Content uploaded by Frederic Godart
Author content
All content in this area was uploaded by Frederic Godart on Nov 02, 2017
Content may be subject to copyright.
Strategic Management Journal
Strat. Mgmt. J.,38: 1232– 1252 (2017)
Published online EarlyView 15 December 2016 in Wiley Online Library (wileyonlinelibrary.com) DOI: 10.1002/smj.2602
Received 27 May 2015;Final revision received 4 October 2016
WHICH BOUNDARIES? HOW MOBILITY NETWORKS
ACROSS COUNTRIES AND STATUS GROUPS AFFECT
THE CREATIVE PERFORMANCE OF ORGANIZATIONS
ANDREW SHIPILOV,*FRÉDÉRIC C. GODART, and JULIEN CLEMENT
INSEAD, Fontainebleau, France
Research summary: Losing key employees to competitors allows an organization to engage in
external boundary-spanning activities. It may benet the organization through access to external
knowledge, but may also increase the risks of leaking knowledge to competitors. We propose that
the destination of departed employees is a crucial contingency: benets or risks only materialize
when employees leave for competitors that differ from the focal organization along signicant
dimensions, such as country or status group. In the context of the global fashion industry, we
nd that key employees’ moves to foreign competitors may increase (albeit at a diminishing rate)
their former employers’ creative performance. Furthermore, rms may suffer from losing key
employees to higher- or same-status competitors, but may benet from losing them to lower-status
competitors.
Managerial summary: Losing key employees to competitors can provide organizations with access
to external knowledge, but increase risks of leaking knowledge to competitors. We nd that an
organization’s access to external knowledge and its risks of knowledge leakage through employee
mobility may be affected by whether its employees leave for competitors in a foreign country
or in a different status group. In the context of the global fashion industry, we show that key
employees’ moves to foreign competitors increase (up to a point) their former employers’ creative
performance. Furthermore, rms may suffer from losing key employees to higher- or same-status
competitors, but benet from losing them to lower-status competitors. Hence, executives in creative
industries and possibly beyond could welcome losing employees to competitors in foreign countries
or to lower-status competitors. Copyright © 2016 John Wiley & Sons, Ltd.
INTRODUCTION
An organization enjoys superior creative perfor-
mance when it consistently generates novel and
useful products or services (Almeida, 1996; Cattani
and Ferriani, 2008; Godart, Shipilov, and Claes,
2014). Creative performance depends on multiple
social determinants (Cattani and Ferriani, 2008;
Keywords: employee mobility; networks; foreign experi-
ence; creative performance; boundary-spanning
*Correspondence to: Andrew Shipilov, Strategy Area, INSEAD,
Boulevard de Constance, Fontainebleau 77300, France. E-mail:
shipilov@insead.edu
Copyright © 2016 John Wiley & Sons, Ltd.
Perry-Smith and Shalley, 2003). One such deter-
minant is the presence of “boundary spanners”,
individuals who build social ties to other organiza-
tions (Barrett et al., 2012; Tushman and Scanlan,
1981). The existing literature provides a good
understanding of the mechanisms through which
organizations can benet from boundary spanners,
who provide access to new ideas from other rms.
As the focal organization recombines this external
knowledge with its own, it can consistently gener-
ate new ideas that result in novel and useful output
(e.g., Rosenkopf and Nerkar, 2001). However, the
benets of boundary spanning may be offset by its
costs: just as boundary spanners can access external
Mobility Networks 1233
knowledge, they can also leak the organization’s
own knowledge to competitors (Aime et al., 2010;
Somaya, Williamson, and Lorinkova, 2008).
Available research assumes that these costs and
benets materialize whenever employees span the
formal boundary that separates their organization
from any other (e.g., Corredoira and Rosenkopf,
2010; Godart et al., 2014; Rosenkopf et al., 2001;
Uzzi and Spiro, 2005). This may be neglecting an
important contingency: boundary spanners should
generate different benets and costs for their orga-
nizations depending on which boundary they span.
Firms are likely to benet only when bound-
ary spanners connect them to competitors whose
knowledge is signicantly new relative to what
the rm already knows. Similarly, a rm should
experience costs only when its boundary spanners
leak knowledge to competitors that can exploit
this knowledge at the expense of the focal rm.
In this article, we focus on geography and status
differences as two factors that affect these bene-
ts and costs. Specically, we examine the con-
sequences of an organization’s losing employees
to competitors in foreign countries or different
status groups.
Employee mobility is a particular form of bound-
ary spanning that can trigger the two mechanisms
discussed above. As current employees of the focal
rm stay in touch with former colleagues or merely
pay attention to what their former colleagues are
doing, they are likely to incorporate competitors’
knowledge into the focal rm’s ideas and output
(Corredoira and Rosenkopf, 2010; Dokko and
Rosenkopf, 2010). Likewise, by hiring the focal
rm’s employees, competitors can learn what the
rm knows and incorporate this knowledge into
their own output (Phillips, 2002). Hence, both by
losing employees to competition and by hiring them
from competition, the rm becomes embedded
in the industrywide mobility network. Arguing
that the benets of access to knowledge and the
costs of knowledge leakage strengthen at different
rates with the degree of external mobility, prior
research has suggested the existence of a nonlinear
relationship between the number of competitors
to which a focal rm loses its employees and its
creative performance (Godart et al., 2014).
Yet, much like other studies on boundary
spanning, research on employee mobility and orga-
nizational creative performance remains agnostic
to differences among the competitors to which a
rm loses its employees, particularly to differences
driven by the competitors’ geographic location and
status (Corredoira and Rosenkopf, 2010; Dokko
and Rosenkopf, 2010; Godart et al., 2014). We
propose that by crossing country or status bound-
aries in search of new jobs, employees may affect
the balance of benets and costs experienced by
their former employer as a result of their mobility.
This is because rms located in different countries
or status groups have different knowledge from
those located within the same country or status
group. Thus, when an organization’s employees
move to competitors in foreign countries or dif-
ferent status groups, they affect both their former
organization’s ability to use knowledge from
these competitors and also the risks of competi-
tors’ benetting from the knowledge of the focal
organization.
We investigate our claims in a global, indus-
trywide longitudinal dataset (2000–2010) compris-
ing information on the mobility of designers across
high-end fashion houses and these houses’ ability to
produce creative fashion collections. Our ndings
suggest that losing employees to foreign competi-
tors has a positive (concave) relationship with the
organization’s creative performance, yet losing peo-
ple to domestic competitors has no impact. We also
nd that rms tend to benet from losing employees
to lower-status competitors, but suffer from losing
employees to the same or higher-status competitors.
This article contributes to research on how
boundary spanning in social networks—and
especially the networks formed through employee
mobility—affects organizational creativity (Cat-
tani and Ferriani, 2008; Dokko and Rosenkopf,
2010; Godart et al., 2014). Our results also offer
more precise prescriptions for organizations than
have been available to date: instead of merely
“spanning boundaries” to become exposed to
novelty, organizations could strive to span specic
types of boundaries, and avoid spanning others.
GEOGRAPHY, STATUS, AND CREATIVE
PERFORMANCE
The notion of formal organizational bound-
aries is a critical building block of organization
theory. Formal external boundaries separate
organizations from the outside world (McEvily,
Soda, and Tortoriello, 2014). These boundaries
emerge as organizations minimize transaction
costs (Williamson, 1981) or take advantage of
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1234 A. Shipilov, F. C. Godart, and J. Clement
efciencies in internal storage and transfer of
knowledge (Kogut and Zander, 1992). Transferring
knowledge across formal boundaries is often
challenging, especially when knowledge is tacit
(Rosenkopf et al., 2001).
