![Theoretical framework [Colour figure can be viewed at wileyonlinelibrary.com]](https://www.researchgate.net/profile/Jianjun-Zhu-2/publication/352719147/figure/fig1/AS:1044911931543552@1626137826078/Theoretical-framework-Colour-figure-can-be-viewed-at-wileyonlinelibrarycom_Q320.jpg)
Content uploaded by Jianjun Zhu
Author content
All content in this area was uploaded by Jianjun Zhu on Jul 13, 2021
Content may be subject to copyright.
British Journal of Management, Vol. 0, 1–25 (2021)
DOI: 10.1111/1467-8551.12537
The Effect of Inter-rm Managerial Social
Ties on Alliance Formation: How Poorly
Embedded Firms Overcome Network
Disadvantages
Ke Yang ,1Jianjun Zhu 2and Michael D. Santoro1
1College of Business, Lehigh University, 621 Taylor Street, Bethlehem, PA, 18015, USA 2College of Business,
New Mexico State University, P.O. Box 30001, Las Cruces, NM, 88003, USA
Corresponding author email: key208@lehigh.edu
Recent research suggests pre-existing network structural dynamics can hamper inter-
organizational alliance formation, particularly for poorly embedded rms. However, this
research fails to fully explain how poorly embedded rms can overcome these structural
barriers when forming alliances. This study adds to the existing knowledge on network
dynamics by proposing how interpersonal relational embeddedness alleviates such con-
straints and facilitates alliance formation of rms poorly embedded in existing alliance
networks. We highlight the informational and sociological benets of inter-rm manage-
rial social ties (IFMSTs) formed through interpersonal social experiences beyond those
originated from prior business exchange activities between potential partnering rms.
Our empirical investigation employs an extensive sample of US public rms and shows
strong evidence that IFMSTs help poorly embedded rms overcome the obstacles imposed
by their inferior network structure embeddedness when forming alliances. The negative
effect of inferior network structural embeddedness is more muted by IFMSTs when both
rms in a dyad are at the periphery of the alliance network than when they occupy asym-
metric structural positions. This mitigation effect of IFMSTs is more salient when the
ties originate from activity-based social experiences than from shared educational afli-
ations.
Introduction
Researchers of social embeddedness have long
recognized the importance of informational and
sociological benets derived from prior inter-
organizational transactional experiences (Gulati,
2007; Powell et al., 2005; Uzzi, 1997). From a
structural embeddedness perspective, the litera-
ture argues that centrally located rms in an ex-
isting alliance network are more likely to form
new alliances with each other due to their ad-
vantageous network positions (Chung, Singh and
Lee, 2000; Gulati and Gargiulo, 1999; Rosenkopf
and Padula, 2008; Rosenkopf, Metiu and George,
2001). This tendency to replicate and reinforce ex-
isting relationships and network structures under-
scores the path-dependent consequences of em-
beddedness, thereby constraining alliance forma-
tion by poorly embedded rms (Gulati, 2007).
However, researchers adopting the structural
perspective have yet to provide an adequate ex-
planation for how poorly embedded (peripheral)
rms in an existing alliance network overcome
the informational and sociological barriers to
form alliances with other rms. For example,
studies of structural homophily (Podolny, 1994)
provide arguments for alliance formation between
rms holding central positions in the existing
network, while leaving the dynamics of alliance
formation by those poorly embedded rms largely
© 2021 British Academy of Management and Wiley Periodicals LLC. Published by John Wiley & Sons Ltd, 9600 Gars-
ington Road, Oxford OX4 2DQ, UK and 350 Main Street, Malden, MA, 02148, USA.
2K. Yang, J. J. Zhu and M. D. Santoro
unexplained. Therefore, the alliance behaviour of
poorly embedded rms still remains a research
question with important theoretical implications,
as the validity of their path-creation potential may
have a profound impact on network dynamics
(Ahuja, Soda and Zaheer, 2012; Ahuja et al.,
2009; Rosenkopf and Padula, 2008; Rosenkopf
et al., 2001).
To explore possible alternatives, researchers
conjecture that peripheral rms in the alliance net-
work can form socially asymmetric alliances with
central rms if they possess unique resources that
a well embedded rm could not obtain elsewhere,
for example radical technological breakthroughs,
and if the peripheral rm is willing to accept un-
favourable trade terms (Ahuja et al., 2009; Gulati
and Gargiulo, 1999). While important to expand
our understanding of how poorly embedded rms
may attract their highly embedded counterparts
to form alliances, these studies also raise a ques-
tion about the sustainability of such socially asym-
metric deals. Furthermore, this social asymmetry
strategy still does not explain how poorly embed-
ded rms ally with each other. To further illus-
trate the signicance of this void in understanding
these network dynamics, over 68% of all alliances
formed among S&P 1500 rms over our sample pe-
riod of 9 years involve at least one peripheral rm,
while one-third of the alliances are among two pe-
ripheral rms.
In this study, we provide new insights into how
poorly embedded rms overcome the barriers to
alliance formation. To do so, we highlight the im-
portance of the interactions between relational
and structural embeddedness. Specically, using
informational and sociological benets of inter-
personal relational embeddedness as our overar-
ching theme, our study explores the role of inter-
rm managerial social ties (henceforth IFMSTs)
among boundary spanners originated from social
experiences. We then take a nuanced approach to
further examine the potential compatibility be-
tween the needs of poorly embedded rms and the
benets engendered by IFMSTs. From a dyadic
perspective, the need for both informational and
reputational benets is greater when both poten-
tial partners face structural obstacles at the periph-
ery of the existing alliance network (Ahuja et al.,
2009). By contrast, such deciency is partially miti-
gated by the structural superiority possessed by the
centrally located rm in an asymmetric dyad, al-
though the structural asymmetry itself may create
both cohesive and divisive forces between the po-
tential partners (Hu, Jain and Delios, 2021). There-
fore, we expect peripheral dyads to benet more
from IFMSTs in overcoming structural barriers
when forming alliances. Moreover, not all interper-
sonal social ties are created equal, which promotes
an intriguing question of whether IFMSTs origi-
nated from different social experiences are equally
effective at alleviating the structural constraints
faced by poorly embedded rms. To provide in-
sights into this question of efcacy, we look to the
multi-dimensional nature of IFMSTs to determine
which types of IFMSTs are more effectual for the
poorly embedded rms in delivering benets piv-
otal to alliance formation.
We dene IFMSTs as interpersonal relationships
formed through a broad array of social experiences
including past education, employment or other so-
cial activities shared among executives and directors
across rms. Our IFMSTs differ from the interper-
sonal relationships used in a large number of prior
studies since they focused on repeated inter-rm
transactional exchanges (e.g. Barden and Mitchell,
2007). This distinction has important theoretical
implications for network dynamics and for the in-
uence of relational and structural embeddedness
on rm behaviour and alliance formation (An-
drevski, Brass and Ferrier, 2016; Polidoro, Ahuja
and Mitchell, 2011; Rowley, Behrens and Krack-
hardt, 2000). One key insight is that relational em-
beddedness empowered by IFMSTs unrelated to
prior inter-rm alliance activities is not subject to
the same path-dependency tendency imposed by
the structural embeddedness of an existing alliance
network. Therefore, it is possible that rms con-
nected through IFMSTs may ally with each other
despite a lack of structural prerogative to ally. Sim-
ilarly dened IFMSTs have been used in other
contexts such as acquisition, underwriting syndi-
cate formation, institutional trading and rms’ IT
components diversity (Cai, Walkling and Yang,
2016; Cohen, Frazzini and Malloy, 2008; Cohen
et al., 2010; Cooney Jr et al., 2015; Engelberg, Gao
and Parsons, 2012; Fracassi, 2016; Ishii and Xuan,
2014; Xue, Yang and Yao, 2018). However, this
study, to our best knowledge, is the rst to theo-
rize a mechanism centred on IFMSTs and provide
empirical support in the context of alliance forma-
tion.
Our focus on the interpersonal social con-
nections among the executives and directors is
twofold. First, top management teams have great
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 3
inuence on key strategic decisions such as alliance
formation (Certo, 2003; Dalton et al., 1998). The
notion that IFMSTs can facilitate information dis-
semination is intuitively appealing, as a handful of
recent studies have documented the informational
impact of IFMSTs on rm behaviours in various
business contexts (Cai, Walkling and Yang, 2016;
Cohen, Frazzini and Malloy, 2008; Cohen et al.,
2010; Ke et al., 2018). For example, in their pio-
neering study, Cohen, Frazzini and Malloy (2008)
investigated information transfer between portfo-
lio managers and executives of publicly traded
rms. They found that portfolio managers perform
signicantly better on the holdings of connected
rms due to information advantages channelled
through their personal ties.
Second, our IFMSTs denition differs from
the well-studied interlock relationships where
two or more rms are connected when the same
director sits on the boards of multiple rms (Gu-
lati and Westphal, 1999). While the interlocking
director represents a direct and observable link
between two rms, it only accounts for a tiny
fraction of an intricate social network among the
large number of corporate boundary spanners, in
which inter-organizational collaboration activities
such as alliance formations are embedded. For
instance, only 0.5% of our sample dyads share
interlocking directors, whereas more than 51.5%
are connected through IFMSTs. Therefore, fo-
cusing only on the interlocking directors ignores
the inuence of the vast majority of other, if not
more important, members of the top management
team – such as CEOs, CFOs and COOs, who are
constantly involved in the inter-organizational
strategic decision-making process. Furthermore,
our IFMSTs measure captures a wide array of
social experiences enabling informational and so-
ciological benets essential to alliance formation,
while interlock directors only capture one such
channel through professional afliation.
Finally, challenged by the empirically un-
bounded nature of inter-organizational col-
laborations, network researchers have chosen
signicantly different empirical settings, such as
the apparel industry in New York (Uzzi, 1997),
the biotech industry in Boston (Owen-Smith and
Powell, 2004) and the chemical industry in West-
ern Europe, Japan and the USA (Ahuja, 2000).
These studies provide us with in-depth knowledge
of inter-organizational collaboration within a
specic sector of interest. However, cross-industry
alliances have become a common and growing
phenomenon, accounting for over 80% of the
alliances in our sample. Therefore, we add to the
growing literature by investigating the inuence
of IFMSTs on alliance formation activities in a
broader context, encompassing organizational
actors in multiple industries with a specic focus
on poorly embedded rms.
Theory and hypotheses
Structural obstacles to alliance formation
Structural embeddedness stresses the importance
of the structural position in an alliance network
in affecting rm behaviours, thereby facilitating
alliance formation by mitigating the uncertainty
arising from information asymmetry and oppor-
tunistic behaviours between partnering rms (Ar-
ranz, Arroyabe and Fernandez de Arroyabe, 2020;
Gulati and Gargiulo, 1999; Jensen and Roy, 2008).
As a result, rms occupying central positions in
the existing alliance network are more likely to
ally with each other, since rms at the periphery
of the network often lack the informational and
sociological benets accrued to those in more cen-
tral positions (Chung, Singh and Lee, 2000; Gulati
and Gargiulo, 1999; Rosenkopf and Padula, 2008;
Rosenkopf, Metiu and George, 2001). This puts
rms at the periphery of an existing alliance net-
work at a disadvantage when it comes to forming
alliances (Ahuja, Polidoro and Mitchell, 2009;
Podolny, 1994). First, peripheral rms lack vis-
ibility with existing alliance network partners
(Rosenkopf and Padula, 2008; Stuart, Hoang and
Hybels, 1999). Second, in addition to obstacles in
disseminating information to successfully market
their own candidacy, poorly embedded rms also
tend to face higher barriers in acquiring infor-
mation about potential partners – such as joint
market opportunities and a potential partner’s
strategic objectives (Gulati and Gargiulo, 1999;
Podolny, 1994). Third, poorly embedded rms
often lack tacit information that is essential for
shaping joint sense-making (Fiol and Lyles, 1985)
and enhancing knowledge integration (Inkpen
and Tsang, 2005; Koka and Prescott, 2002). Com-
bined, this heightened information asymmetry and
reduced visibility diminishes peripheral rms’ per-
ceived attractiveness as potential alliance partners.
