community (Spence 2016), the reputational loss for such
firms would be higher if MSIs manage to communicate the delisting decision to relevant
local actors (e.g., local NGOs, business associations, municipalities). Given that many
initiatives have created local network structures, such communication should, in
principle, be possible. MSIs’ current approach – to communicate delistings globally (e.g.,
via their website) and through professional services (like Bloomberg) – is unlikely to
influence SME behavior. Further, MSIs have to contextualize their assistance to SMEs
better. SMEs work at different levels of maturity when it comes to CSR and hence they
also require different levels of assistance (e.g., in preparing reports and tailored guidance
materials; see also Wickert 2016, p. 813).
UN Global Compact. One key recommendation would be to make COP reporting
a condition for entering the UNGC, instead of making it an outcome of participation. Our
results suggest that a large number of firms do not seem to be interested in aligning their
business practices and strategies with the UNGC’s ten principles. Hence, those who argue
34
that some firms misuse the initiative for “blue-washing” (Berliner & Prakash 2015) may
indeed have a point. Making the submission of a COP report a precondition for
participation would slightly increase entry barriers (see above) and also offer a baseline
from which to judge future implementation progress. The COP report could then outline
how the ten principles are addressed at the time of joining, and what plans exist for the
future. Such a policy change would also ensure that new participants have gained some
experience with reporting.
Our results also suggest that the UNGC currently has no mechanism in place that
allows deviant companies to be punished. The UNGC assumes that delisted firms are
punished through the public nature of delistings – that is, all expelled firms are listed on
a “name and shame” board on the initiative’s website (Hamid & Johner 2010). Our
findings suggest that this mechanism is likely to influence larger publicly traded firms,
while it does not show much effect on SMEs. As the exit option is not costly for SMEs,
there are incentives for these firms to free ride (i.e. join the initiative for three years
without any reporting). We thus recommend requiring asset-specific investments from all
new participants. Such investments create sunk costs and hence make the exit option more
costly (Prakash & Potoski 2007). For instance, such investments can be required by
making participation in UNGC local networks mandatory for all participants. This would
require firms to participate in and contribute to collective action activities. Because such
collective action activities are only accessible by network members, participants’
investment in them would be hard to regain.
35
CONCLUSION AND FUTURE RESEARCH
Given the growing attention to principle-based MSIs and their relevance for addressing
social and environmental problems, it is crucial to better understand what types of firms
are affected by this mode of private regulation. Our results show that firm size, ownership
structure, join year, and the existence of local networks influence whether firms decide
to leave MSIs. We do not claim that our findings hold for all types of MSIs. Our research
is based on an in-depth investigation of the UNGC – the largest principle-based MSI over
the last two decades. Our results are likely to be applicable in the context of other
principle-based initiatives, such as, but not limited to: the Principles of Responsible
Investment and the Equator Principles. Although these initiatives differ in some respects,
they also share some features (e.g., they require annual implementation reports and delist
non-compliant participants).
We see different avenues for future research. First, we need more empirical
evidence as to why certain types of firms decide to leave principle-based MSIs in general
and the UNGC in particular. While our study provides evidence that certain firms decide
to leave, we cannot say anything about the underlying reasons for this decision. Asking
why firms accept to be delisted seems important, because, as indicated above, at least in
the context of the UNGC, participants seem to agree to the delisting and do not attempt
to work against it. Looking at the recent literature on the UNGC (see e.g., Jastram &
Klingenberg 2018; Schembera 2018), a number of reasons seem possible but have not
been studied empirically so far. For instance, firms’ expectations regarding the private
benefits of participation may not have been met. Also, as the field of MSIs continues to
grow, firms may look for initiatives that better fit their individual industry-based needs,
36
while the UNGC remains a broad umbrella initiative that serves almost all industries.
Further, companies may go through internal change processes (e.g., change in leadership
or budget) resulting in a lower prioritization of UNGC participation. While the rationale
behind delistings can be explored quantitatively as well as qualitatively, we believe that
a qualitative multi-case-study research design (Stewart 2012) could help to gather first
insights which could then be further tested through quantitative studies.