Some individuals— boundary spanners—have
social relationships that cut across formal
organizational boundaries and help their cur-
rent employer access external knowledge. External
boundary-spanning network ties represent a dis-
tinct form of exploration (March, 1991), as they
enable an organization to acquire new knowledge
that is not available internally (Henderson and
Cockburn, 1994). Provided that boundary spanners
share knowledge inside their organization, other
employees can then incorporate these new inputs
into the creative process and generate novel idea
combinations.
External boundary spanning can take a vari-
ety of forms. As engineers or scientists feel the
need to access external knowledge for R&D
work, they build on the research and patents of
other companies (Rosenkopf et al., 2001). More
recently, cross-company employee mobility has
emerged as an important purveyor of external
boundary-spanning ties. Many industries are
increasingly characterized by “boundaryless
careers” (Bidwell and Briscoe, 2010): individuals
regularly change employers, often leaving one rm
for its competitors or clients. Employee mobility to
clients is considered uniformly benecial because
it helps the rm acquire new business (Somaya
et al., 2008). Yet there are also advantages to
losing employees to competitors. Indeed, departed
individuals often keep communicating with their
former colleagues (Corredoira and Rosenkopf,
2010: 161). This helps their former employer
learn what competitors are currently working on.
Even in the absence of explicit communication
between the departed individual and his or her
former colleagues, an employee’s movement to
a competitor can prompt former colleagues to
pay more attention to what this competitor is
doing. This occurs because losing employees to a
competitor gives saliency to this competitor’s work
in the eyes of the focal rm’s members. Consistent
with these arguments, Corredoira and Rosenkopf
(2010) found that rms were more likely to cite
the patents of competitors that hired their departed
employees.
Despite these benets, boundary-spanning rela-
tionships can also be costly: losing employees to
competitors means transferring knowledge to them.
An employee who moves from one rm to another
not only carries abstract appreciation of the rm’s
practices and representations, but can also export
his former employer’s most recent knowledge. In
line with this argument, Phillips (2002) nds that
law rms that lost employees to competitors have a
high likelihood of failure, especially when departed
employees occupied highly ranked positions.
A shortcoming of research on boundary spanning
in the context of boundaryless careers (Corredoira
and Rosenkopf, 2010; Dokko and Rosenkopf, 2010;
Godart et al., 2014) is its ambivalence as to how
the competitors to which a rm loses its employ-
ees are meaningfully different from the focal rm.
Geography and status are two important sources of
such differences. Employee mobility across geogra-
phies or status groups links the focal organization
to different communities of practice. Communi-
ties are dened as “relatively homogenous sectors
occupied by actors who are likely to share com-
mon values, attitudes and interests” (Knoke et al.,
1996: 23), demarcated by boundaries with different
degrees of institutionalization. Some boundaries are
supported by stable and objective sustaining mech-
anisms, whereas others are more exible and are
sustained by intersubjective mental maps of indus-
try members. Country borders and status groups
represent two extremes of this continuum. On the
one hand, country borders are highly institution-
alized boundaries that are sustained by geography
and the state (Elias and Jephcott, 1982), and cross-
ing them involves mobility across what Bell and
Zaheer (2007) aptly label “geographic holes.” On
the other hand, boundaries between status groups
are institutionalized to a much lesser extent and are
based on intersubjective agreement between differ-
ent stakeholders regarding the relative quality rank-
ings of market participants (Phillips and Zucker-
man, 2001).
The diversity of knowledge is likely to be lower
within status and geographic boundaries than across
them. Furthermore, relative quality rankings— on
the basis of which the status boundaries are built
(Podolny, 2005)—may affect the distribution of
audiences’ attention on the producers. This has
implications for the perceptions of producers’ cre-
ativity, as higher-status producers tend to attract
more attention and receive greater credit for their
ideas than lower-status producers. Taken together,
status and geographic positions of competitors
represent salient environmental and institutional
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1235
factors that affect the diversity of knowledge as well
as the intensity of competition for ideas to which
an organization is exposed when it generates and
implements novel and useful ideas. This, in turn,
should affect the payoffs which rms can obtain
from employee mobility (Mawdsley and Somaya,
2016).1
Crossing national boundaries
Knowledge pools are geographically localized
(Almeida, 1996), so that knowledge is more
homogenous within geographic locations than
across them (Bell and Zaheer, 2007). The homog-
enization of knowledge within locations happens
because co-located individuals meet and exchange
knowledge at lower cost than those who are
geographically separated (Saxenian, 1996) and
co-located individuals are likely to share similar
belief systems (James, 2007). In addition to sim-
ple geographic proximity, national borders also
demarcate different pools of knowledge (Oettl
and Agrawal, 2008). These borders are highly
institutionalized in that they delimit societies with
different cultures and state institutions (Barkey
and Parikh, 1991). One can easily nd cultural
differences across national borders: commonplace
practices in one country may appear vastly foreign
in another (Hofstede, 1980; Schwartz, 1994).
There is a reason to expect that companies will
neither benet nor suffer from personnel loss to
competitors in the same country. When domestic
competitors hire the focal rm’s employees, the
remaining employees will not learn ideas through
the knowledge-sharing and attention-focusing
channels that would give them a competitive edge
vis-à-vis their competitors. This is because domes-
tic competitors are exposed to similar cultural
practices and have the same belief systems as
the focal rm. Furthermore, the costs of outward
mobility (i.e., from the focal rm to competitors)
within an organization’s home country will also
be limited because the practices transferred by
1Our theory assumes a ceteris paribus clause with respect to
the heterogeneity of individuals. That is, when we argue for the
knowledge ow and attention-focusing effects resulting from a
rm’s personnel loss to competitors, we assume that the quality of
its former employees’ human capital is held constant (Chadwick
and Dabu, 2009; Coff, 1997). We designed our empirical analyses
to directly control for the quality of mobile individuals as well as
for the quality of those who remain in the fashion houses.
departed individuals to competitors are not cultur-
ally novel or useful from their competitors’ point
of view. Hence, one could expect that there will be
no relationship between the number of domestic
competitors to which the rm loses its employees
and a rm’s creative performance.
However, there could be a nonlinear (inverted
U-shaped) relationship between mobility to foreign
competitors and creative performance. When
mobility to foreign competitors initially starts to
increase, the rm will obtain signicant benets
relative to the baseline case of not losing employ-
ees to foreign competition. This is because the
organization’s remaining employees are likely to
keep having conversations and interactions with
departed employees who now work in cultural
environments with different values, attitudes, and
interests, or simply to pay more attention to what
their former colleagues’ new employers are doing
in these environments. By contrast, the employees
in a rm that experiences no personnel loss to
foreign competitors will not receive a comparable
exposure (or will not pay as much attention) to
foreign cultural environments. The benets from
access to novel cross-cultural ideas will satiate,
however, when a rm loses people to too many for-
eign competitors: once the rm receives too many
culturally diverse ideas from prior employees, its
current employees will either start ignoring some
of them or feel overwhelmed by the breadth of
possible choices (Grant and Schwartz, 2011).
On the cost side, the departures of employees
to foreign competitors can increase the likelihood
of transferring the organization’s practices to them
(Aime et al., 2010), which is less likely to occur
when the rm does not lose employees to foreign
competitors. Given that knowledge pools are geo-
graphically localized, foreign competitors typically
have access to their own culturally appropriate set
of practices. By importing practices from another
country through hiring, a competitor’s set of choices
for product and service ideas increases. These costs
escalate with an increase in the number of for-
eign rms to which the focal rm loses its former
employees.