© 2021 British Academy of Management and Wiley Periodicals LLC.
4K. Yang, J. J. Zhu and M. D. Santoro
IFMSTs and rms’ structural and relational
embeddedness
Network researchers have differentiated relational
embeddedness from structural embeddedness.
Relational embeddedness refers to the inuences
of dyadic relationships, since the mechanisms of
relational embeddedness operate at both interper-
sonal and inter-organizational levels (Ahuja, 2000;
Barden and Mitchell, 2007; BarNir and Smith,
2002; Gulati and Westphal, 1999; Mitsuhashi and
Min, 2016; Rider, 2012). In contrast, as mentioned
above, structural embeddedness stresses the im-
portance of the structural position in an alliance
network in affecting rm behaviours (Arranz, Ar-
royabe and Fernandez de Arroyabe, 2020; Gulati
and Gargiulo, 1999; Jensen and Roy, 2008). Stud-
ies on structural embeddedness document that the
characteristics of a network (Rowley, Behrens and
Krackhardt, 2000) and rms’ structural position
in an alliance network (Ahuja, 2000; Moran,
2005; Rosenkopf and Padula, 2008; Shipilov, Li
and Greve, 2011) endow the actors with informa-
tional and sociological benets that facilitate the
subsequent formation of new relationships and
reinforce the existing network structure.
Despite the parallel importance of both rela-
tional embeddedness and structural embedded-
ness in enhancing our understanding of inter-
organizational behaviours and outcomes (Zaheer,
Gözübüyük and Milanov, 2010), only a handful of
studies have engaged both perspectives in exam-
ining inter-organizational activities (Andrevski,
Brass and Ferrier, 2016; Moran, 2005; Rowley,
Behrens and Krackhardt, 2000). For example,
Polidoro, Ahuja and Mitchell (2011) studied how
relational and structural embeddedness induces
tie stability, while Ahuja (2000) theorized how
prior inter-organizational collaboration ties and
structural holes interplay to affect the rm’s sub-
sequent innovation output. However, when both
relational embeddedness and structural embed-
dedness originate from the same network of prior
inter-organizational business exchange activities,
they are likely subject to the same self-reproducing
tendencies imposed by the network. For example,
executives of rms poorly embedded in the exist-
ing alliance network likely lack connections with
their counterparts at other rms when the social
context is limited to prior alliance activities.
In this study, we examine how relational and
structural embeddedness jointly impact alliance
formation by focusing on interpersonal relation-
ships between executives across rms. Specically,
how IFMSTs mitigate the negative effect of infe-
rior structural embeddedness, constraining periph-
eral rms when it comes to alliance formation. We
depict our research framework in Figure 1.
Hypothesis development
Interpersonal ties among executives established
through rms’ prior business exchange experi-
ences can facilitate future collaborations. However,
IFMSTs developed through a much broader ar-
ray of social experiences can deliver informational
benets beyond what has been captured by the
existing literature. IFMSTs create a unique open
network based on interpersonal relations that can
affect the volume and characteristics of informa-
tion exchange (Burt, 1992; Koka and Prescott,
2002). First, IFMSTs enhance mutual awareness
and being more visible increases a rm’s likelihood
of perhaps being chosen as an alliance partner
(Santoro, 2013). Second, IFMSTs provide rms
with a unique conduit for exchanging both ex-
plicit and tacit information relevant to evaluat-
ing the attractiveness of potential alliance part-
ners. Information such as a rm’s idiosyncratic
characteristics (Gulati, 1998), needs, objectives,
preferences and strategic orientation (Gulati and
Gargiulo, 1999), as well as proprietary informa-
tion on business operations and practices (Uzzi,
1997). Executives in a more embedded relation-
ship are more willing to share these various types
of otherwise hard-to-obtain aspects of organiza-
tional knowledge (Uzzi, 1999; Uzzi and Lancaster,
2004). Moreover, besides the increased quantity of
information exchange, IFMSTs may also enhance
the perceived quality of information, which fur-
ther induces more condence in the connected rm
as a potential alliance partner (McAllister, 1995).
Finally, prior studies show that people gravitate to-
wards relying more on information received from
people with whom they are socially connected than
those they are not (Hong, Kubik and Stein, 2005;
Huberman, 2001; Seasholes and Zhu, 2010).
Following these notions, IFMSTs can be effec-
tive in alleviating the challenges imposed by infe-
rior structural embeddedness endured by periph-
eral rms. We therefore propose the following:
H1a: IFMSTs positively moderate the negative ef-
fect of poor structural embeddedness (i.e.
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 5
Relational Embeddedness
IFMSTs
(Inter-firm Managerial Social Ties)
Activity-based IFMSTs
Education-based IFMSTs
Structural Embeddedness
Peripheral Dyad
(two poorly embedded firms)
Asymmetric Dyad
(one poorly and one centrally
embedded firm)
Alliance
Formation
Variation across
IFMSTs Types
Activity-Based Vs
Education-Based
IFMSTs
H1a
H1b
H2
H1a
H1b
Controlled Dyad Factors
Board Interlock
Past alliance knowledge
Firm financials and features
H2
Figure 1. Theoretical framework [Colour gure can be viewed at wileyonlinelibrary.com]
peripheral dyads with two peripheral rms
or asymmetric dyads with one peripheral
and one central rm) on alliance formation.
At a more nuanced level, a peripheral dyad will
benet more from IFMSTs since both partners,
due to their positions in the network structure, are
hindered by their informational and reputational
deciencies. In comparison, these deciencies are
partially mitigated by the social legitimacy and in-
formational advantages possessed by the centrally
located rm in an asymmetric dyad. Therefore,
ceteris paribus, IFMSTs’ informational and so-
ciological benets are marginally less important
to the asymmetric dyads. Our research question
focuses on the incremental effect of IFMSTs
in moderating the negative effect of structural
barriers on dyads facing varying degrees of infor-
mational and sociological challenges. Thus, our
arguments do not contradict the notion that when
IFMSTs reduce information asymmetry between
the partners, peripheral dyads may still be less
likely to form alliances than asymmetric dyads.
Moreover, we do not preclude other factors, such
as the value relevancy of potential alliances, from
further interacting with IFMSTs’ moderating
effect – both within and across these two types of
dyads. Following these notions, we posit:
H1b: IFMSTs exhibit a stronger inuence in re-
ducing the negative effect of poor structural
embeddedness on peripheral dyads than on
asymmetric dyads in alliance formation.
IFMSTs and their underlying social experiences
The efcacy of the alternative social mechanisms
engendered by IFMSTs in moderating the negative
effect of structural inferiority varies with the po-
tential compatibility between the benets provided
and the needs of poorly embedded rms seeking
alliance opportunities. For instance, prior studies
found that the inuence of structural homophily
is limited to highly embedded rms. However, the
similarity in structural inferiority, while it may in-
duce an incentive for coalition, does not help to
overcome the informational barrier nor address
the lack of mutual visibility that hinders alliance
formation among poorly embedded rms (Ahuja,
Polidoro and Mitchell, 2009). Our denition of
IFMSTs captures a wide variety of social experi-
ences, ranging from those resulting from having
similar educational afliation to more engaging
social activities such as professional interactions.
This raises an intriguing question: Are IFMSTs
from alternative social interactions equally effec-
tive at alleviating the constraints faced by periph-
eral rms?
To further understand IFMSTs’ inuence, we
take a nuanced approach to examine the efcacy
of ties originated from different social experiences
© 2021 British Academy of Management and Wiley Periodicals LLC.
6K. Yang, J. J. Zhu and M. D. Santoro
in helping poorly embedded rms overcome struc-
tural inferiority (Berends, van Burg and van Raaij,
2011; Butler and Gurun, 2012; Hong, Kubik and
Stein, 2005; McAllister, 1995; Shipilov, 2012). We
therefore differentiate IFMSTs cultivated through
more engaging social activities from those likely
rooted in a similarity-based prior educational
afliation (Cai, Walkling and Yang, 2016). Empir-
ically, operating within the boundary of the social
experiences captured by our sample and data
sources, we dene two subgroups of IFMSTs: (1)
activity based, dened as interpersonal ties devel-
oped through social activities such as working on
the same top management team, serving on the
board of trustees of a local art museum or playing
at the same golf club, etc. and (2) education based,
dened as ties established through a shared alma
mater, a widely recognized base for homophilic re-
lationships (Gomerps, Mukharlyamov and Xuan,
2016).
It has been well established that similarity
among individuals helps to discourage idiosyn-
cratic behaviour and foster trust as well as
reciprocity, which may help to facilitate inter-
organizational collaboration (Brass, 1985; Gra-
novetter, 1973). Therefore, to the extent that
the principle of homophily shapes individuals’
decision-making processes in a variety of settings,
executives may lean towards a homophily-driven
choice of alliance partners in lieu of a costly
search process. Educational background, gender
and ethnicity are among the characteristics widely
used by researchers to operationally identify
similarity among individuals (McPherson, Smith-
Lovin and Cook, 2001). For example, Gomerps,
Mukharlyamov and Xuan (2016) nd that getting
a degree from the same school increases the like-
lihood of two venture capitalists syndicating with
each other by 33.3%.
By contrast, ties cultivated through activity-
based social experiences, while they may well
constitute a basis for homophilic bias to come
into play, are more effectual in delivering both
the informational and sociological benets to
facilitate alliance formation than education-based
ties. We offer four reasons. First, the strength of
the relationship as an information dissemination
and acquisition channel is likely to be greater for
activity-based ties due to not only its more salient
engaging nature but also the sheer size difference
between the typical focal groups of a corporate
board versus an alumni network (Barbulescu,
2015; Brown, Gianiodis and Santoro, 2015; Hall,
2011). That is, the chances of direct interactions
between two directors serving on the same board
are signicantly higher than between two individ-
uals graduating from the same university. Second,
activity-based IFMSTs reinforce the affective
foundations for generating stronger emotional
bonds, attachment and kinship between individ-
uals (Lewis and Weigert, ), which often results in
more condence in each other’s track record of
productivity, sense of responsibility, and norms
of fairness and reciprocity (Evans and Wurster,
1997). This enhances mutual trust, further facili-
tates the exchange of information and knowledge,
and promotes agreement on working styles and
business solutions (Selnes and Sallis, 2003; Uzzi,
1996). Anecdotally to this point, an article recently
appeared in Forbes (Dosh, 2016) that claims 80%
of surveyed Fortune 500 executives say playing
golf enables them to establish new business re-
lationships. Third, activity-based ties are more
likely to strengthen the initial mutual awareness in
searching for potential alliance partners, especially
among poorly embedded rms impeded by height-
ened search costs due to information asymmetry.
Finally, activity-based IFMSTs may provide added
benets through fostering a cognitive belief in each
other’s competence, a sense of responsibility and
norms of fairness and reciprocity (Butler, 1991).
It is important to point out that our discussion
of the relative strength of activity-based IFMSTs
does not preclude benets derived from education-
based ties. Nonetheless, in comparison, activity-
based IFMSTs are likely to be more powerful than
education-based ties in affecting alliance forma-
tion. Formally, we propose:
H2: IFMSTs activity-based ties exhibit a stronger
positive moderation on the negative effect
of poor structural embeddedness (i.e. periph-
eral dyads and asymmetric dyads) on alliance
formation than do IFMSTs education-based
ties.