Second, future research also needs to clarify the role of regulatory intermediaries
(Abbott et al. 2017) in the context of principle-based MSIs. Our results show that publicly
listed firms are less likely to leave MSIs. One reason for such behavior is the strong
influence of investors who see delisting as increasing their own risk. Understood in this
way, investors assume the role of indirect monitors that facilitate implementation and
prevent delisting. Future research needs to gain deeper insights into whether other
stakeholders than investors allow MSIs to indirectly influence participants. For instance,
it is likely that governmental actors can play a similar role in those national contexts
where public procurement decisions are impacted, among other things, by firms’
participation in MSIs.
37
ENDNOTES
1
!We thank Reviewer #3 for pointing this out.!!
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!
38
TABLES
Table 1. Descriptive statistics and correlations (dependent and independent variables)
#
Variable
Mean
SD
Min
Max
1
2
3
4
5
6
7
8
1
Delisted
0.58
0.49
0
1
1
2
SME
0.60
0.49
0
1
0.29***
1
3
FT500
0.02
0.13
0
1
-0.15***
-0.16***
1
4
Publicly traded
0.10
0.30
0
1
-0.24***
-0.36***
0.38***
1
5
Early adopter
0.53
0.50
0
1
0.21***
-0.12***
0.07***
0.06***
1
6
Local network: none
0.12
0.32
0
1
0.12***
-0.02***
-0.03***
-0.04***
0.10***
1
7
Local network: weak
0.26
0.44
0
1
-0.08***
0.00
0.05***
0.06***
0.01
-0.22***
1
8
Local network: strong
0.62
0.48
0
1
-0.01
0.02*
-0.02*
-0.02*
-0.08***
-0.47***
-0.76***
1
Note: * p < 0.05, ** p < 0.01, *** p < 0.001;
39
Table 2. Testing hypotheses with descriptive statistics
Hypothesis
Variable
Total
Listed
Delisted
H1a
SMEs
n
7,035
2,064
4,819
(%)
(100%)
(29.99%)
(70.01%)
MNEs
n
4,780
2,734
1,882
(%)
(100%)
(59.23%)
(40.77%)
H1b
FT500
n
187
183
4
(%)
(100%)
(97.86%)
(2.14%)
Other
n
11,307
4,614
6,693
(%)
(100%)
(40.81%)
(59.19%)
H2
Publicly traded firms
n
1,168
899
269
(%)
(100%)
(76.97%)
(23.03%)
Other
n
10,331
3,899
6,432
(%)
(100%)
(37.74%)
(62.26%)
H3
Early adopters
n
6,048
1,943
4,105
(%)
(100%)
(32.13%)
(67.87%)
Late adopters
n
5,451
2,855
2,596
(%)
(100%)
(52.38%)
(47.62%)
H4a/b
No network/ no
information
n
1,368
344
1,024
(%)
(100%)
(25.15%)
(74.85%)
Weak network
n
2,970
1,440
1,530
(%)
(100%)
(48.48%)
(51.52%)
Strong network
n
7,161
3,014
34,147
(%)
(100%)
(42.09%)
(57.91%)
Total
n
(%)
11,499
(100%)
4,798
(41.73%)
6,701
(58.27%)
40
Table 3. MMGLM and MESTREG regression estimates of H1 to H4 on being delisted from the UNGC
Variables
(1)
Delisted
Odds ratios
(2)
Delisted
Odds ratios
(3)
Delisted
Hazard ratios
Hypothesis H1a/b
SMEs
3.472***
3.469***
2.313***
(0.292)
(0.292)
(0.163)
MNEs
ref.
ref.
ref.
FT 500
0.058***
0.058***
0.050***
(0.021)
(0.021)
(0.018)
Other firms
ref.
ref.
ref.
Hypothesis H2
Publicly traded firms
0.343***
0.343***
0.354***
(0.052)
(0.052)
(0.047)
Other firms
ref.
ref.
ref.
Hypothesis H3
Early adopters
2.928***
2.940***
0.537***
(0.285)
(0.286)
(0.040)
Late adopters
ref.
ref.
ref.
Hypothesis H4a/b
no network/ no information
ref.
ref.
ref.
local network exists
.419***
(0.067)
weak network
0.327***
0.662***
(0.048)
(0.064)
strong network
0.450***
0.716***
(0.067)
(0.069)
log likelihood
-6,366
-6,361
-17,916
ICC
8.65%
8.65%
12.62%
n Groups
141
141
141
N
11,497
11,497
11,499
Note: * if p < 0.05, ** if p < 0.01, *** if p < 0.001. Odds ratios/ Hazard ratios are presented, robust standard errors in parentheses. The dependent
variable is the dummy delisted, the controls include dummies of industry sectors on the firm level and a dummy for OECD on the country
level. MEST REG carried out Weibull distribution.