Given the preceding arguments, one should
expect the marginal benets of employee loss to
competitors in different countries to exceed the
marginal costs when the number of foreign com-
petitors to which a rm has lost employees is low
to moderate. This will result in an initial increase
of creative performance relative to the baseline
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1236 A. Shipilov, F. C. Godart, and J. Clement
case of no personnel loss to foreign competitors.
When the number of foreign competitors to which
the rm has lost employees is high, the marginal
costs could offset marginal benets, resulting in
the subsequent reduction of creative performance.
Thus, we propose the rst hypothesis:
Hypothesis 1 (H1): There is an inverted
U-shaped relationship between the number of
foreign competitors to which a rm loses its
employees and this rm’s creative performance.
Crossing status boundaries
Status signals quality, especially when quality
is difcult to determine ex ante (Podolny, 2005).
Status stratication emerges when actors are ranked
according to their relative status (Bothner, Kim,
and Smith, 2012), so that most industries have
commonly “agreed-upon” high-, medium-, and
low-status groups (Phillips and Zuckerman, 2001).
Such status hierarchies are not formalized, but they
are enacted and sustained via the stakeholders’
intersubjective agreement (Podolny, 2005).
The products of higher-status rms attract more
attention from customers than those of lower-status
rms (Podolny, 1993). Likewise, the ideas gener-
ated by higher-status rms are considered more
legitimate than those of lower-status rms. As
higher-status players tend to be trendsetters in their
industries (Rao, Greve, and Davis, 2001), their
opinions often inuence other industry members.
By contrast, lower-status rms frequently possess
ideas that break from the industry’s conventions
(Cattani and Ferriani, 2008). For example, many
twentieth-century breakthrough innovations in
art (e.g., cubism [Sgourev, 2013]), science (e.g.,
relativity theory), or even nancial markets
(e.g., junk bonds) initially came from low-status
players.
Status positions, and the knowledge to which
individuals who occupy such positions are exposed,
are maintained through homophilous attachment
and socialization. Higher-status actors tend to
associate with other high-status actors to preserve
their exclusivity and avoid leaking their status to
lower-status actors or be “contaminated” by them;
therefore the latter are forced to associate with
one another (Podolny, 2005). Knowledge exchange
through social ties socializes actors of similar status
into having similar kinds of ideas. For example,
high-status fashion houses employ high-status
designers—many of whom received elite training
in prestigious schools (e.g., the Parsons School of
Design in New York or the École de la Chambre
Syndicale de la Couture Parisienne in Paris). These
individuals would, for example, agree that clothing
has to be properly stitched together. Yet low-status
designers might try to challenge this convention.
This happens because a peripheral position helps
low-status actors champion divergent ideas without
the fear of conicting with the accepted norms
of the eld (Perry-Smith and Shalley, 2003;
Phillips and Zuckerman, 2001).
Much like a former employee crossing a
national border may help their prior employer
explore novel ideas, a former employee crossing
a status boundary—through the acceptance of
a job with a competitor belonging to a different
status group—can facilitate their past employer’s
exploratory search through spanning status bound-
aries. Employee mobility across status groups
could provide the former employer with access
to new ideas— i.e., ideas available in the other
status group that are either unknown to the focal
organization or to which the focal organization has
not yet paid enough attention. Depending on the
destination of employee mobility, the organization
can get two kinds of ideas: new legitimate ideas
from higher-status competitors that do not involve
radical departure from the industry convention, or
new radical ideas from lower-status competitors
that may even be new to the industry (Cattani and
Ferriani, 2008).
Losing employees to higher-status competitors is
likely to generate signicant benets and costs. This
type of mobility establishes knowledge exchange
and attention-focusing channels around legitimate
yet currently untapped (new) ideas that the rm’s
current employees can incorporate into new prod-
ucts. By contrast, an organization that does not
lose employees to higher-status competitors will not
have access to such ideas. The benets of exposure
to these ideas may satiate with an increase in the
number of higher-status competitors that hire the
rm’s employees, as the rm’s employees become
overwhelmed and start ignoring some of their ideas.
At the same time, the status of a former
employee’s new employer is also likely to dramat-
ically affect the costs of outward mobility for the
focal rm. Indeed, status hierarchies give rise to a
“Matthew effect” whereby high-status actors obtain
greater recognition for a given accomplishment
than lower-status actors receive for a similar
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1237
achievement (Merton, 1968). If a higher-status
competitor learns new ideas from another rm’s
former employees and incorporates them into its
own products, then this higher-status competitor
is likely to get more credit for these ideas than the
focal rm where they originated. This competitor’s
output is also likely to attract more of audiences’
attention if the competitor has higher status than
the rm where the original ideas came from. As a
result, the true originator of these ideas is likely to
be perceived as offering less-novel products. The
costs of employee loss to higher-status competitors
will grow with an increase in the number of com-
petitors to which the rm has lost its talent. This
is because audiences will be increasingly likely to
forget (or might not even bother to observe) that
such competitors’ ideas originated within the focal
rm. By contrast, the absence of employee losses to
higher-status competitors can help the organization
to avoid the risks of feeding the competitors’
creative pipelines.
When the number of higher-status competitors
that recruit the rm’s employees starts to increase
from low to moderate, one should expect the
marginal benets of employee loss to higher-status
competitors to initially exceed the marginal costs.
This is because access to some new legitimate
ideas is better than no access, and also because the
organization is leaking its ideas to only a hand-
ful of higher-status competitors. This will result
in an initial increase of creative performance rel-
ative to the baseline case of no personnel loss
to higher-status competitors. When the number of
such higher-status competitors is high, the marginal
costs of employee loss would offset its marginal
benets. This is because the benets of exposure to
legitimate ideas will satiate, as the rm cannot take
advantage of all of them, but the costs of feeding the
higher-status competitors’ pipelines will escalate,
as now there is a very large number of higher-status
competitors that could appropriate the focal rm’s
ideas. This will eventually result in the reduction of
a rm’s creative performance. Thus, we propose the
second hypothesis:
Hypothesis 2 (H2): There is an inverted-
U-shaped relationship between the number
of higher-status competitors to which a rm
loses its employees and this rm’s creative
performance.
Losing employees to lower-status competitors
is also likely to impact an organization’s creative
performance because it involves mobility across
a status boundary. When a rm loses employees
to lower-status competitors, it gains more access
to new nonconformist ideas that break with the
mainstream, relative to the baseline case of no
mobility to such competitors. Yet when a rm’s
employees leave for lower-status competitors, the
resulting transfer of practices is unlikely to generate
signicant costs because the focal rm’s higher
status will always attract the audience’s attention
to its own products, including in instances when
the rm incorporates ideas from the lower-status
competitors. In turn, even if lower-status com-
petitors incorporate the focal rm’s ideas into
their products, the audiences’ judgments will not
penalize the products of the focal rm because of
its higher status. Since the benets of increased
access to sources of nonconformist ideas will grow
with an increase in the number of lower-status
competitors that hire the rm’s employees, but the
costs do not, then the following should be true:
Hypothesis 3 (H3): There is a positive relation-
ship between the number of lower-status com-
petitors to which a rm loses its employees and
this rm’s creative performance.
By contrast, when a rm loses employees to
same-status competitors, its ability to access new
legitimate or nonconformist ideas is limited because
it likely already has access to similar ideas inter-
nally. Likewise, the costs of transferring the rm’s
ideas to same-status competitors are unlikely to
be high because competitors already share similar
ideas within the same status group. Hence, we do
not expect a signicant relationship between the
number of same-status competitors that hire the
focal rm’s employees and its creative performance.