Research methods
Sample and data
To create our novel and comprehensive dataset,
we merged information from three well-recognized
data sources: (1) BoardEx by Management Diag-
nostic Limited for individual executives’ and
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 7
directors’ prole information; (2) CRSP/
Compustat merged Financial Database for rm-
level nancial information; and (3) SDC Joint
Ventures and Alliance Database by Thomson for
strategic alliance information.
BoardEx covers rms representing approxi-
mately 90% of US stock market capitalization
and offers a balanced representation of rms
of different sizes and business sectors. For each
individual covered by the database, BoardEx iden-
ties her/his name and traces her/his employment
history as board of director or senior executive. It
also includes educational background and social
activities based on information collected from
multiple sources, such as rms’ public lings and
individual curriculum vitae listedonrmwebsites.
This database has been used in a variety of organi-
zational behaviour settings to examine the roles of
social networks among rm executives and direc-
tors across several different disciplines (e.g. Cai,
Walkling and Yang, 2016; Cohen et al., 2010; En-
gelberg, Gao and Parsons, 2012; Fracassi and Tate,
2012; Ke et al., 2018; Xue, Yang and Yao, 2018).
To construct our sample of rm dyad-years
suitable for our empirical framework, we rst
used the BoardEx executives’ and directors’ prole
information to identify their afliations for each
sample year. We then constructed an annual time-
series measure of IFMSTs by cross-referencing
individuals’ employment history, educational
background and other social activity records that
pre-date the sample year of interest. Specically,
to identify educational ties between two individ-
uals, we rely on both the name of the university
they graduated from and the unique ID assigned
to each university by BoardEx. We applied both
programming algorithms and manual verication
to correct potential data errors and carefully trace
back individuals’ employment history to account
for the potential impact of signicant corporate
events – such as mergers and acquisitions during
the sample period.
We then aggregated the individual-level inter-
personal social ties to the rm dyadic level. Next,
we matched the BoardEx-based social tie measures
with data from the CRSP/Compustat databases to
obtain nancial and accounting information for
rms in the dyads. Lastly, we collected the sample
rms’ alliance activities from the SDC database.
This process generated a sample of 8,190 rm-year
observations for the years 2000–2008. Subsequent
triangulation produced a sample of 3,735,575
dyad-year observations with non-missing dyad in-
formation. The number of rm dyads for each year
wascalculatedasN×(N−1)/2, where N is the num-
ber of rms in each sample year. We summarize
the distribution of the number of rms, dyads and
alliances by sample years in Table 1.
Similar to prior research (Wang and Zajac,
2007), our study focuses on US public rms due to
data availability, such as information on IFMSTs
and rm nancials. However, this did limit the
number of identied sample alliances as well.
While limited due to data triangulation, the re-
sulting number of 452 alliances formed between
sample rm dyads in our study is in line with that
of alliances among public US rms reported in
previous studies. Based on one of the widely ac-
cepted sector categorization systems of Fama and
French (1997) that groups rms into 48 sectors,
366 (80.97%) out of the 452 alliances are identied
as cross-industry. This large prevalence of cross-
industry alliance formation conrms our choice
to conceptualize the network in a multi-industry
setting.
Measures
To the extent that alliance formation is a joint deci-
sion based on a bilateral search process where there
is ‘no obvious correct way to order the two rms in
a dyad’ (Wang and Zajac, 2007, p. 1304), a model
with dyadic-level rm characteristics is superior to
a one-sided focal rm approach. Therefore, our
analysis adopts the dyadic perspective in building
our models and constructing our variables.
Dependent variable
Alliance formation. Our dependent variable is di-
chotomous; coded as 1 if at least one alliance was
formed by the dyad in a given year and 0 other-
wise. Consistent with the literature, in this study
an alliance refers to non-equity/equity alliances
and inter-organizational collaborations that in-
clude marketing, R&D, manufacturing, licensing,
patent or logistics agreements (Rowley, Behrens
and Krackhardt, 2000).
Independent variables
Inter-rm managerial social ties. Following the
extant literature, we constructed the social tie
proxy based on the two individuals’ current and
© 2021 British Academy of Management and Wiley Periodicals LLC.
8K. Yang, J. J. Zhu and M. D. Santoro
Table 1. Number of rms, dyads and alliances across years
Year Number of rms Number of dyads Number of alliances
2000 774 299,151 42
2001 869 377,146 59
2002 906 409,965 49
2003 937 438,516 54
2004 950 450,775 53
2005 954 454,581 44
2006 945 446,040 50
2007 919 421,821 45
2008 936 437,580 56
Total 8,190 3,735,575 452
Notes: The number of rm dyads for each year was calculated as N×(N−1)/2, where N is the number of rms in each sample year (e.g.
299,151 rm dyads were formed among the 774 sample rms in year 2000: 774×(774−1)/2).
past overlap arising from a variety of social ex-
periences (Cai, Walkling and Yang, 2016; Cohen,
Frazzini and Malloy, 2008; Engelberg, Gao and
Parsons, 2012; Xue, Yang and Yao, 2018). To
examine the efcacy of alternative social mech-
anisms engendered by IFMSTs, we focus on two
types of interpersonal connections. (1) Activity-
based IFMSTs include social connections between
two individuals established through either over-
lapping employment experiences at a third-party
rm different from their current afliations by
working together on the top management team or
by serving on the same board (e.g. CEO and CFO
at Xerox or directors serving on IBM’s board).
This measure also includes common participation
in managing roles at other social activities, such as
serving on non-prot organization boards, univer-
sity board of trustees, charities or members at the
same social club. We acknowledge that individuals
holding memberships (rather than playing manag-
ing roles) at large organizations such as the New
York State Bar Association are unlikely to know
each other. Therefore, following Cai, Walkling and
Yang (2016), we excluded individuals whose role
is reported by BoardEx as ‘member’ or ‘unknown’
at the afliated organizations. (2) Education-based
IFMSTs are identied when two individuals are
connected through their educational background
(i.e. shared alma mater), and specically when
they attended the same university and graduated
with the same degree within 1 year of each other.
Taken together, IFMSTs are measured as the
number of distinct socially connected executive
and director pairs for each rm dyad in a given
year. We log-transformed this variable to correct
for right-skewness. Figure 2 provides a detailed
example of how we constructed the IFMSTs for a
sample dyad.
Structural position of dyadic rms – asymmet-
ric and peripheral dyad. We adopted a social
networks methodology for detecting the cen-
tral/peripheral structure of the network (Borgatti
and Everett, 2000). The ideal central/peripheral
structure for the network indicates that there is
a cohesive subgroup of central (core) rms in
which these central rms are well connected to
each other. We followed the procedures used in
prior studies and computed rms’ centrality scores
to capture the rm’s network position (e.g. Ahuja,
Polidoro and Mitchell, 2009; Pollock et al., 2015).
To do so, we rst constructed the network for
each year based on the alliance activities of S&P
1500 rms over the prior 4 years (results remain
consistent with 3- or 5-year rolling windows). We
then computed Bonacich’s beta centrality measure,
also known as Bonacich power centrality, with
UCINET 6 (Bonacich, 1987; Borgatti, Everett and
Freeman, 1999). This is a more generalized form of
eigenvector centrality used in the seminal paper by
Ahuja et al. (2009). The column centrality vector
is given by
c(α, β)=α(I−βR)−1Rl
where R is the adjacency matrix, with diagonal el-
ements zero. I is the identity matrix. l is a column
vector of all ones. αis a scaling vector, which is
set to normalize the score (to reect the number
of rms in the network in our case). A parame-
ter β(positive or negative) denes the degree of
the dependence of each rm’s score on the score
of other rms. This measure is more exible and
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 9
x1
Firm X Firm Y
x2
x3
y1
y2
y3
y4
y5
Activity-based IFMSTs
Education-based IFMSTs
Figure 2. Example of inter-rm managerial social ties (IFMSTs)
Notes: This gure presents an example of how we construct the
inter-rm managerial social ties between three executives at Firm
X (i.e. x1, x2 and x3) and ve executives at Firm Y (i.e. y1, y2,
y3, y4 and y5). The IFMSTs measure the number of distinct pairs
of socially (i.e. via education-based experience and/or activity-
based experience) connected executives aggregated to the rm-
dyad level. In this example, x1 and y2 share social ties through
overlaps attending the same college and overlaps in past employ-
ment at Firm Z, while x1 and y4 both serve on the same university
board of trustees. x3 and y2 overlapped through previous employ-
ment at Firm W. x3 and y5 shared social ties through overlaps at-
tending the same college. By denition, there are four unique pairs
of socially connected individuals for rm-dyad X–Y (i.e. IFMSTs
=4; x1 and y2, x1 and y4, x3 and y2, x3 and y5), among which
three pairs are connected through activity-based ties (x1 and y2,
x1 and y4, x3 and y2) and two pairs via education-based ties (x1
and y2, x3 and y5). Accordingly, activity-based IFMSTs =3 and
education-based IFMSTs =2.
[Colour gure can be viewed at wileyonlinelibrary.com]
generalizable than eigenvector centrality. In our
study context, Bonacich’s (1987) beta centrality
measure appropriately results in higher centrality
scores for rms that are allied with many rms and
higher scores if their alliance partners have entered
into more alliances with others.
To dene peripheral and central rms, we mean-
split the beta centrality score for our sample data
by year. For each year, rms that have a central-
ity score below the mean are dened as peripheral,
and central otherwise. Following prior literature
(Ajuha et al., 2009), we created a dummy variable
for two types of dyads involving peripheral rms
to test our hypotheses.
Peripheral dyad. At the rm dyad level, we fol-
low prior literature and identify a peripheral dyad
when both rms in the dyad have a centrality score
below the sample mean (Ajuha et al., 2009). The
dummy variable equals one for peripheral dyad,
and zero otherwise.
Asymmetric dyad. This is dened as a binary
variable that equals one if one of the rms in the
dyad has a centrality score lower than the sample
mean while the other has a centrality score equal
to or greater than the sample mean, and zero oth-
erwise (Ajuha et al., 2009).
Control variables
Since our tests of alliance formation consider a
number of key aspects by building on the dyadic
perspective of alliance formation, we constructed
rm-pair specic characteristics (interlock,busi-
ness similarity,general and partner-specic alliance
knowledge). We then converted other conventional
rm-level characteristics into dyadic-level rela-
tive measures (rm size,growth potential,ROA,
marketing resources) while keeping the maximum
value of those characteristics within the dyad as
an additional control (dyad max rm size,dyad
max growth potential,dyad max ROA,dyad max
marketing resources). Prior studies nd that inter-
locking directors play a role in alliance formation
(Gulati and Westphal, 1999). Thus, we include in-
terlock as a control variable, which we dene as the
log-transformed number of interlocking directors
shared by the rm dyad. Villalonga and McGa-
han (2005) argue that business similarity affects
alliance formation. Following their approach, we
dene business similarity as a dummy variable that
equals 1 if two rms belong to the same two-digit
SIC industry, and 0 otherwise. Following Wang
and Zajac (2007), we also control for relative size,
calculated as the absolute value of the total assets
difference between two rms scaled by the sum
of their total assets. Furthermore, we used two
measures to control the rm’s performance effect
on alliance formation: growth potential (proxied
by the market-to-book ratio) and return on assets
(ROA). Market-to-book ratio, as a value-based
measure, captures a rm’s long-term growth po-
tential (Barclay and Smith, 1995). For the measure
of growth potential, we took the logarithm of both
rms’ market-to-book ratios (M/B) and scaled the
© 2021 British Academy of Management and Wiley Periodicals LLC.