41
Figure 1: MESTREG survival regression results for early and late adopters
42
REFERENCES
Abdelzaher D, Fernandez WD, Schneper WD (2019) Legal rights, national culture and social
networks: Exploring the uneven adoption of United Nations Global Compact.
International Business Review 28, 12–24.
Abbott KW, Levi-Faur D, Snidal D (2017) Theorizing Regulatory Intermediaries. The
ANNALS of the American Academy of Political and Social Science 670, 14–35.
Amer E (2018) The penalization of non-communicating UN Global Compact’s companies by
investors and its implications for this initiative’s effectiveness. Business & Society 57,
255-291.
Arevalo JA, Aravind D, Ayuso S, Roca M (2013) The Global Compact: An analysis of the
motivations of adoption in the Spanish context. Business Ethics: A European Review 22,
1–15.
Baccaro L, Mele V (2011) For Lack of Anything Better? International Organizations and
Global Corporate Codes. Public Administration 89, 451–470.
Bartley T (2007) Institutional Emergence in an Era of Globalization: The Rise of
Transnational Private Regulation of Labor and Environmental Conditions. American
Journal of Sociology 113, 297–351.
Baumann-Pauly D, Wickert C, Spence L, Scherer A (2013) Organizing corporate social
responsibility in small and large firms: Size matters. Journal of Business Ethics 115, 693–
705.
Berliner D, Prakash A (2012) From norms to programs: The United Nations Global Compact
and global governance. Regulation & Governance 6, 149–166.
Berliner D, Prakash A (2014) The United Nations Global Compact: An institutionalist
perspective. Journal of Business Ethics 122, 217–23.
Berliner D, Prakash A (2015) “Bluewashing” the firm? Voluntary regulations, program design,
and member compliance with the United Nations Global Compact. Policy Studies Journal
43, 115–138.
Bitektine AB (2011) Towards a theory of social judgment of organizations: the case of
legitimacy, reputation, and status. Academy of Management Review 36, 151–79.
Boström M (2006) Regulatory credibility and authority through inclusiveness: Standardization
organizations in cases of eco-labelling. Organization 13, 345–367.
Brammer S, Pavelin S (2006) Voluntary environmental disclosures by large UK companies.
Journal of Business Finance & Accounting 33, 1168–88.
Bromley P, Powell W (2012) From Smoke and Mirrors to Walking the Talk: Decoupling in
the Contemporary World. Academy of Management Annals 6, 483-530.
Brown D, Knudsen JS (2012) Managing Corporate Responsibility Globally and Locally:
Lessons from a CR Leader. Business & Politics, 14(3), 1-29.
Clark M, Kozar JS (2011) Comparing sustainable forest management certifications standards:
A Meta-Analysis. Ecology and Society, 16(1) (online).
Coulmont M, Berthelot S (2015) The financial benefits of a firm’s affiliation with the UN
Global Compact. Business Ethics: A European Review 24, 144–157.
43
David P, Bloom M, Hillman AJ (2007) Investor activism, managerial responsiveness, and
corporate social performance. Strategic Management Journal 28, 91–100.
de Bakker FGA, Rasche A, Ponte S (2019) Multi-Stakeholder Initiatives on Sustainability: A
Cross-Disciplinary Review and Research Agenda for Business Ethics. Business Ethics
Quarterly 29, 343-383.
Dhaliwal DS, Li OZ, Tsang A, Yang YG (2011) Voluntary nonfinancial disclosure and the cost
of equity capital: The initiation of corporate social responsibility reporting. Accounting
Review 86, 59–100.
Drees JM, Heugens P (2013) Synthesizing and Extending Resource Dependence Theory: A
Meta-Analysis. Journal of Management 39, 1666–1698.
Eccles RG, Klimenko S (2019) The investor revolution: Shareholders are getting serious
about sustainability. Harvard Business Review 97, 106–117.
Edelman (2017) Edelman Trust Barometer 2017. Edelman, New York, NY.