DATA AND METHODS
Study setting
We conducted this study in the context of the global
high-end fashion industry. This industry offers an
excellent opportunity to explore the link between
the mobility of employees and the creativity of their
ex-rms: in fashion, creativity is paramount to com-
mercial success and talent mobility is the norm
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1238 A. Shipilov, F. C. Godart, and J. Clement
(Caves, 2000; Godart and Mears, 2009). Despite
the rivalries between their fashion houses, design-
ers consider themselves as members of a creative
community that transcends the houses’ boundaries
and encompasses the whole industry. The indus-
try is lled with examples of “reverse knowledge
transfers” as a result of losing employees to com-
petition. Appendix 1 in File S1 provides two salient
examples that illustrate the transfer of knowledge
across country boundaries and status boundaries.
Data collection and variables
We constructed the industrywide dataset on the
global high-end fashion industry over 21 fash-
ion seasons—covering both Fall/Winter and
Spring/Summer collections—and 261 fashion
houses on which we did not have missing values
for variables used in this study. Our data spans
the period from 2000 through 2010. We acquired
this data mostly from public sources, such as
industry publications and company websites. First,
we identied rms by collecting the names of all
houses that had organized major fashion shows
over the period considered (Breward, 2003; Kawa-
mura, 2005, 2011). Second, we collected career
histories of designers who worked for these houses
from industry encyclopedias (Price Alford and
Stegemeyer, 2009; Vergani, 2010), as well as from
leading industry publications (e.g., Wome n ’s Wear
Daily,Journal du Textile,Vogue). We also used
websites (e.g., fmd.com, nymag.com, style.com)
and Factiva. Data on designers span a period starting
in the 1930s and ending in 2010. We collected infor-
mation on designers before and after they became
creative directors, which allowed us to avoid
sampling on designers’ mobility during a particular
career stage. We also conducted over 40 interviews
with industry insiders between 2007 and 2014.
Dependent variable
To compute our dependent variable— Creative Per-
formance—we accessed the fashion collections’
creativity ratings in the reference French trade mag-
azine Journal du Textile (JdT) (Barkey and Godart,
2013; Crane, 1997). To construct this measure,
JdT asks industrial buyers to evaluate the creativ-
ity of fashion collections in both Fall/Winter and
Spring/Summer fashion shows. To buyers, creativ-
ity is a combination of the novelty of a collection
as well as their assessment of the probability that
this collection will actually get sold in stores. JdT
asks the buyers to give 20 points to the most cre-
ative collection for the given season, and 0 points
to collections that they do not think are creative.
Each buyer is able to give points to a maximum of
20 houses. We collected the ratings that each indi-
vidual buyer gave to each individual collection. Our
dependent variable incorporates the average num-
ber of points that a given fashion house’s collection
received from all of the buyers for a particular sea-
son.2
Independent variables
The fashion industry boasts substantial within-
country and between-country mobility. Appendix 2
in File S1 presents a matrix of employee mobility
events within and across countries between 2000
and 2010. Diagonal values show how many individ-
uals have changed jobs within a country. Entries in
row (i) and column (j) indicate how many people
left a house located in country (i) to work in a house
located in country (j).
We used six-year windows to construct mobility-
related theoretical variables. When an organization
loses employees to competition, it becomes embed-
ded in an industry’s mobility network. We computed
aDomestic Mobility Out-Degree as the number of
houses within the same country to which the focal
fashion house is connected through outward mobil-
ity of its past designers in a given six-year period.
For example, if between 2000 and 2005, six design-
ers left a U.S. house and joined four different U.S.
houses, then the focal house’s Domestic Mobility
Out-Degree would be four. Following Godart et al.
(2014), we computed Foreign Mobility Out-Degree
as the number of houses across different countries to
which the focal fashion house becomes connected
through designers’ outward mobility. For example,
if six designers left a U.S. house and joined two
2Two previous papers used JdT scores under somewhat different
labels. Godart et al. (2014) used the sum of points given by the
raters and labeled this measure Creative Market Performance,
while Godart et al. (2015) used the average of points and labeled
their measure Creative Innovations. The different labels, although
being synonymous and both referring to the creativity of fashion
houses, highlight slightly different angles— market success in a
competitive environment for talent in the former case, and the
creativity of collections (product innovations) in the latter. In the
present study we compute our dependent variable in the same
wayasinGodartet al. (2015) and label it for simplicity Creative
Performance.
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1239
different houses in France, then the house’s Foreign
Mobility Out-Degree would be two.
To operationalize mobility within/across status
groups, we followed Godart et al. (2014) and cre-
ated a measure that would accurately reect the
status of the fashion houses. We did so through
identifying the extent to which houses shared fash-
ion models during the fashion shows. We exploited
the fact that during every fashion season there is
a shortage of sought-after fashion models. Since
a model cannot appear in an unlimited number
of fashion shows during the same fashion week,
the agencies that represent the models have to dis-
criminate among booking requests from different
houses. These agencies are “doing the grading of
the fashion houses,” with fashion models being
the “exchange currency” (Mears, 2011). Thus, as
the status of a house gets higher, it is able to
book more sought-after models. Operationally, we
constructed sociomatrices in which the value of a
cell cij is the number of models shared by fash-
ion houses iand jover two seasons during a
calendar year.Annual aggregation is appropriate
because modeling agencies develop annual model
allocation strategies. Based on each sociomatrix, we
computed each fashion house’s normalized power
centrality in the model-sharing network (Podolny,
1993). High centrality means that the fashion house
gets to share sought-after models with other houses
that use sought-after models, which is consistent
with the industry’s conception of status. This mea-
sure has high face validity: the houses with top
scores include such high-status brands as Alexander
McQueen, Chanel, Christian Dior, Dolce & Gab-
bana, Fendi, etc.
The resulting status measure was used to con-
struct the rank order of the fashion houses for
every year. Then we identied the high-, medium-,
and low-status groups in the fashion industry by
computing scores corresponding the 33rd and 66th
percentiles of status in a given year. It was impor-
tant to empirically identify status groups because
these tend to be relatively stable communities of
houses and the industry stakeholders recognize
them as such. Next we identied which mobility
events corresponded to employee loss to a higher,
same, or lower-status group. For instance, between
the years 2000 and 2005, a focal fashion house
located in the medium-status group might have lost
designers to two houses located in the higher-status
group, to three fashion houses located in the
lower-status group, and to one house located in the
same-status group. Then this house’s Higher-Status
Mobility Out-Degree would be equal to two, its
Same-Status Mobility Out-Degree equal to one,
and its Lower-Status Mobility Out-Degree equal
to three. Importantly, we determined the relative
positions of houses in the status groups based on
the status scores in the same year during which
mobility events actually occurred. For example,
within the 2000–2005 time window, designer
mobility in the year 2000 would be evaluated using
the status scores computed for the year 2000 and
his or her mobility in 2001 evaluated using 2001
status scores.
Control variables
Our models include a range of controls. To begin,
we controlled for the absolute value of House Sta-
tus, using the normalized power centrality measure.
This helped hold the status level of the house con-
stant in regressions examining mobility across or
within status groups.