10 K. Yang, J. J. Zhu and M. D. Santoro
higher value with the lower value. Following Wang
and Zajac (2007), we measured the relative ROA
with the ratio of the absolute value of the differ-
ence between the two rms’ exponentiated ROAs
and the sum of these transformed ROAs. Villa-
longa and McGahan (2005) suggest that a rm’s
marketing resources determine inter-rm collabo-
ration. We calculated relative marketing resources
by dividing the absolute value of the difference of
both rms’ exponentiated advertisement-to-sales
ratios by the sum of these transformed ratios.
Prior studies show that geographic similarity also
inuences alliance formation (Ahuja et al., 2009;
Reuer and Lahiri, 2014). We constructed a relative
distance-based measure to capture the geographic
similarity between a dyad of rms in comparison
to their distances from other potential partners
in their respective pairing rms’ industry. Specif-
ically, we measured the distance in miles between
headquarter locations. For each rm in the dyad,
we then calculated its mean distance to the pairing
rm’s industry peers based on the two-digit SIC
code. Finally, we scaled the distance between the
dyad by the average of the two mean distances.
The accumulated alliance-related experience and
subsequent knowledge obtained by potential al-
liance partners can inuence a focal rm’s decision
in identifying future alliance partners (Hoang and
Rothaermel, 2010). For each sample year, the
general alliance knowledge is the total number
of alliances formed by the two rms in the dyad
with any other sample rms during the previous
5years.Thepartner-specic alliance knowledge
is the total number of alliances formed between
the focal dyad during the previous 5 years. Both
alliance knowledge variables were log-transformed
to correct for the right skewness. We also included
their quadratic terms to capture the potential non-
linear effects. Lastly, we included year dummies to
control for secular trends (Wang and Zajac, 2007).
Table 2 reports the descriptive statistics and
correlations for all variables. This shows each
rm-dyad has on average two pairs of executives
and directors sharing social ties. Following the
Fama and French (1997) sector categorization sys-
tem, Table 3 reports the sector prole information
along several key dimensions such as alliance ac-
tivities and average rm size, growth potential and
market resources. These statistics show signicant
variations in rm characteristics across sectors.
These sectors correspond well with other, often
studied industries such as chemicals, computers
and pharmaceutical products that are associated
with signicantly higher centrality measures, con-
rming the active alliance engagement by rms
in these sectors. Table 4(Panel A) presents the
characteristics of central versus peripheral rms
by sample years. Similar to Table 3, the sample
mean of rm characteristics is reported. Central
rms are on average larger and, not surprisingly,
more active in alliance activities than peripheral
rms. Table 4(Panel B) reports the mean value of
key variables for central dyads, asymmetric dyads
and peripheral dyads separately. Relative to cen-
tral dyads, peripheral dyads, on average, share less
socially connected executives and board members
and are less experienced with alliance activities.
Models and estimation methods
We built our model following Paruchuri, Goossen
and Phelps (2019) and construct variables from
a dyadic perspective following Rothaermel and
Boeker (2008). A logit model with maximum like-
lihood estimation was used for hypothesis testing
(Greene, 2003):
PIFMSTsij,t−1,Xij,t−1
=exp αIFMSTsij,t−1+Xij,t−1β+τt
1+exp αIFMSTsij,t−1+Xij,t−1β+τt
where yij,t is the alliance formation dummy for rm
i and j at time t, Xij,t−1is a vector of independent
variables and τtis the year xed effect. Moderating
effects are tested by adding interaction terms be-
tween peripheral dyad/asymmetric dyad and IFM-
STs. All independent variables were lagged by 1
year. To investigate how IFMSTs of different types
interact with structural embeddedness, we break
down IFMSTs into education-based and activity-
based IFMSTs and repeat the above analyses.
Results
Effects of IFMSTs on alliance formation
Table 5 presents our estimation results. Model 1
presents the baseline model and shows that consis-
tent with prior studies, the negative effect of poor
structural embeddedness is more salient when both
potential partners are at the periphery of the cur-
rent alliance network. The estimated coefcient
of IFMSTs is positive and signicant (α=1.057,
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 11
Table 2. Descriptive statistics and pairwise correlation matrix for the sample of rm dyads
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Alliance formation
(1)
1.000
IFMSTs (2) 0.021*** 1.000
IFMSTs – activity
based (3)
0.021*** 0.966*** 1.000
IFMSTs – education
based (4)
0.012*** 0.483*** 0.282*** 1.000
Peripheral dyad (5) −0.010*** −0.123*** −0.117*** −0.076*** 1.000
Asymmetric dyad
(6)
0.007*** 0.114*** 0.110*** 0.069*** −0.665*** 1.000
Interlock (7) 0.002*** 0.246*** 0.257*** 0.038*** −0.012*** 0.014*** 1.000
Business similarity
(8)
0.008*** 0.053*** 0.055*** 0.009*** −0.010*** 0.011*** 0.009*** 1.000
Max dyad size (9) 0.001*** 0.000 0.000 0.000 −0.057*** 0.053*** 0.000 0.000 1.000
Relative size (10) 0.000 0.045*** 0.040*** 0.027*** −0.055*** 0.068*** −0.010*** −0.032*** 0.000 1.000
Max dyad growth
potential (11)
0.007*** −0.033*** −0.037*** 0.003*** −0.080*** 0.121*** −0.004*** 0.013*** −0.181*** 0.005*** 1.000
Relative growth
potential (12)
0.000 0.022*** 0.021*** 0.014*** 0.002*** 0.006*** 0.002*** −0.015*** 0.000 0.053*** 0.098***
Max dyad ROA (13) 0.005*** −0.037*** −0.038*** −0.007*** −0.049*** 0.040*** −0.004*** −0.014*** −0.114*** −0.013*** 0.619***
Relative ROA (14) 0.002*** −0.074*** −0.072*** −0.034*** −0.010*** 0.040*** −0.010*** 0.007*** 0.000 0.041*** 0.284***
Max dyad
marketing
resource (15)
0.001 0.051*** 0.049*** 0.033****** −0.038*** 0.044*** 0.006*** −0.029*** 0.005*** 0.028*** 0.103***
Relative marketing
resource (16)
0.000 0.024*** 0.024*** 0.012*** −0.021*** 0.028*** 0.001*** −0.015*** 0.085*** 0.034*** 0.072***
Geographic
similarity (17)
−0.002*** −0.124*** −0.123*** −0.057*** −0.003****** 0.004*** −0.045*** −0.009*** 0.109*** −0.012*** 0.035***
General alliance
knowledge (18)
0.011*** 0.126*** 0.121*** 0.075*** −0.188*** 0.231*** 0.013*** 0.045*** 0.006*** 0.048*** 0.074***
General alliance
knowledge square
(19)
0.010*** 0.109*** 0.105*** 0.063*** −0.161*** 0.204*** 0.011*** 0.049*** 0.001*0.039*** 0.066***
Partner-specic
alliance
knowledge (20)
0.045*** 0.037*** 0.037*** 0.020*** −0.017*** 0.002*** 0.002*** 0.012*** 0.125*** −0.002*** 0.006***
© 2021 British Academy of Management and Wiley Periodicals LLC.
12 K. Yang, J. J. Zhu and M. D. Santoro
Table 2. (Continued)
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
Partner-specic
alliance
knowledge square
(21)
0.045*** 0.036*** 0.036*** 0.020*** −0.017*** 0.002*** 0.002*** 0.011*** 0.114*** −0.002*** 0.006***
Mean 0.000 0.663 0.582 0.141 0.731 0.140 0.003 0.050 2.099 0.608 0.831
Min 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 1.148 0.000 −0.705
Max 1.000 6.987 6.984 2.708 1.000 1.000 1.792 1.000 6.934 1.000 4.364
St. dev. 0.011 0.816 0.795 0.323 0.443 0.347 0.050 0.218 1.117 0.299 0.511
(12) (13) (14) (15) (16) (17) (18) (19) (20) (21)
Alliance formation
(1)
IFMSTs (2)
IFMSTs – activity
based (3)
IFMSTs – education
based (4)
Peripheral dyad (5)
Asymmetric dyad
(6)
Interlock (7)
Business similarity
(8)
Maxdyadsize(9)
Relative size (10)
Max dyad growth
potential (11)
Relative growth
potential (12)
1.000
Max dyad ROA (13) 0.042*** 1.000
Relative ROA (14) 0.047*** 0.314*** 1.000
Max dyad
marketing
resource (15)
−0.001 0.053*** 0.134*** 1.000
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 13
Table 2. (Continued)
(12) (13) (14) (15) (16) (17) (18) (19) (20) (21)
Relative marketing
resource (16)
0.000 0.012*** 0.219*** 0.801*** 1.000
Geographic
similarity (17)
0.006*** 0.016*** 0.019*** 0.022*** 0.012*** 1.000
General alliance
knowledge (18)
−0.001 −0.017*** 0.064*** 0.031*** 0.020*** 0.008*** 1.000
General alliance
knowledge square
(19)
−0.003*** −0.020*** 0.066*** 0.021*** 0.015*** 0.005*** 0.974*** 1.000
Partner-specic
alliance
knowledge (20)
−0.001*0.003*** −0.001 0.002*** 0.001*−0.004*** 0.017*** 0.016*** 1.000
Partner-specic
alliance
knowledge square
(21)
−0.001*0.003*** 0.001 0.002*** 0.001*−0.004*** 0.017*** 0.015*** 0.982*** 1.000
Mean 4.086 0.092 0.041 0.079 0.009 0.620 0.462 1.678 0.001 0.000
Min −60.716 −1.146 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000
Max 88.270 0.619 0.994 1.000 0.580 1.528 6.209 38.547 1.386 1.922
St. dev. 14.822 0.063 0.052 0.062 0.029 0.283 1.210 4.949 0.013 0.010
†0.00012.
Notes: The sample of r m dyads consists of 3,735,575 rm dyad-year observations over the sample period of 9 years from 2000 to 2008. Alliance dummy equals 1 if a rm dyad announces
at least one alliance formation in a given year, and 0 otherwise. Inter-rm managerial social ties are the logarithm of number of distinct socially connected executives and director pairs
for each rm dyad in a given year.
The statistical signicance is denoted as *** p<0.01, ** p<0.05 and *p<0.1 for two-tailed tests.
© 2021 British Academy of Management and Wiley Periodicals LLC.