Finnemore M, Sikkink K (1998) International norm dynamics and political change.
International Organization 52, 887–917.
Fransen L (2012) Multi-stakeholder governance and voluntary programme interactions:
Legitimation politics in the institutional design of corporate social responsibility. Socio-
Economic Review 10, 163–192.
Fransen LW, Kolk A (2007) Global rule-setting for business: A critical analysis of multi-
stakeholder standards. Organization 14, 667–684.
Garsten C, Jacobsson K (2011) Transparency and legibility in international institutions: the
UN Global Compact and post-political global ethics. Social Anthropology/Anthropologie
Sociale 19, 378–393.
Gest J, Armstrong C, Carolan E et al. (2013) Tracking the Process of International Norm
Emergence: A Comparative Analysis of Six Agendas and Emerging Migrants’ Rights.
Global Governance 19, 153–186.
Gilbert DU, Rasche A, Waddock S (2011) Accountability in a global economy: The emergence
of international accountability standards. Business Ethics Quarterly 21, 23–44.
Glasbergen P (2013) Legitimation of Certifying Partnerships in the Global Market Place.
Environmental Policy and Governance 23, 354–367.
Gond J-P, Piani V (2013) Enabling institutional investors’ collective action: The role of the
Principles for Responsible Investment initiative. Business & Society 52, 64–104.
Gulbrandsen LH (2014) Dynamic governance interactions: Evolutionary effects of state
responses to non-state certification programs. Regulation & Governance 8, 74–92.
Haack P, Pfarrer MD, Scherer AG (2014) Legitimacy-as-feeling: How affect leads to vertical
legitimacy spillovers in transnational governance. Journal of Management Studies 51,
634–66.
Haack P, Schoeneborn D, Wickert C (2012) Talking the Talk, Moral Entrapment, Creeping
Commitment? Exploring Narrative Dynamics in Corporate Responsibility
Standardization. Organization Studies 33, 815–845.
Hamid U, Johner O (2010) The United Nations Global Compact communication on progress
policy: Origins, trends and challenges. In: Rasche A, Kell, G. (eds) The United Nations
44
Global Compact: Achievements, Trends and Challenges, pp. 265–280. Cambridge
University Press, Cambridge.
Helmchen C (2010) Running a Global Compact Local Network: Insights from the Experience
in Germany. In: Rasche A., Kell, G. (eds) The United Nations Global Compact:
Achievements, Trends and Challenges, pp. 355–369. Cambridge University Press,
Cambridge.
Hox J (2002) Multilevel analysis: Techniques and applications. Lawrence Erlbaum Associates
Publishers, London.
Jastram S, Klingenberg J (2018) Assessing the Outcome Effectiveness of Multi-Stakeholder
Initiatives in the Field of Corporate Social Responsibility – The Example of the United
Nations Global Compact. Journal of Cleaner Production 189, 775–784.
Kaan C, Liese A (2011) Public Private Partnerships in Global Food Governance: Business
Engagement and Legitimacy in the Global Fight against Hunger and Malnutrition.
Agriculture and Human Values 28, 385–399.
Kell G (2013) 12 years later: Reflections on the growth of the UN Global Compact. Business
& Society 52, 31–52.
Kell G, Levin D (2003) The Global Compact Network: An Historic Experiment in Learning
and Action. Business & Society Review 108, 151-181.
Knudsen J (2011) Company delistings from the UN Global Compact: Limited business demand
or domestic governance failure? Journal of Business Ethics 103, 331–349.
Kostova T, Zaheer S (1999) Organizational legitimacy under conditions of complexity: The
case of the multinational enterprise. Academy of Management Review 24, 64–81.
KPMG (2015) Currents of change: The KPMG survey of corporate responsibility reporting
2015. KPMG, Amsterdam.
KPMG, GRI, UNEP, Centre for Corporate Governance in Africa. (2016). Carrots & sticks:
Global trends in sustainability reporting regulation and policy. Available at:
https://home.kpmg.com/xx/en/home/insights/2016/05/carrots-and-sticks-global-trends-
in-sustainability-reporting-regulation-and-policy.html (Accessed: 25 January 2017).
Manning S, von Hagen O (2010) Linking local experiments to global standards: How project
networks promote global institution-building. Scandinavian Journal of Management 26,
398–416.