In order to show that our results are indeed
driven by knowledge ows and attention focusing
to and from the departed designers, we had to con-
trol for an important alternative explanation: results
may be driven by the quality of human capital,
so that employees leave for different types of new
employers (especially employers from different sta-
tus groups) depending on their talent. While it is
very hard to directly observe the quality of individ-
uals in archival research, we fortunately can turn to
an indirect proxy, namely, the awards that industry
peers give to designers to recognize their talent and
achievements. To that end, we collected data on the
recipients of major awards: the Council of the Fash-
ion Designers of America (CFDA) Fashion Awards
and British Fashion Awards. These are globally
renowned awards (e.g., CFDA Awards are often
referred to as “the Oscars of Fashion”) and, despite
the words “America” and “British” in the titles, any
foreign designer at any stage of their career can
win either (or both) of these awards. For the time
period considered, neither Milan- nor Paris-based
institutions offer equivalent awards. Even though
our dataset covers collections presented from 2000
through 2010, we identied the winners of all of
these awards from their inception (i.e., 1984 for
the CFDA Award and 1989 for the British Fash-
ion Award). Individuals who won these awards are
considered to be star designers—i.e., profession-
als whose career mobility could entail a signicant
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1240 A. Shipilov, F. C. Godart, and J. Clement
transfer of human capital and, thus, act as an alter-
native explanation for our relationships.
We used the information on these awards to
compute several variables. One variable is labeled
Number of Hired Stars and is a count of how
many award-winning designers a fashion house
hired in the same time window on the basis of which
our mobility out-degree measures were computed.
We also computed a variable labeled Number of
Lost Stars, a count of how many award-winning
designers a fashion house lost in the same window.
Since the loss (or hire) of a star might be particularly
felt right after the event, we computed indicator
variables Recent Star Loss and Recent Star Hire set
to one if the house lost or hired at least one star
designer within the 12 months preceding the fashion
show that the buyers were asked to evaluate. We also
controlled for the availability of talent inside the
fashion houses. To that end, we computed a variable
Star Creative Director that captured how many
award-winning creative directors were responsible
for the focal collection. We also constructed a
variable called House’s Number of Awards that
reected the total number of awards designers
earned while working for the focal house as opposed
to this house’s “acquisition” of awards by hiring
stars. House’s Number of Awards controls for the
extent to which a house is a hotbed of talent—i.e.,
has unique management practices that allow it to
develop talent, as opposed to hiring talent from the
outside.
The payoffs to designer mobility can also be
affected by the existence of a company’s former
employees both in the industry in general and in
the houses to which it lost employees in particular.
To capture whether there is an alumni network in
the specic destination houses or, alternatively, in
the industry as a whole, we have constructed two
dummies: Alumni Network in Destination House
and Alumni Network in Industry. The rst was set
to one if a house has lost employees to at least
one competing house that had already hired its
employees in the past (e.g., Tom Ford moves to
Gucci from Perry Ellis and then another designer
moves from Perry Ellis to Gucci) and the second
variable was set to one if there was at least one
former employee of the focal fashion house working
in any other fashion house (e.g., Tom Ford moves
to Gucci from Perry Ellis and then another designer
moves from Perry Ellis to Chanel).
France and Italy are two key countries in the
global fashion industry (Godart, 2012). When
someone moves to work in any French house, this
person may be expected to learn from the best
industry experts on how to design fashion items.
When someone moves to work in Italy, this person
can be expected to learn from the experts on how
to actually manufacture fashion items. No other
countries in the world have such stature. Thus,
we controlled for how many individuals have left
the focal fashion house to work for competitors in
these two countries, which gave us two variables,
Designers Lost to Italian Houses and Designers
Lost to French Houses.
We also controlled for the amount of Media Cov-
erage of the focal fashion house by computing
the log of the unity plus the total number of arti-
cles covered by Factiva about each house for the
time period preceding a fashion season. This vari-
able was a proxy for stakeholders’ attention to the
focal fashion house (Godart and Mears, 2009). We
controlled for the house’s age by computing how
many years had elapsed from its founding (variable
House Age).
Additionally, we computed variables associated
with creative directors who were responsible for
the collections evaluated by buyers: directors’ age
(Age of Creative Director), length of director’s
tenure at the current fashion house (Creative
Director’s Tenure), the number of different fashion
houses this person worked at (Creative Director’s
Number of Houses), and whether the director’s
position involved working with other creative
directors (Team of Creative Directors). We also
counted the logged value of the total number of
designers working in the fashion house, including
the designer assistants (this variable was labeled
Number of Designers). We also computed an
indicator variable Creative Director’s Fashion
Education coded as one if creative directors had
studied fashion in school. When variables (like Age,
Tenure, or Fashion Education) captured teams of
creative directors, we took the average of relevant
characteristics.
Just like the fashion houses have outward
centrality in the mobility network from losing
employees, they also have inward centrality in the
mobility network through hiring employees from
competitors in the same or different countries as
well as in the same or different status strata. Using
similar heuristics as in computing outward cen-
trality, we computed Domestic Mobility In-Degree,
Foreign Mobility In-Degree, Same-Status Mobility
In-Degree, Higher-Status Mobility In-Degree and
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1241
Lower-Status Mobility In-Degree. We also included
season and country xed effects.
Analytical approach
Despite including these control variables, our anal-
yses may still be affected by endogeneity due to
omitted variable bias or reverse causality affecting
our mobility in-/out-degree measures. For example,
individuals from different companies might be
drawn to join the focal house if its creative director
is known to be easy to work with. Having such a
director at the helm of a house can also increase
creative performance because people surrounding
this person are more creative. The absence of an
“easy/hard to work with” variable or other unob-
servables linked to the personality of the current
creative director that correlate both with personnel
mobility variables and creative performance in the
models could inate the standard errors for our
theoretical variables or alter the signs of these vari-
ables’ coefcients. Reverse causality may also be at
play: when a rm performs well, its employees are
likely to be in demand by higher-status competitors;
yet when a rm performs poorly, it may lay off some
of its employees, who will most likely end up with
lower-status competitors. Ordinary least squares
(OLS) estimation is likely to provide biased results
in our context.
A typical approach to control for endogeneity is
to conduct two-stage regression analyses whereby
one instruments an endogenous regressor with vari-
ables that affect the dependent variable’s error term
only through the instrumental variable’s effect on
the endogenous regressor. The problem is that we
have multiple potentially endogenous regressors
(i.e., each of the mobility out- and in-degrees).
The only framework that realistically allows us to
instrument this high number of variables at the
same time is the Arellano-Bover/Blundell-Bond
(AB/BB) two-step system generalized method of
moments (GMM) estimator (Arellano and Bond,
1991; Arellano and Bover, 1995; Blundell and
Bond, 1998).
ANALYSIS AND RESULTS
Table 1 reports descriptive statistics and correlation
coefcients for our variables. Our initial sample
contained 2,427 house-season observations, but due
to the need to accommodate lags of the variables
which we needed to instrument, we performed our
regressions on the remaining sample of 2,023 obser-
vations. Table 2 contains our regression results.
Using the Arellano-Bond family of estimators
requires that the model’s error terms be rst-order
serially correlated (as evidenced by the signicance
of the AR(1) test), yet not second-order correlated
(as evidenced by the lack of signicance for the
AR(2) test). In addition, a nonsignicant Hansen
test (as well as difference-in-Hansen test) of overi-
dentifying restrictions is required to further suggest
that the specic lagged values are valid instruments
(Milanov and Shepherd, 2013; Suarez, Cusumano,
and Kahl, 2013) and that the chosen lag structure
is appropriate (Roodman, 2009). In each model,
we report tests for serial correlation—AR(1) and
AR(2), the model’s overall Hansen test to eval-
uate joint validity of the full instrument set,3
difference-in-Hansen tests to evaluate the qual-
ity of variables’ lags as instruments for the lev-
els equation in the system GMM, as well as the
difference-in-Hansen tests to evaluate the qual-
ity of the lags of dependent variable as instru-
ments. Robust standard errors (Windmeijer, 2005)
are reported for each model. These are needed in
this context to accommodate the fact that while a
two-step estimator is more efcient than a one-step
estimator, the former produces articially low stan-
dard errors. We also report the number of instru-
ments as well as the number of panels (i.e., fashion
houses).