14 K. Yang, J. J. Zhu and M. D. Santoro
Table 3. Sector prole
Industry No. rms
Growth
potential Size (asset) ROA
Marketing
resources
General alliance
knowledge
Specic alliance
knowledge
Bonacich power
centrality
Aircraft 67 0.7447 7.6710 0.0419 0.0051 3.3806 0.1791 1.5425
Agriculture 31 2.1092 7.5896 0.0493 0.0111 6.3954 0.1722 7.9214
Automobiles and Trucks 219 1.7380 6.8247 0.0416 0.0135 2.4110 0.0639 0.8636
Banking 282 2.7843 9.1095 0.0297 0.0072 0.7110 0.0337 0.7369
Beer & Liquor 27 −0.5604 7.9638 0.0603 0.0325 8.9630 0.1852 0.0684
Construction Materials 92 2.7681 8.1672 0.0543 0.0073 3.2391 0.2174 12.0298
Printing and Publishing 200 1.8027 7.5465 0.0500 0.0129 3.0625 0.1550 10.1609
Shipping Containers 19 0.1443 8.5379 0.0583 0.0124 −1.8421 0.1316 0.0461
Business Services 648 3.1229 7.6278 0.0497 0.0088 3.4660 0.1103 4.1365
Chemicals 280 3.1983 7.7463 0.0536 0.0139 4.5179 0.2089 10.3652
Electronic Equipment 703 2.9217 7.8533 0.0548 0.0068 3.4805 0.1531 11.0407
Apparel 86 1.9121 7.8716 0.0596 0.0072 1.1337 0.1221 0.2608
Construction 102 1.5086 7.8891 0.0461 0.0075 7.7745 0.1961 25.1232
Coal 22 2.7934 7.2347 0.0393 0.0063 0.1136 0.1364 0.1390
Computers 540 3.4290 10.0335 0.0611 0.0069 21.8387 0.3548 62.2926
Pharmaceutical Products 445 1.7736 7.7240 0.0648 0.0127 4.0600 0.1100 23.6969
Electrical Equipment 92 3.3636 7.4285 0.0439 0.0125 7.2500 0.1087 2.4159
Fabricated Products 14 3.7041 6.8836 0.0610 0.0045 4.2500 0.3214 7.8162
Trading 316 3.4741 7.7388 0.0515 0.0087 3.7247 0.1535 5.6251
Food Products 209 1.4170 7.5534 0.0400 0.0064 2.3110 0.0837 1.8380
Entertainment 20 1.4160 5.9335 0.0573 0.0100 −0.2250 0.1500 23.6576
Precious Metals 2 0.2347 12.6374 0.0133 0.0053 0.0000 0.0000 0.0000
Defense 17 1.3769 6.1529 0.0667 0.0011 2.7647 0.0294 0.5909
Healthcare 149 2.7719 7.9006 0.0451 0.0088 4.7819 0.1309 4.9941
Consumer Goods 133 3.0904 7.1805 0.0565 0.0126 6.8045 0.1541 12.1864
Insurance 307 2.6603 7.3491 0.0496 0.0090 4.2248 0.1451 5.5181
Measuring and Control
Equipment
163 2.1006 7.0978 0.0452 0.0085 4.5920 0.1288 1.5070
Machinery 323 2.4266 7.5106 0.0473 0.0088 2.5031 0.1176 3.6635
Restaurants, Hotels, Motels 164 3.1895 7.6017 0.0560 0.0175 3.2805 0.1463 8.7722
Medical Equipment 184 1.4742 7.1235 0.0545 0.0079 8.3859 0.1685 8.1131
Non-Metallic and Industrial
Metal Mining
7 1.2555 7.6073 0.0694 0.0113 23.3571 0.2143 0.1167
Petroleum and Natural Gas 331 3.2853 7.9483 0.0498 0.0097 3.8399 0.0861 6.2250
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 15
Table 3. (Continued)
Industry No. rms
Growth
potential Size (asset) ROA
Marketing
resources
General alliance
knowledge
Specic alliance
knowledge
Bonacich power
centrality
Others 26 1.5733 6.2467 0.0352 0.0095 3.7343 0.1779 0.9856
Business Supplies 99 3.0656 8.3695 0.0443 0.0043 3.7929 0.1414 3.9988
Personal Services 66 −0.1978 7.5264 0.0528 0.0050 0.4697 0.0682 0.0590
Real Estate 24 0.1347 6.4982 0.0585 0.0062 1.6458 0.1250 0.0272
Retail 431 1.9299 7.6953 0.0482 0.0081 3.4130 0.1531 9.4354
Rubber and Plastic Products 27 0.8092 6.3365 0.0337 0.0043 0.0926 0.1667 1.7983
Shipbuilding, Railroad
Equipment
10 1.4676 6.3792 0.0555 0.0021 0.0000 0.0500 0.0269
Tobacco Products 9 1.8190 8.7276 0.0525 0.0049 −0.2222 0.2778 0.4547
Candy & Soda 50 2.1514 7.8495 0.0506 0.0080 4.4011 0.1360 8.1616
Steel Works 106 2.8513 7.5434 0.0443 0.0096 3.4292 0.0991 6.0514
Communication 143 2.7769 8.8103 0.0418 0.0065 8.6748 0.1538 6.6407
Recreation 74 1.1649 7.2606 0.0541 0.0153 1.7365 0.1014 9.0609
Transportation 132 1.2980 7.8983 0.0462 0.0136 9.6818 0.2008 9.9327
Textiles 23 1.7139 7.1647 0.0542 0.0080 0.5000 0.3696 23.1792
Utilities 542 2.0291 7.4936 0.0439 0.0080 3.1116 0.1356 5.7110
Wholesale 234 1.9967 7.4718 0.0525 0.0092 6.5940 0.1389 12.1152
All 8,190 2.3834 7.6685 0.0494 0.0095 4.0640 0.1358 7.7033
†Since 366 out of the 452 alliances included in our sample are cross-industry, we count these alliances in the respective sectors of both partners, which leads to the total reported number
of alliances in this table: 366+366+(452−366)=818. To keep the table at a reasonable length for reporting purposes, we report the summary statistics according to one of the widely
accepted yet more aggregated sector categorization systems (Fama and French, 1997).
© 2021 British Academy of Management and Wiley Periodicals LLC.
16 K. Yang, J. J. Zhu and M. D. Santoro
Table 4. Summary statistics of rm (rm-dyad) prole information by centrality
Panel A: Firm prole
Year No. rms
Growth
potential Size (asset) ROA
Marketing
resources
General
alliance
knowledge
Partner-
specic
alliance
knowledge
Bonacich
power
centrality
Central rm
2000 246 3.3452 8.1641 0.0585 0.0125 6.6171 0.2537 25.0844
2001 278 2.2733 8.5538 0.0565 0.0106 6.2416 0.3084 19.6609
2002 282 2.3861 8.7726 0.0508 0.0085 7.9502 0.3299 21.0948
2003 286 2.3255 8.9304 0.0489 0.0088 5.9451 0.3092 24.7825
2004 286 2.2027 8.7672 0.0489 0.0090 6.5973 0.2572 16.9784
2005 295 2.9625 9.0637 0.0452 0.0115 6.8304 0.2492 14.0072
2006 293 2.0600 9.0659 0.0492 0.0100 4.4563 0.2126 16.3657
2007 275 1.8327 9.3357 0.0496 0.0105 3.9799 0.1784 12.7700
2008 355 2.0948 9.4073 0.0440 0.0104 4.0826 0.2110 56.2811
Overall 2,596 2.3683 8.9209 0.0498 0.0102 5.7999 0.2553 23.8627
Peripheral rm
2000 528 3.3939 6.9389 0.0495 0.0077 3.3281 0.0592 0.0236
2001 591 2.5260 7.0259 0.0508 0.0088 3.7905 0.0730 0.0208
2002 624 2.5101 7.1056 0.0493 0.0089 2.9563 0.0683 0.0234
2003 651 1.6954 7.0577 0.0484 0.0097 3.2714 0.0885 0.0287
2004 664 2.0161 7.1726 0.0487 0.0090 3.3681 0.0933 0.0246
2005 659 2.5065 7.1239 0.0485 0.0100 3.5880 0.0755 0.0247
2006 652 2.1092 7.2090 0.0501 0.0100 3.7843 0.0871 0.0254
2007 644 1.9992 7.1068 0.0475 0.0090 2.1400 0.0925 0.0228
2008 581 3.0363 7.0010 0.0510 0.0086 3.1139 0.0804 0.0256
Overall 5,594 2.3904 7.0873 0.0493 0.0091 3.2584 0.0804 0.0244
Panel B: Dyad prole
Central dyad Asymmetric dyad Peripheral dyad
Alliance formation likelihood (%) 0.0375 0.0135 0.0053
IFMSTs 0.7614 0.7028 0.6049
IFMSTs – activity based 0.6722 0.6193 0.5280
IFMSTs – education based 0.1661 0.1504 0.1269
Interlock 0.0033 0.0033 0.0027
Business similarity 0.0514 0.0511 0.0487
Relative size 0.6099 0.6182 0.5982
Relative growth potential 3.7417 4.1625 4.0895
Relative ROA 0.0371 0.0424 0.0406
Relative marketing resource 0.0089 0.0094 0.0086
Geographic similarity 0.6209 0.6199 0.6195
General alliance knowledge 0.6520 0.4960 0.3248
Partner-specic alliance knowledge 0.0009 0.0002 0.0001
Number of alliances 141 219 92
Number of observations 376,422 1,615,040 1,744,113
se =0.043, p <0.01), indicating the positive main
effect of IFMSTs on alliance formation.
Model 2 presents our results of how IFMSTs
interplay with the alliance network structure
in moderating the negative effects of inferior
structural embeddedness (H1a and H1b). The
estimated coefcients of interaction terms, IFM-
STs ×Peripheral dyad (β=0.238, se =0.095,
p=0.01) and IFMSTs ×Asymmetric dyad (β
=0.158, se =0.092, p =0.09), capture IFMSTs’
moderating effects on peripheral and asymmetric
dyads in comparison to central dyads. Both are
signicantly positive. For a non-linear model like
Logit regression, it is inappropriate to use the
estimated coefcient directly to make inferences
on the true relationship between the independent
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 17
Table 5. IFMSTs, structural embeddedness and the likelihood of alliance formation
Aggregated IFMSTs Activity vs. education-based IFMSTs Hypothesis
Model 1 Model 2 Model 3 Model 4
IFMSTs’ moderation on the effects of structural embeddedness
IFMSTs ×Peripheral dyad 0.238** H1a, H1b
(0.095)
IFMSTs ×Asymmetric
dyad
0.158*
(0.092)
IFMSTs – Activity based
×Peripheral dyad
0.280*** H2
(0.102)
IFMSTs – Education
based ×Peripheral dyad
−0.146
(0.242)
IFMSTs – Activity based
×Asymmetric dyad
0.190*
(0.099)
IFMSTs – Education
based ×Asymmetric
dyad
−0.136
(0.228)
Main effect of IFMSTs
IFMSTs 1.057*** 0.918***
(0.043) (0.072)
IFMSTs – activity based 0.938*** 0.773***
(0.045) (0.077)
IFMSTs – education based 0.491*** 0.589***
(0.097) (0.164)
Controls
Peripheral dyad −1.969*** −2.526*** −1.948*** −2.492***
(0.168) (0.272) (0.168) (0.262)
Asymmetric dyad −1.049*** −1.478*** −1.028*** −1.436***
(0.144) (0.267) (0.145) (0.256)
Interlock −2.503*** −2.497*** −2.347*** −2.336***
(0.503) (0.502) (0.505) (0.504)
Business similarity 0.975*** 0.971*** 0.983*** 0.980***
(0.121) (0.121) (0.122) (0.121)
Dyad maximum size 0.233*** 0.223*** 0.233*** 0.223***
(0.043) (0.043) (0.043) (0.043)
Relative size −0.100 −0.080 −0.131 −0.111
(0.155) (0.155) (0.155) (0.155)
Dyad maximum growth
potential
0.729*** 0.723*** 0.725*** 0.716***
(0.111) (0.110) (0.111) (0.110)
Relative growth potential −0.007** −0.007** −0.007** −0.007**
(0.003) (0.004) (0.004) (0.003)
Dyad maximum ROA 1.919** 1.950** 1.901** 1.935**
(0.840) (0.836) (0.841) (0.836)
Relative ROA −0.105 −0.089 −0.055 −0.033
(0.995) (0.993) (0.990) (0.987)
Dyad maximum marketing
resource
−4.248*** −4.275*** −4.392*** −4.439***
(1.262) (1.259) (1.279) (1.274)
Relative marketing
resource
4.277 4.356 4.320 4.417
(3.393) (3.374) (3.509) (3.477)
© 2021 British Academy of Management and Wiley Periodicals LLC.