March JG, Olsen JP (1998) The Institutional Dynamics of International Political Orders.
International Organization 52, 943–969.
McWilliams A, Siegel D (2001) Corporate social responsibility: A theory of the firm
perspective. Academy of Management Review 26, 117–127.
Mena S, Palazzo G (2012) Input and output legitimacy of multi-stakeholder initiatives. Business
Ethics Quarterly 22, 527–556.
Moog S, Spicer A, Böhm S (2015) The politics of multi-stakeholder initiatives: The crisis of
the Forest Stewardship Council. Journal of Business Ethics 128, 469–493.
Oosterveer P, Adjei BE, Vellema S, Slingerland M (2014) Global sustainability standards and
food security: Exploring unintended effects of voluntary certification in palm oil. Global
Food Security 3, 220–226.
45
Overdevest C, Zeitlin J (2014) Assembling an Experimentalist Regime: Transnational
Governance Interactions in the Forest Sector. Regulation & Governance 8, 22–48.
Payne RA (2001) Persuasion, Frames and Norm Construction. European Journal of
International Relations 7, 37–61.
Pfeffer J, Salancik G (1978) The external control of organizations: A resource dependence
perspective. Harper and Row, New York.
Prakash A, Potoski M (2007) Collective action through voluntary environmental programs: A
club theory perspective. Policy Studies Journal 35, 773–92.
Rasche A, Kell G (2010) The United Nations Global Compact: Achievement, trends and
challenges. Cambridge University Press, Cambridge.
Rasche A (2012) Global policies and local practice: Loose and tight couplings in multi-
stakeholder initiatives. Business Ethics Quarterly 22, 679–708.
Rasche A, Waddock S, McIntosh M (2013) The United Nations Global Compact: Retrospect
and prospect. Business & Society 52, 6–30.
Reinecke J, Manning S, von Hagen O (2012) The emergence of a standards market: Multiplicity
of sustainability standards in the global coffee industry. Organization Studies 33, 791–
814.
Ruggie JG (2002) Trade, Sustainability and Global Governance. Columbia Journal of
Environmental Law 27, 297–307.
Scharpf FW (2009) Legitimacy in the multilevel European polity. European Political Science
Review 1, 173–204.
Schembera S (2018) Implementing Corporate Social Responsibility: Empirical Insights on the
Impact of the UN Global Compact on Its Business Participants. Business and Society 57,
783–825.
Sethi S, Schepers D (2014) United Nations Global Compact: The Promise-Performance Gap.
Journal of Business Ethics 122, 193–208.
Shiu Y-M, Yang S-L (2017) Does engagement in corporate social responsibility provide
strategic insurance-like effects? Strategic Management Journal 38, 455–70.
Sigwatch (2016) Corporations that NGOs loved and hated in 2015. Sigwatch, London.
Spence LJ (2016) Small Business Social Responsibility: Expanding Core CSR Theory.
Business & Society 55, 23–55.
Stewart J (2012) Multiple-case study methods in governance-related research. Public
Management Review 14, 67-82.
Tolbert PS, Zucker LG (1983) Institutional Sources of Change in the Formal Structure of
Organizations: The Diffusion of Civil Service Reform, 1880-1935. Administrative
Science Quarterly 28, 22–39.
Torgler B (2008) Trust in international organizations: An empirical investigation focusing on
the United Nations. Review of International Organizations 3, 65-93.
UN Global Compact. (2011) Annual Review of Business Policies and Actions to Advance
Sustainability. United Nations Global Compact Office, New York.
UN Global Compact. (2014) Local networks: Building a better future, together. United Nations
Global Compact Office, New York.
46
UN Global Compact (2015) Local network report 2015. United Nations Global Compact
Office, New York.
UN Global Compact (2017) Communication on Progress Policy, Available at:
https://www.unglobalcompact.org/docs/communication_on_progress/COP_Policy.pdf
(Accessed: 13 January 2017).
Wickert C (2016) “Political” Corporate Social Responsibility in Small- and Medium-Sized
Enterprises: A Conceptual Framework. Business & Society 55, 792–824.
Whelan N (2010) Building the United Nations Global Compact local network model: History
and highlights. In: Rasche A, Kell, G. (eds) The United Nations Global Compact:
Achievement, trends and challenges, pp. 317–339, Cambridge University Press,
Cambridge.