Our baseline is Model 1 that only includes
controls. We introduce a linear and quadratic
term of Foreign Mobility Out-Degree in Model
2. The positive coefcient on the linear term
(𝛽=0.40, SE =0.16, p=0.01, CI4[0.09; 0.72])
and negative coefcient on the squared term
(𝛽=-0.06, SE =0.02, p =0.01, CI [-0.10; -0.01])
3In AB/BB estimation, the number of instruments is inuenced
by the number of lags of the endogenous variables, the lagged
values used to instrument the lagged dependent variable, as well as
the exogenous control variables. The appropriate lag structure can
vary for each model; the best one results in the signicant AR(1)
test, not signicant AR(2) test, not signicant overall Hansen tests,
as well as not signicant difference-in-Hansen tests for the subsets
of instruments based on the lagged dependent variable and that
based on other lags (Milanov et al., 2013; Suarez et al., 2013).
At the same time, Hansen tests cannot have the perfect values of
p=1, which is an indication that the model may have too many
instruments to be capable of expunging endogeneity, nor should
the test values be substantially less than 0.25, which indicates a
high risk of instrumental variables correlating with the dependent
variable’s error term (Roodman, 2009).
4We report 95percent condence intervals (CIs) in this article.
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1242 A. Shipilov, F. C. Godart, and J. Clement
Table 1. Descriptive statistics and correlations
Variable 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
1Creative Performance 1.00
2Alumni Network in Industry 0.24 1.00
3Star Creative Director 0.17 0.38 1.00
4Alumni Network in Destination House 0.05 0.21 0.10 1.00
5Designers Lost to French Houses 0.25 0.44 0.21 0.29 1.00
6Designers Lost to Italian Houses 0.13 0.34 0.18 0.42 0.20 1.00
7House’s Number of Awards 0.21 0.23 0.45 −0.03 0.18 0.14 1.00
8House Age 0.20 0.39 0.20 0.09 0.33 0.15 −0.06 1.00
9Recent Star Hire 0.02 0.04 0.30 0.03 0.04 0.02 0.11 0.04 1.00
10 Recent Star Loss −0.02 0.03 −0.06 −0.03 0.06 -0.02 0.00 −0.01 −0.03 1.00
11 Number of Lost Stars 0.24 0.33 0.36 0.27 0.43 0.37 0.14 0.23 0.02 0.03 1.00
12 Number of Hired Stars 0.06 0.30 0.43 0.12 0.29 0.19 0.10 0.21 0.05 0.03 0.41 1.00
13 Media Coverage (log) 0.42 0.48 0.38 0.14 0.36 0.24 0.30 0.45 0.08 0.00 0.28 0.22 1.00
14 House Status 0.33 0.32 0.31 0.03 0.26 0.14 0.25 0.38 0.06 −0.03 0.22 0.26 0.61 1.00
15 Age of Creative Director 0.08 0.24 0.26 0.13 0.07 0.15 0.02 0.25 −0.04 −0.11 0.12 0.13 0.38 0.26 1.00
16 Tenure of Creative Director 0.16 0.10 0.21 0.12 0.04 0.22 0.10 0.12 −0.06 −0.18 0.18 −0.06 0.27 0.15 0.72
17 Creative Director’s Number of Houses 0.13 0.28 0.27 0.20 0.33 0.30 0.10 0.32 −0.01 0.03 0.38 0.35 0.32 0.36 0.27
18 Team of Creative Directors −0.10 0.04 −0.02 0.01 −0.01 −0.03 0.01 0.00 0.04 0.00 −0.09 −0.09 −0.04 −0.01 −0.07
19 Creative Director’s Fashion Education 0.09 −0.03 0.06 −0.10 0.00 −0.07 0.09 −0.08 0.04 0.05 0.05 0.04 −0.06 −0.01 −0.37
20 Number of Designers (log) 0.11 0.38 0.25 0.17 0.35 0.27 0.22 0.22 0.08 −0.01 0.18 0.09 0.29 0.14 0.07
21 Domestic Mobility In-Degree 0.00 0.28 0.22 0.03 0.25 0.14 0.14 0.15 0.02 0.08 0.03 0.39 0.20 0.15 0.10
22 Foreign Mobility In-Degree 0.11 0.40 0.20 0.06 0.49 0.29 0.21 0.27 0.09 0.11 0.27 0.36 0.23 0.23 −0.10
23 Higher-Status Mobility In-Degree −0.07 0.05 0.12 −0.03 0.11 −0.02 0.15 −0.10 0.09 0.13 0.03 0.19 0.07 0.15 −0.07
24 Same-Status Mobility In-Degree 0.01 0.22 0.11 −0.03 0.29 0.03 0.11 0.13 -0.01 0.15 0.06 0.35 0.13 0.14 0.00
25 Lower-Status Mobility In-Degree 0.10 0.38 0.21 0.00 0.33 0.19 0.05 0.29 0.16 0.08 0.25 0.31 0.33 0.31 −0.01
26 Foreign Mobility Out-Degree 0.26 0.53 0.26 0.19 0.64 0.46 0.31 0.25 0.07 0.06 0.39 0.35 0.33 0.26 0.01
27 Domestic Mobility Out-Degree 0.24 0.52 0.39 0.28 0.46 0.41 0.30 0.24 −0.02 −0.02 0.46 0.24 0.42 0.25 0.31
28 Higher-Status Mobility Out-Degree 0.06 0.27 0.10 −0.05 0.39 0.17 0.16 0.06 0.05 0.07 0.09 0.13 0.25 0.12 −0.01
29 Same-Status Mobility Out-Degree 0.09 0.44 0.18 0.12 0.43 0.15 0.17 0.20 −0.01 0.05 0.34 0.21 0.23 0.15 0.09
30 Lower-Status Mobility Out-Degree 0.25 0.40 0.34 0.11 0.35 0.26 0.24 0.25 0.07 0.04 0.40 0.36 0.42 0.39 0.10
Mean 1.24 0.40 0.35 0.03 0.29 0.25 0.29 26.72 0.03 0.03 0.18 0.17 4.40 13.17 43.32
S.D. 2.39 0.49 0.50 0.18 0.65 0.62 0.67 30.94 0.18 0.17 0.49 0.44 1.42 8.82 11.67
Min 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.69 0.00 22.00
Max 12.57 1.00 2.00 1.00 3.00 4.00 5.00 172.00 1.00 1.00 4.00 2.00 7.98 50.34 79.00
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1243
Table 1. continued
Variable 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
16 Tenure of Creative Director 1.00
17 Creative Director’s Number of Houses 0.10 1.00
18 Team of Creative Directors −0.11 −0.16 1.00
19 Creative Director’s Fashion Education −0.22 0.04 −0.03 1.00
20 Number of Designers (log) 0.04 0.03 0.46 0.00 1.00
21 Domestic Mobility In-Degree −0.12 0.22 0.05 0.03 0.28 1.00
22 Foreign Mobility In-Degree −0.18 0.29 0.00 0.11 0.32 0.10 1.00
23 Higher-Status Mobility In-Degree −0.13 0.17 0.00 0.10 0.02 0.18 0.32 1.00
24 Same-Status Mobility In-Degree −0.25 0.21 0.03 0.04 0.14 0.45 0.46 0.09 1.00
25 Lower-Status Mobility In-Degree −0.09 0.15 0.09 0.12 0.30 0.14 0.62 0.09 0.24 1.00