18 K. Yang, J. J. Zhu and M. D. Santoro
Table 5. (Continued)
Aggregated IFMSTs Activity vs. education-based IFMSTs Hypothesis
Model 1 Model 2 Model 3 Model 4
Geographic similarity 0.043 0.037 0.058 0.053
(0.153) (0.153) (0.154) (0.153)
Partner-specic alliance
knowledge
4.851*** 4.934*** 4.829*** 4.956***
(1.149) (1.138) (1.140) (1.122)
Partner-specic alliance
knowledge2
−2.143 −2.305*−2.305*−2.384*
(1.449) (1.398) (1.398) (1.376)
General alliance
knowledge
0.464*** 0.459*** 0.455*** 0.448***
(0.118) (0.118) (0.118) (0.118)
General alliance
knowledge2
−0.080*** −0.065** −0.063*** −0.062**
(0.027) (0.028) (0.027) (0.028)
Constant −11.544*** −11.278*** −11.429*** −11.174***
(0.255) (0.272) (0.254) (0.267)
Year dummy Yes Yes Yes Yes Yes
N 3,735,575 3,735,575 3,735,575 3,735,575
−2LogLikelihood 7,331 7,325 7,330 7,322
AIC 7,385 7,383 7,386 7,386
Pseudo Adj-R20.191 0.192 0.191 0.192
Notes: Standard errors and p-values for two-tailed tests are reported in parentheses.
The statistical signicance is denoted as *** p<0.01, ** p<0.05 and *p<0.1 for two-tailed tests.
and dependent variables (Ai and Norton, 2003).
Therefore, we followed Wiersema and Bowen
(2009) to compute the true interaction and key
marginal effects. The true interaction is 0.002 (se
=0.001, p <0.01, ranging from 0.001 to 0.012)
for IFMSTs ×Peripheral dyads and 0.001 (se =
0.000, p <0.01, ranging from 0.001 to 0.015) for
IFMSTs ×Asymmetric dyads. Holding other vari-
ables at their means, the marginal negative effect
of peripheral dyads is reduced from −0.005 (se =
0.001, p <0.01) to −0.002 (se =0.001, p <0.01),
and the marginal negative effect of asymmetric
dyads reduced from −0.003 (se =0.001, p <0.01)
to −0.001 (se =0.000, p <0.01) when estimated at
low versus high values of IFMSTs. Both modera-
tion effects are validated. Thus, H1a is supported.
We then compared the coefcient of the interac-
tion term between IFMSTs and peripheral dyads
with the coefcient of the interaction term between
IFMSTs and asymmetric dyads. The chi-square
test shows that the former is statistically greater
than the latter (chi-sqdf =1=3.816, p =0.05), pro-
viding support for H1b.
Model 3 presents the baseline model differen-
tiating activity-based IFMSTs (α=0.938, se =
0.045, p <0.01) from education-based IFMSTs (α
=0.491, se =0.097, p <0.01). The chi-square test
shows that the coefcient for activity-based IFM-
STs is statistically greater than the coefcient for
education-based IFMSTs (chi-sqdf =1=13.532,
p<0.01), conrming the superior efcacy of
activity-based IFMSTs in facilitating alliance for-
mation in comparison to education-based IFM-
STs.
Model 4 reports the results of the respective
interplay of activity-based and education-based
IFMSTs with the indictors of poorly embedded
dyads. The interaction terms, Activity-based IFM-
STs ×Peripheral dyads (β=0.280, se =0.102, p
<0.01) and Activity-based IFMSTs ×Asymmet-
ric dyads (β=0.190, se =0.099, p =0.06), are
both positive as expected. The true interaction is
0.002 (se =0.001, p <0.01, ranging from 0.001
to 0.013) for Activity-based IFMSTs ×Peripheral
dyads and 0.001 (se =0.000, p <0.01, ranging
from 0.001 to 0.010) for Activity-based IFMSTs
×Asymmetric dyads. Holding other variables at
their means, the marginal negative effect of periph-
eral dyads reduces from −0.005 (se =0.001, p <
0.01) to −0.003 (se =0.001, p <0.01), and the
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 19
marginal negative effect of asymmetric dyads re-
duces from −0.003 (se =0.001, p <0.01) to −0.002
(se =0.000, p <0.01) when estimated at low ver-
sus high values of IFMSTs. Both moderation ef-
fects are validated. However, the interaction terms,
Education-based IFMSTs ×Peripheral dyads (β
=−0.146, se =0.242, p =0.55) and Education-
based IFMSTs ×Asymmetric dyads (β=−0.136,
se =0.228, p =0.55) are both statistically in-
signicant. Taken together, activity-based IFMSTs
are effective in assuaging the negative inuence
of structural embeddedness on alliance formation
for both peripheral and asymmetric dyads, while
education-based IFMSTs are not. These results
support H2.
Robustness tests and additional studies
We ran a number of additional tests to check the
robustness of our ndings. First, we adopted a
two-stage regression analysis to address potential
endogeneity issues (Petrin and Train, 2010). It’s
possible that a rm appoints executives who are
connected to the potential partnering rm in antic-
ipation of the alliance formation, in which case the
observed association between the IFMSTs and the
likelihood of alliance formation can be spurious.
We identied the number of rms’ executives and
directors as a legitimate instrument variable, be-
cause the size of top management teams is likely
to be positively associated with the number of
individual-level social ties between the two rms
but unlikely to be directly associated with the like-
lihood of alliance formation between them. The
rst column of Table 6 reports the estimates of
our key variables, which are consistent and robust.
Second, we used alternative ways to measure the
central position of a rm (Borgatti and Everett,
2000) and to classify central versus peripheral
rms. First, we median-split Bonacich’s beta
centrality measure to identify central versus pe-
ripheral rms. Second, we re-ran our analysis
with the degree centrality measure, which is nu-
merically computed as the number of distinct
alliance partners that the focal rm has in the
alliance network. Third, we adopted the discrete
central/peripheral model in UCINET 6 to directly
output the algorithm’s identication of central
and peripheral rms. Fourth, we then applied the
central/peripheral algorithm in the social network
analysis tool-UCINET 6 to generate a continuous
centrality score to detect the central/peripheral
structure of the network (Borgatti and Everett,
2000). The estimates for key variables constructed
using these alternative approaches are reported
in columns 2–5 of Table 6, respectively, and are
consistent with the results from our proposed
model.
Third, it is plausible that the breadth of our
activity-based IFMSTs blurs the roles of different
types of social activities at play (e.g. professional
vs. leisure activities). To address this potential
concern, we constructed two alternative activity-
based IFMSTs measures. We rst limited the ties
to those originating from professional activities,
such as working on the same top management
team or serving on the same corporate board. Sec-
ond, we limited the ties to those cultivated from
other non-professional activities, such as serving
on non-prot organization boards of trustees or
playing at the same golf club. We then repeated
our analyses and report the estimation results
in columns 6 and 7 of Table 6, respectively. The
alternative measures demonstrate similar effects
on attenuating the negative effects of inferior
structural embeddedness for peripheral rms in
alliance formation, which further validates our
composite measure of the activity-based IFMSTs.
Fourth, studies using rm dyadic structure typ-
ically deal with large sample sizes but a relatively
small number of ‘event rm dyads’. We addressed
the ‘rare event’ issue by rst adopting a rare event
logistic regression for our analysis (King and Zeng,
2001). Next, we ran a number of tests based on a
reduced sample by repeating our main tests using
randomly selected subsamples of 10 to 1,000 times
as many non-event dyads in the analysis (see Wang
and Zajac, 2007). The estimates for key study vari-
ables of these alternative approaches are reported
in columns 1–6 of Table 7, respectively, and the re-
sults remain consistent and robust.
Lastly, we conducted split-sample analyses to
validate the revealed interaction effects in our
model (Gelman and Park, 2009). We separated our
sample into three subgroups based on a dyad’s
relative structural embeddedness: central dyads,
asymmetric dyads and peripheral dyads. We then
re-ran our main models and compared the coef-
cients of IFMSTs across the subsamples to vali-
date the interaction effect of IFMSTs. The results
remain robust. For brevity we do not include the
results here, but they are available upon request.
© 2021 British Academy of Management and Wiley Periodicals LLC.
20 K. Yang, J. J. Zhu and M. D. Santoro
Table 6. Robustness check: endogeneity control and alternative approaches for structural identication and IFMST measures
Robustness check (1) (2) (3) (4) (5) (6) (7)
Alternative approaches for peripheral/central rm identication
Alternative measure of activity-based
IFMSTs
Model with
endogeneity
control
Model using median
split Bonacich’s beta
centrality
Model using
degree
centrality
Model using
central
peripheral
algorithm
Model using
centralness score
Employment-related
social activities
Non-professional
social activities
IFMST (aggregate measure)
IFMSTs’ moderation on the effects of structural embeddedness
IFMSTs ×Peripheral
dyad
0.285*** 0.314*** 0.266*** 0.364*** 0.224**
(0.097) (0.108) (0.093) (0.105) (0.109)
IFMSTs ×Asymmetric
dyad
0.211** 0.199*0.166** 0.236** 0.185
(0.094) (0.110) (0.092) (0.107) (0.094)
Control variables and year
xed effects included
Ye s Ye s Ye s Ye s Ye s
Activity-based IFMST and education-based IFMST
IFMSTs’ moderation on the effects of structural embeddedness
IFMSTs – Activity based
×Peripheral dyad
0.281*** 0.314*** 0.233*** 0.313*** 0.274*** 0.374*** 0.362***
(0.111) (0.113) (0.099) (0.111) (0.103) (0.097) (0.119)
IFMSTs – Education
based ×Peripheral
dyad
0.068 0.003 0.148 0.200 −0.064 −0.065 −0.177
(0.296) (0.244) (0.220) (0.241) (0.236) (0.237) (0.246)
IFMSTs – Activity based
×Asymmetric dyad
0.201*0.246** 0.184*0.214** 0.233** 0.235*0.236*
(0.109) (0.116) (0.098) (0.113) (0.101) (0.090) (0.116)
IFMSTs – Education
based ×Asymmetric
dyad
0.111 −0.265 −0.169 0.031 −0.205 −0.081 −0.171
(0.288) (0.252) (0.217) (0.246) (0.233) (0.223) (0.231)
Control variables and year
xed effects included
Ye s Ye s Ye s Ye s Ye s Yes Ye s
Notes: Standard errors and p-values for two-tailed tests are reported in parentheses.
The statistical signicance is denoted as *** p<0.01, ** p<0.05 and *p<0.1 for two-tailed tests.
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 21
Table 7. Robustness check: additional logistic regression analyses of rare events
(1) (2) (3) (4) (5) (6)
Robustness check
Rare event
logit
Random
sample with
10×
non-events
Random
sample with
50×
non-events
Random
sample with
100×
non-events
Random
sample with
500×
non-events
Random
sample with
1,000×
non-events
IFMST (aggregate measure)
IFMSTs’ moderation on the effects of structural embeddedness
IFMSTs ×Peripheral
dyad
0.305*** 0.326** 0.317** 0.262** 0.207** 0.213**
(0.095) (0.148) (0.139) (0.110) (0.102) (0.099)
IFMSTs ×
Asymmetric dyad
0.163*0.267*0.237** 0.183*0.164*0.158*
(0.091) (0.140) (0.117) (0.112) (0.095) (0.096)
Control variables and
year xed effects
included
Ye s Ye s Ye s Ye s Ye s Ye s
Activity-based IFMST and education-based IFMST
IFMSTs’ moderation on the effects of structural embeddedness
IFMSTs – Activity
based ×Peripheral
dyad
0.317*** 0.338** 0.332*** 0.260*** 0.235** 0.260**
(0.105) (0.152) (0.128) (0.120) (0.109) (0.106)
IFMSTs – Education
based ×Peripheral
dyad
−0.004 0.224 0.050 −0.377 −0.053 0.109
(0.262) (0.382) (0.307) (0.284) (0.254) (0.252)
IFMSTs – Activity
based ×
Asymmetric dyad
0.178*0.256*0.275** 0.219** 0.183** 0.166*
(0.100) (0.153) (0.128) (0.120) (0.107) (0.101)
IFMSTs – education
based ×
Asymmetric dyad
−0.059 −0.025 −0.163 −0.372 −0.096 −0.215
(0.247) (0.372) (0.301) (0.280) (0.243) (0.237)
Control variables and
year xed effects
included
Ye s Ye s Ye s Ye s Ye s Ye s
Notes: Standard errors and p-values for two-tailed tests are reported in parentheses.