26 Foreign Mobility Out-Degree −0.03 0.27 0.00 0.09 0.36 0.20 0.72 0.12 0.36 0.49 1.00
27 Domestic Mobility Out-Degree 0.34 0.29 −0.03 0.02 0.30 0.33 0.21 0.06 0.09 0.23 0.31 1.00
28 Higher-Status Mobility Out-Degree −0.03 0.09 −0.04 0.03 0.13 0.10 0.43 0.31 0.32 0.27 0.46 0.19 1.00
29 Same-Status Mobility Out-Degree 0.05 0.24 −0.01 0.02 0.13 0.11 0.36 0.06 0.40 0.23 0.48 0.34 0.19 1.00
30 Lower-Status Mobility Out-Degree 0.09 0.23 −0.02 0.09 0.26 0.15 0.41 0.09 0.17 0.59 0.53 0.46 0.22 0.20 1
Mean 11.27 3.03 0.20 0.65 0.88 0.50 0.63 0.17 0.39 0.20 0.64 0.47 0.13 0.26 0.33
St. Dev. 9.91 2.16 0.40 0.46 0.27 0.86 1.14 0.47 0.78 0.60 1.17 0.89 0.41 0.57 0.84
Min 0.00 1.00 0.00 0.00 0.69 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Max 53.00 13.00 1.00 1.00 2.08 7.00 6.00 3.00 7.00 5.00 7.00 7.00 3.00 4.00 6.00
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1244 A. Shipilov, F. C. Godart, and J. Clement
Table 2. Regression results
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9
coef SE coef SE coef SE coef SE coef SE coef SE coef SE coef SE coef SE
Season Fixed Effects Yes Yes Yes Yes Ye s Yes Yes Ye s Yes
Country Fixed Effects Ye s Yes Yes Yes Ye s Yes Yes Ye s Yes
Lagged Dependent Variable
(Creative Performance)
1.05 (0.12) 0.96 (0.08) 1.08 (0.10) 0.94 (0.10) 0.94 (0.10) 1.02 (0.10) 0.96 (0.10) 0.87 (0.08)
House Status −0.01 (0.01) −0.00 (0.01) 0.02 (0.02) −0.01 (0.02) 0.00 (0.02) −0.01 (0.02) −0.01 (0.01) 0.01 (0.01)
Number of Hired Stars 0.18 (0.14) 0.02 (0.10) 0.16 (0.16) −0.02 (0.18) −0.00 (0.17) 0.13 (0.16) 0.06 (0.13) −0.21 (0.27)
Number of Lost Stars −0.15 (0.12) −0.11 (0.08) 0.02 (0.12) 0.06 (0.18) 0.06 (0.17) 0.12 (0.15) 0.09 (0.11) −0.66 (0.33)
Recent Star Loss 0.35 (0.16) 0.23 (0.11) 0.02 (0.15) 0.07 (0.17) 0.02 (0.14) −0.14 (0.20) −0.15 (0.14) 0.04 (0.17)
Recent Star Hire 0.35 (0.19) 0.09 (0.12) 0.34 (0.16) 0.11 (0.14) 0.14 (0.14) 0.08 (0.13) 0.12 (0.14) −0.35 (0.28)
Star Creative Director −0.38 (0.23) −0.14 (0.12) −0.58 (0.20) −0.15 (0.20) −0.29 (0.24) −0.26 (0.14) −0.24 (0.13) −0.06 (0.19) 0.50 (0.31)
House’s Number of Awards −0.11 (0.20) 0.08 (0.15) −0.25 (0.25) −0.05 (0.18) −0.08 (0.16) −0.22 (0.21) −0.15 (0.19) 0.17 (0.20)
Alumni Network in Destination
House
0.41 (0.20) 0.43 (0.26) 0.09 (0.26) 0.09 (0.35) 0.03 (0.31) −0.08 (0.33) −0.18 (0.28) 0.01 (0.46)
Alumni Network in Industry −0.16 (0.24) −0.11 (0.16) −0.81 (0.23) −0.29 (0.27) −0.38 (0.31) −0.55 (0.30) −0.30 (0.20) 0.20 (0.26)
Designers Lost to Italian Houses −0.25 (0.09) −0.27 (0.10) −0.12 (0.16) −0.14 (0.21) −0.16 (0.16) −0.04 (0.19) 0.10 (0.13) −0.56 (0.28)
Designers Lost to French
Houses
−0.29 (0.10) −0.29 (0.14) −0.21 (0.21) −0.17 (0.29) −0.13 (0.25) 0.02 (0.25) 0.14 (0.16) −0.66 (0.18)
Media Coverage (log) 0.07 (0.12) −0.07 (0.11) 0.19 (0.24) 0.02 (0.12) 0.04 (0.15) 0.07 (0.14) 0.15 (0.12) 0.33 (0.13)
House Age 0.01 (0.01) 0.01 (0.01) −0.01 (0.01) −0.00 (0.01) 0.00 (0.01) −0.01 (0.01) −0.01 (0.01) −0.01 (0.00)
Age of Creative Director −0.02 (0.02) −0.02 (0.01) 0.04 (0.02) −0.01 (0.03) −0.01 (0.03) 0.01 (0.02) 0.01 (0.02) 0.00 (0.02)
Tenure of Creative Director 0.05 (0.02) 0.05 (0.02) −0.05 (0.02) 0.00 (0.03) 0.00 (0.03) −0.02 (0.03) −0.03 (0.02) 0.00 (0.02)
Creative Director’s Number of
Houses
0.01 (0.03) −0.00 (0.03) −0.02 (0.03) −0.02 (0.03) −0.03 (0.03) −0.02 (0.03) −0.04 (0.02) −0.01 (0.08) 0.14 (0.11)
Team of Creative Directors −0.58 (0.28) −0.29 (0.21) 0.13 (0.34) 0.05 (0.29) −0.03 (0.24) 0.27 (0.29) 0.12 (0.20) 0.05 (0.20)
Number of Designers (log) −0.25 (0.39) −0.47 (0.31) 0.76 (0.50) −0.35 (0.51) −0.21 (0.58) −0.13 (0.47) 0.02 (0.28) 0.52 (0.37)
Creative Director’s Fashion
Education
−0.49 (0.44) −0.31 (0.39) 0.97 (0.79) 0.07 (0.84) 0.15 (0.88) 0.32 (0.54) 0.42 (0.33) 0.04 (0.37)
Domestic Mobility In-Degree 0.15 (0.08) 0.10 (0.07) 0.24 (0.11) 0.19 (0.13) 0.17 (0.14) 0.29 (0.10) 0.20 (0.10) −0.02 (0.06) −0.31 (0.15)
Foreign Mobility In-Degree 0.40 (0.20) 0.14 (0.09) 0.06 (0.19) 0.16 (0.15) 0.07 (0.20) 0.32 (0.19) 0.24 (0.13) 0.01 (0.09) −0.53 (0.22)
Higher-Status Mobility
In-Degree
−0.42 (0.26) −0.04 (0.15) −0.49 (0.24) −0.08 (0.23) −0.01 (0.25) −0.18 (0.24) 0.02 (0.19) 0.36 (0.15)
Same-Status Mobility In-Degree −0.11 (0.11) 0.01 (0.09) −0.30 (0.09) −0.07 (0.11) −0.06 (0.14) −0.31 (0.18) −0.19 (0.11) 0.00 (0.10) 0.49 (0.19)
Lower-Status Mobility
In-Degree
−0.06 (0.11) 0.10 (0.09) −0.10 (0.12) −0.07 (0.15) −0.05 (0.13) −0.25 (0.18) −0.32 (0.15) −0.03 (0.10) 0.49 (0.15)
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1245
Table 2. continued
Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8 Model 9
coef SE coef SE coef SE coef SE coef SE coef SE coef SE coef SE coef SE
Foreign Mobility Out-Degree 0.40 (0.16) 0.55 (0.15) 0.58 (0.26) 0.66 (0.28) 0.57 (0.24) 0.40 (0.18) 0.32 (0.16) 0.61 (0.21)
Foreign Mobility Out-Degree
(Squared)
−0.06 (0.02) −0.09 (0.03) −0.08 (0.04) −0.08 (0.04) −0.09 (0.03) −0.08 (0.03) −0.06 (0.03) −0.03 (0.03)
Domestic Mobility Out-Degree 0.40 (0.30) 0.18 (0.24) 0.24 (0.19) 0.12 (0.20) 0.02 (0.13) 0.10 (0.10) 0.30 (0.14)
Domestic Mobility Out-Degree
(Squared)
−0.05 (0.05)
Higher-Status Mobility
Out-Degree
−0.70 (0.53) −0.49 (0.35) −0.55 (0.33) −0.57 (0.28) −0.33 (0.19) 0.00 (0.21)
Higher-Status Mobility
Out-Degree (Squared)
0.18 (0.22)
Same-Status Mobility
Out-Degree