The statistical signicance is denoted as *** p<0.01, ** p<0.05 and *p<0.1 for two-tailed tests.
Discussion and conclusions
Our study explores alternative social mechanisms
that help poorly embedded rms overcome the
structural constraints in forming alliances by
examining the interactions between relational
embeddedness and structural embeddedness.
Our study illuminates three key ndings. First,
the negative effect of inferior structural embed-
dedness that hampers alliance participation by
poorly embedded rms is signicantly reduced
when the potential alliance partners are connected
through IFMSTs. Second, this moderating effect
is stronger when both potential partners are at the
periphery of the existing alliance network. Third,
activity-based IFMSTs further reduce the negative
effect of poor structural embeddedness in com-
parison to ties originated from shared educational
afliations.
Theoretical implications
This study adds to the nascent research stream to
better understand the alliance activity of poorly
embedded rms and thereby offers important
theoretical implications on rm network dynam-
ics (Ahuja, Soda and Zaheer, 2012; Ahuja et al.,
2009; Chakravarty, Zhou and Sharma, 2020).
© 2021 British Academy of Management and Wiley Periodicals LLC.
22 K. Yang, J. J. Zhu and M. D. Santoro
Recent studies demonstrate that by accepting
unfavourable terms of trade, poorly embedded
rms can increase their initial attraction to highly
embedded alliance partners, though at the expense
of limiting their own future expansion potentials
(Ahuja et al., 2009). Our ndings generate further
implications for the viability of a path-creation
potential for poorly embedded rms by gradually
changing the network dynamics through allying
with not only centrally embedded alliance part-
ners, but also other poorly embedded rms. One
key insight is that when originated from social
experiences different from the transactional inter-
organizational activities that dene the structure
of the existing alliance network, relational em-
beddedness can alleviate the path-dependence
constraints hindering peripheral rms from form-
ing alliances. Therefore, our study is the rst to
shed light on how, based on IFMSTs, the infor-
mational and sociological benets derived from
relational embeddedness moderate the negative
effect of structural barriers of poorly embedded
rms when it comes to alliance formation.
Second, our study contributes to the burgeon-
ing literature that largely focuses on the direct
impact of IFMSTs on rms’ behaviour in vari-
ous inter-organizational business settings (Cooney
et al., 2015; Ishii and Xuan, 2014). In particu-
lar, our study conceptualizes the granularity in
the strength of the informational and sociologi-
cal benets channelled through interpersonal so-
cial ties based on different social experiences, in-
cluding both activity-based and education-based
IFMSTs (Koka and Prescott, 2002; Uzzi and
Lancaster, 2004). Our ndings also reveal the im-
portance of compatibility between the benets
provided by different underlying social mecha-
nisms and the needs of rms in various structural
embeddedness conditions seeking alliance oppor-
tunities. Failing to meet the needs for informa-
tional and reputational capital vital to overcom-
ing structural barriers, the same factors that limit
the strength of structural homophily in predicting
the alliance behaviour of poorly embedded rms
(Ahuja et al., 2009) may also explain the muted
moderating effect of other homophilic relations,
such as those derived from shared educational af-
liations.
Third, peripheral rms are often seen as the
weaker party when it comes to partnering with
more central rms not only from their perceived
or real informational and reputational decien-
cies, but also due to power differences resulting
from network asymmetry (Hu, Jain and Delios,
2021). Hu, Jain and Delios (2021) added to the
dearth of literature on peripheral rms by suggest-
ing that peripheral rms could gain more durable
alliances by partnering with a rm that has mod-
erately higher centrality, which in turn would help
them expand their network and progress to more
central positions in the network. We further add to
this literature by suggesting that IFMSTs may not
only allow peripheral rms to partner with more
central rms in the network by mitigating informa-
tional and reputational deciencies, but by perhaps
levelling the eld when it comes to power asymme-
try due to network positioning.
Managerial implications
This research offers managerial implications that
have strategic signicance. First, some of these
peripheral rms may have pre-competitive or
basic knowledge that must be combined with the
resources and competencies of other peripheral
or central rms via alliances in order to commer-
cialize that knowledge into technologies that can
be put into immediate use (Santoro, 2013). Given
that poorly embedded rms participated in over
68% of all alliances captured in our sample, our
study draws managerial attention to the strategic
and perhaps economic signicance of fostering
interpersonal relationships among these rms in
order to facilitate alliance formation. Second,
our ndings on the boundary of the IFMSTs’
inuence through different types of ties provide
possible ways for top managers, especially those
managing poorly embedded rms, to utilize the
social capital derived from personal networks.
In particular, our ndings highlight the varying
efcacy of personal ties originated from different
social experiences in facilitating alliance forma-
tion. In doing so, our ndings do not invalidate
or preclude the potential sociological benets of
educational homophily, but rather highlight the
stronger salience of the moderating effects of
activity-based interpersonal social ties. Therefore,
rms, especially poorly embedded rms, should
recognize and perhaps try to exploit the strategic
importance of their top managers’ professional
and social networking activities when seeking
alliance formation opportunities. From a broader
perspective, our study contributes to a growing
literature in providing managerial implications
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 23
on the value and relevancy of a rm’s key exec-
utives and directors’ social networks in shaping
corporate policy decisions. For example, a study
by Engelberg, Gao and Parsons (2013) documents
that an average pay-for-external-connectivity pre-
mium of $17,000 is awarded to the CEO for each
additional connection the CEO has with a coun-
terpart at another rm. These ndings suggest
that IFMSTs could affect other strategic actions
beyond the formation of strategic alliances, which
would appear to be rich avenues for future research
inquiry as we discuss in the next section.
Limitations and future research
We believe our ndings could promote research
questions that merit further investigation. First, to
the extent that data availability presents an almost
ubiquitous challenge to empirical studies in the
eld of inter-organizational activities, it would be
useful to focus data collection efforts along two
additional dimensions for future investigation: (1)
more ne-grained collaboration-level information
to recognize the varying degrees of executives’
involvement in the inter-organizational activities
of interest (Bierly III and Galagher, 2007) and (2)
more detailed social networking data to directly
capture rather than having to infer the strength of
social ties. Second, it would be useful to conduct
inquiries into the moderating effect of IFMSTs on
the formation of different types of alliances based
on the various reasons for collaboration, inten-
sity of the collaboration and equity investments
provided by each partner. These research avenues
could provide information on additional ways
specic social mechanisms can further benet
poorly embedded rms when it comes to alliance
formation. Third, we suggest extending our work
to the large literature on mergers and acquisitions
(Capron and Shen, 2007), to explore the role of
social ties and compare the relationship dynamics
between these different contexts as they can often
have very different motivations, processes and
outcomes. This appears to be a potentially rich
area for further research. Finally, while we focused
on differentiating the effect of education-based
social ties from that of activity-based ties, an in-
vestigation into other dimensions of interpersonal
relationship dynamics, for example inter- and
intra-nationality relationships, warrants further
inquiry.
References
Ahuja, G. (2000). ‘Collaboration networks, structural holes, and
innovation:a longitudinal study’, Administrative Science Quar-
terly,45, pp. 425–455.
Ahuja, G., F. Polidoro and W. Mitchell (2009). ‘Structural ho-
mophily or social asymmetry? The formation of alliances by
poorly embedded rms’, Strategic Management Journal,30,
pp. 941–958.
Ahuja, G., G. Soda and A. Zaheer (2012). ‘The genesis and dy-
namics of organizational networks’, Organization Science,23,
pp. 434–448.
Ai, C. and E. C. Norton (2003). ‘Interaction terms in logit and
probit models’, Economics Letters,80, pp. 123–129.
Andrevski, G., D. J. Brass and W. J. Ferrier (2016). ‘Alliance port-
folio congurations and competitive action frequency’, Jour-
nal of Management,42, pp. 811–837.
Arranz, N., M. F. Arroyabe and J. C. Fernandez de Arroyabe
(2020). ‘Network embeddedness in exploration and exploita-
tion of joint R&D projects: a structural approach’, British
Journal of Management,31, pp. 421–437.
Barbulescu, R. (2015). ‘The strength of many kinds of ties: un-
packing the role of social contacts across stages of the job
search process’, Organization Science,26, pp. 1040–1058.
Barclay, M. and C. Smith (1995). ‘The priority structure of cor-
porate liabilities’, The Journal of Finance,50, pp. 899–917.
Barden, J. Q. and W. Mitchell (2007). ‘Disentangling the
inuences of leaders’ relational embeddedness on inter-
organizational exchange’, Academy of Management Journal,
50, pp. 1440–1461.
BarNir, A. and K. A. Smith (2002). ‘Interrm alliances in the
small business: the role of social networks’, Journal of Small
Business Management,40, pp. 219–232.
Berends, H., E. van Burg and E. M. van Raaij (2011). ‘Con-
tacts and contracts: cross-level network dynamics in the de-
velopment of an aircraft material’, Organization Science,22,
pp. 940–960.
Bierly III, P. E. and S. Gallagher (2007). ‘Explaining alliance
partner selection: t, trust and strategic expediency’, Long
Range Planning,40 (2), pp. 134–153.
Bonacich, P. (1987). ‘Power and centrality: a family of measures’,
American Journal of Sociology,92, pp. 1170–1182.
Borgatti, S. P. and M. G. Everett (2000). ‘Models of
core/periphery structures’, Social Networks,21, pp. 375–
395.
Borgatti, S. P., M. G. Everett and L. C. Freeman (1999). Ucinet
5.0 for Windows. Natick: Analytic Technologies.
Brass, D. J. (1985). ‘Men’s and women’s networks: a study of in-
teraction patterns and inuence in an organization’, Academy
of Management Journal,28, pp. 327–343.
Brown, J. A., P. T. Gianiodis and M. D. Santoro (2015). ‘Fol-
lowing doctors’ orders: organizational change as a response
to human capital bargaining power’, Organization Science,26,
pp. 1284–1300.
Burt, R. S. (1992). Structural Holes: The Social Structure of Com-
petition. Cambridge, MA: Harvard University Press.
Butler, A. and U. G. Gurun (2012). ‘Educational networks, mu-
tual fund voting patterns, and CEO compensation’, Review of
Financial Studies,25, pp. 2533–2562.
© 2021 British Academy of Management and Wiley Periodicals LLC.
24 K. Yang, J. J. Zhu and M. D. Santoro
Butler, J. K. (1991). ‘Towards understanding and measuring con-
ditions of trust: Evolution of a conditions of trust inventory’,
Journal of Management,17 (3), pp. 643–663.
Cai, J., R. Walkling and K. Yang (2016). ‘The price of street
friends: social networks, informed trading, and shareholder
costs’, Journal of Financial and Quantitative Analysis,51,pp.
801–837.
Capron, L. and J. C. Shen (2007). ‘Acquisitions of private vs. pub-
lic rms: private information, target selection, and acquisition
returns’, Strategic Management Journal,28, pp. 891–911.
Certo, S. T. (2003). ‘Inuencing initial public offering investors
with prestige: signaling with board structures’, Academy of
Management Review,28, pp. 432–446.
Chakravarty, A., C. Zhou and A. Sharma (2020). ‘Effect of
alliance network asymmetry on rm performance and risk’,
Journal of Marketing,84, pp. 74–94.
Chung, S., H. Singh and K. Lee (2000). ‘Complementarity, status
similarity and social capital as drivers of alliance formation’,
Strategic Management Journal,21, pp. 1–22.
Cohen, L., A. Frazzini and C. Malloy (2008). ‘The small world
of investing: board connections and mutual fund managers’,
Journal of Political Economy,116, pp. 951–979.
Cohen, L., A. Frazzini and C. Malloy (2010). ‘Sell-side school
ties’, Journal of Finance,65, pp. 1409–1437.
Cooney, J. W. Jr, L. Madureira, A. K. Singh and K. Yang (2015).
‘Social ties and IPO outcomes’, Journal of Corporate Finance,
33, pp. 129–146.
Dalton, D. R., C. M. Daily, A. E. Ellstrand and J. L. Johnson
(1998). ‘Met analytic reviews of board composition, leadership
structure, and nancial performance’, Strategic Management
Journal,19, pp. 269–290.
Dosh, K. (2016). ‘Golfers make better business executives’,
Forbes, May 16.
Engelberg, J., P. Gao and C. A. Parsons (2012). ‘Friends with
money’, Journal of Financial Economics,103, pp. 169–188.
Engelberg, J., P. Gao and C. A. Parsons (2013). ‘The price of
a CEO’s Rolodex’, Review of Financial Studies,26, pp. 79–
114.
Evans, P. B. and T. S. Wurster (1997). ‘Strategy and the new eco-
nomics of information’, Harvard Business Review,75, pp. 71–
78.
Fama, E. and K. R. French (1997). ‘Industry costs of equity’,
Journal of Financial Economics,43, pp. 153–193.
Fiol, C. M. and M. A. Lyles (1985). ‘Organizational learning’,
Academy of Management Review,10, pp. 803–813.
Fracassi, C. (2016). ‘Corporate nance policies and social net-
works’, Management Science,63, pp. 2420–2438.
Fracassi, C. and G. Tate (2012). ‘External networking and inter-
nal rm governance’, Journal of Finance,67, pp. 153–194.
Gelman, A. and D. K. Park (2009). ‘Splitting a predictor at the
upper quarter or third and the lower quarter or third’, The
American Statistician,63, pp. 1–8.
Gomerps, P. A., V. Mukharlyamov and Y. Xuan (2016). ‘The
Cost of friendship’, Journal of Financial Economics,119 (3),
pp. 626–644.
Granovetter, M. (1973). ‘The strength of weak ties’, American
Journal of Sociology,78, pp. 1360–1380.
Greene, W. (2003). Econometric Analysis, 5th edn. Englewood
Cliffs, NJ: Prentice Hall.
Gulati, R. (1998). ‘Alliances and networks’, Strategic Manage-
ment Journal,19, pp. 293–317.
Gulati, R. (2007). Managing Network Resources: Alliance, Afli-
ations and Other Relational Assets. Oxford: Oxford University
Press.
Gulati, R. and M. Gargiulo (1999). ‘Where do inter-
organizational networks come from?’, American Journal
of Sociology,104, pp. 1439–1493.
Gulati, R. and J. D. Westphal (1999). ‘Cooperative or control-
ling? The effects of CEO–board relations and the content of
interlocks on the formation of joint ventures’, Administrative
Science Quarterly,44, pp. 473–506.
Hall, S. (2011). ‘Educational ties, social capital and the translocal
(re)production of MBA alumni networks’, Global Networks,
11, pp. 118–138.
Hoang, H. and F. T. Rothaermel (2010). ‘Leveraging internal
and external experience: exploration, exploitation, and R&D
project performance’, Strategic Management Journal,31,pp.
734–758.
Hong, H., J. D. Kubik and J. C. Stein (2005). ‘The neighbor’s
portfolio: word-of-mouth effects in the holdings and trades of
money managers’, Journal of Finance,60, pp. 2801–2814.
Hu, T., A. Jain and A. Delios (2021). ‘Centrality asymmetry and
partner complementarity as inuences on alliance dissolution’,
British Journal of Management,32, pp. 59–79.
Huberman, G. (2001). ‘Familiarity breeds investment’, Review of
Financial Studies,14, pp. 659–680.
Inkpen, A. C. and W. K. Tsang (2005). ‘Social capital, network
and knowledge transfer’, Academy of Management Review,30,
pp. 146–165.
Ishii, J. and Y. Xuan (2014). ‘Acquirer–target social ties and
merger outcomes’, Journal of Financial Economics,112,pp.
344–363.
Jensen, M. and A. Roy (2008). ‘Stagingexchange partner choices:
when do status and reputation matter?’, Academy of Manage-
ment Journal,51, pp. 495–516.
King, G. and L. Zeng (2001). ‘Explaining rare events in interna-
tional relations’, International Organization,55, pp. 693–715.
Ke, R., M. Li, Z. Ling and Y. Zhang (2018). ‘Social connections
within executive teams and management forecasts’, Manage-
ment Science,65, pp. 439–457.
Koka, B. and J. E. Prescott (2002). ‘Strategic alliances as social
capital: a multidimensional view’, Strategic Management Jour-
nal,23, pp. 795–816.
McAllister, D. (1995). ‘Affect- and cognition-based trust as
foundations for interpersonal cooperation in organizations’,
Academy of Management Review,38, pp. 24–59.
McPherson, J. M., L. Smith-Lovin and J. M. Cook (2001). ‘Birds
of a feather: homophily in social networks’, Annual Review of
Sociology,27, pp. 415–444.
Mitsuhashi, H. and J. Min (2016). ‘Embedded networks and
suboptimal resource matching in alliance formations’, British
Journal of Management,27, pp. 287–303.
Moran, P. (2005). ‘Structural vs. relational embeddedness: social
capital and managerial performance’, Strategic Management
Journal,26, pp. 1129–1151.
Owen-Smith, J. and W. W. Powell (2004). ‘Knowledge networks
as channels and conduits: the effects of spillovers in the Boston
biotechnology community’, Organization Science,15,pp.5–
21.
Paruchuri, S., M. C. Goossen and C. Phelps (2019). ‘Conceptual
foundations of multilevel social networks’. In S. E. Humphrey
and J. M. LeBreton (Eds), The Handbook of Multilevel
© 2021 British Academy of Management and Wiley Periodicals LLC.
The Effect of Inter-rm Managerial Social Ties on Alliance Formation 25
Theory, Measurement, and Analysis. https://ssrn.com/
abstract=3323503
Petrin, A. and K. Train (2010). ‘A control function approach to
endogeneity in consumer choice models’, Journal of Marketing
Research,47, pp. 3–13.
Podolny, J. M. (1994). ‘Market uncertainty and the social charac-
ter of economic exchange’, Administrative Science Quarterly,
39, pp. 458–483.
Polidoro, F., G. Ahuja and W. Mitchell (2011). ‘When the social
structure overshadows competitive incentives: the effects of
network embeddedness on joint venture dissolution’, Academy
of Management Journal,54, pp. 203–223.
Pollock, T. G., P. M. Lee, K. Jin and K. Lashley (2015).
‘(Un)Tangled: exploring the asymmetric coevolution of new
venture capital rm’s reputation and status’, Administrative
Science Quarterly,60, pp. 482–517.
Powell, W. W., D. R. White, K. W. Koput and J. Owen-Smith
(2005). ‘Networks dynamics and eld evolution: the growth of
inter-organizational collaboration in the life sciences’, Ameri-
can Journal of Sociology,110, pp. 1132–1205.
Reuer, J. J. and N. Lahiri (2014). ‘Searching for alliance partners:
effects of geographic distance on the formation of R&D col-
laborations’, Organization Science,25, pp. 283–298.
Rider, C. (2012). ‘How employees’ prior afliations constrain or-
ganizational network change: a study of U.S. venture capital
and private equity’, Administrative Science Quarterly,57,pp.
453–483.
Rosenkopf, L. and G. Padula (2008). ‘Investigating the mi-
crostructure of network evolution: alliance formation in the
mobile communications industry’, Organization Science,19,
pp. 669–687.
Rosenkopf, L., A. Metiu and V. P. George (2001). ‘From the bot-
tom up? Technical committee activity and alliance formation’,
Administrative Science Quarterly,46, pp. 748–772.
Rothaermel, F. and W. Boeker (2008). ‘Old technology meets new
technology: complementarities, similarities, and alliance for-
mation’, Strategic Management Journal,29, pp. 47–77.
Rowley, T. J., D. Behrens and D. Krackhardt (2000). ‘Re-
dundant governance structures: an analysis of structural
and relational embeddedness in the steel and semiconduc-
tor industries’, Strategic Management Journal,21, pp. 369–
386.
Santoro, M. D. (2013). ‘Uncertainty, governance alignment, and
rm performance in biotechnology alliances’, International
Journal of Biotechnology,12, pp. 225–248.
Seasholes, M. S. and N. Zhu (2010). ‘Individual investors and
local bias’, Journal of Finance,65, pp. 1987–2010.
Selnes, F. and J. Sallis (2003). ‘Promoting relationship learning’,
Journal of Marketing,67, pp. 80–95.
Shipilov, A. (2012). ‘Strategic multiplexity’, Strategic Organiza-
tion,10, pp. 215–222.
Shipilov, A. V., S. X. Li and H. R. Greve (2011). ‘The prince and
the pauper: search and brokerage in the initiation of status-
heterophilous ties’, Organization Science,22, pp. 1418–1434.
Stuart, T. E., H. Hoang and R. C. Hybels (1999). ‘Inter-
organizational endorsements and the performance of en-
trepreneurial ventures’, Administrative Science Quarterly,44,
pp. 315–349.
Uzzi, B. (1996). ‘The sources and consequences of embeddedness
for the economic performance of organizations’, American So-
ciological Review,61, pp. 674–698.
Uzzi, B. (1997). ‘Social structure and competition in interrm
networks: the paradox of embeddedness’, Administrative Sci-
ence Quarterly,42, pp. 35–67.
Uzzi, B. (1999). ‘Embeddedness in the making of nancial cap-
ital: how social relations and networks benet rms seeking
nancing’, American Sociological Review,64, pp. 481–505.
Uzzi, B. and R. Lancaster (2004). ‘Embeddedness and price for-
mation in the corporate law market’, American Sociological
Review,69, pp. 319–344.
Villalonga, B. and A. M. McGahan (2005). ‘The choice among
acquisitions, alliances, and divestures’, Strategic Management
Journal,26, pp. 1183–1208.
Wang, L. and E. J. Zajac (2007). ‘Alliance or acquisition?
A dyadic perspective on interrm resource combinations’,
Strategic Management Journal,28, pp. 1291–1317.
Wiersema, M. and H. P. Bowen (2009). ‘The use of limited depen-
dent variable techniques in strategy research: issues and meth-
ods’, Strategic Management Journal,30, pp. 679–692.
Xue, L., K. Yang and Y. Yao (2018). ‘Examining the effects of
interrm managerial social ties on IT components diversity:
an agency perspective’, MIS Quarterly,42, pp. 679–694.
Zaheer, A., R. Gözübüyük and H. Milanov (2010). ‘It’s the con-
nections: the network perspective in inter-organizational re-
search’, Academy of Management Perspectives,24, pp. 62–77.
Ke Yang is an Associate Professor of Finance at the College of Business, Lehigh University. Her re-
search focuses on mergers and acquisitions, corporate governance and social networks.
Jianjun Zhu is an Assistant Professor of Marketing at the College of Business, New Mexico State
University. His research focuses on strategic alliance, innovation, business ethics, international man-
agement, crowdsourcing, sharing economy, mobile marketing, cyber security and marketing strategy.
Michael D. Santoro is Professor of Management and holds the William R. Kenan Jr. Professorship at
the College of Business, Lehigh University. His research focuses on strategic alliances, strategic change
and the external sourcing of knowledge in driving technological innovation.
© 2021 British Academy of Management and Wiley Periodicals LLC.