−0.37 (0.37) −0.35 (0.18) −0.22 (0.16) −0.23 (0.12) −0.21 (0.13) −0.24 (0.19)
Same-Status Mobility
Out-Degree (Squared)
0.04 (0.15)
Lower-Status Mobility
Out-Degree
0.06 (0.11) −0.10 (0.14) −0.13 (0.13) −0.44 (0.20)
Lower-Status Mobility
Out-Degree (Squared)
0.07 (0.03) 0.06 (0.03) 0.08 (0.06)
Observations 2,023 2,023 2,023 2,023 2,023 2,023 2,023 2,023 2,023
Number of houses 261 261 261 261 261 261 261 261 261
Number of instruments 94 106 84 92 85 87 96 80
AB test for AR(1) in rst
differences (p-value) 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
AB test for AR(2) in rst
differences (p-value) 0.93 0.83 0.99 0.88 0.88 0.75 0.72 0.75
Overall Hansen test (p-value) 0.84 0.60 0.93 0.91 0.93 0.87 0.90 0.84
Difference-in-Hansen test for
exogeneity
of GMM instruments for levels
(p-value) 0.50 0.49 0.53 0.50 0.52 0.41 0.41 0.25
Difference-in-Hansen test for
exogeneity of
Lagged dependent variable
(p-value) 0.83 0.33 0.53 0.74 0.61 0.74 0.58 0.64
Robust standard errors in parentheses; p-values can be obtained by dividing coefcients by standard errors.
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1246 A. Shipilov, F. C. Godart, and J. Clement
are suggestive of the inverted-U-shaped function
we hypothesize (H1). In Model 3, we enter a
linear and quadratic term of Domestic Mobility
Out-Degree for comparison.The positive linear
(𝛽=0.55, SE =0.15, p=0.00, CI[0.25; 0.85]) and
negative quadratic (𝛽=−0.09, SE =0.03, p=0.00,
CI[−0.14; −0.04]) terms of Foreign Mobility
Out-Degree are signicant, whereas the effects of
Domestic Mobility Out-Degree are not signicant.
Following Bothner et al. (2012), we further
wanted to probe whether the relationship between
Foreign Mobility Out-Degree and Creative Per-
formance is indeed inverted-U-shaped or rather
characteristic of a “diminishing returns” func-
tion whereby the effect is initially positive, then
diminishes, yet never turns negative. To this end,
we estimated Model 3bis (reported in Table 3) in
which we replaced Foreign Mobility Out-Degree
with dummies that take a value of one for different
values of this variable. The reference category is the
dummy Foreign Mobility Out-Degree=4, which
is the peak of the effect. As Model 3bis illustrates,
the dummies below four have negative and mostly
signicant effects, whereas the dummies above four
do not. The results of these intermediary models
therefore suggest partial support for Hypothesis
1: the relationship between Foreign Mobility
Out-Degree and Creative Performance seems to
be a “diminishing returns” concave function rather
than a true inverted-U.
To test Hypotheses 2 and 3, we now return to
Table 2 where we enter linear and squared terms
of Higher Status-Mobility Out-Degree as well
as Same Status-Mobility Out-Degree in Model
4. In Model 5, we drop the quadratic effects of
Higher Status-Mobility Out-Degree as well as
that of the Same Status-Mobility Out-Degree as
they are not signicant in Model 4.We then add a
linear term of Lower-Status Mobility Out-Degree
in Model 6. As this term is not signicant, we
experiment with adding a quadratic term for
the Lower-Status Mobility Out-Degree in Model
7, whose effect turns out to be signicant and
positive. This suggests support for Hypothesis 3,
albeit with a slight twist: Lower-Status Mobility
Out-Degree seems to have a positive impact on
Creative Performance, with increasingly positive
returns. Interestingly, effects of mobility to higher
or same-status competitors both become signicant
only in Model 7 where we introduce quadratic
effects for Lower-Status Mobility Out-Degree and
both Higher Status-Mobility Out-Degree and Same
Table 3. Regression results
Model 3bis
Foreign Mobility Out-Degree=0−1.17 (0.53)
Foreign Mobility Out-Degree=1−0.68 (0.40)
Foreign Mobility Out-Degree=2−0.61 (0.32)
Foreign Mobility Out-Degree=3−0.56 (0.25)
Foreign Mobility Out-Degree=50.89 (0.77)
Foreign Mobility Out-Degree>5−0.49 (0.67)
Controls from Model 3 Yes
House’s country xed effects Yes
Season’s xed effects Yes
Number of houses 261
Number of instruments 133
AB test for AR(1) in rst differences
(p-value) 0
AB test for AR(2) in rst differences
(p-value) 0.84
Overall Hansen test (p-value) 0.78
Difference-in-Hansen test for
exogeneity of GMM instruments for
levels (p-value)
0.47
Difference-in-Hansen test for
exogeneity of lagged dependent
variable (p-value)
0.83
Observations 2,023
Robust standard errors in parentheses.
Status-Mobility Out-Degree. This can probably be
explained by the fact that models in which we don’t
properly account for all three kinds of mobility
within, as well as across, status boundaries are not
properly specied.
Model 7 is our full model. There is evidence
that Hypothesis 1 is supported with respect to
cross-country mobility: fashion houses benet
from losing designers to competitors from different
countries, but this effect satiates at high values of
mobility. This is evidenced by the positive linear
effect of Foreign Mobility Out-Degree (𝛽=0.40,
SE =0.18, p=0.03, CI [0.04; 0.75]) and a neg-
ative quadratic effect of this variable (𝛽=-0.08,
SE =0.03, p=0.00, CI[-0.13; -0.03]). We did
not observe a signicant effect of within-country
mobility. Hypothesis 2 is not supported: the full
model shows that fashion houses uniformly suffer
from losing designers to competitors of higher sta-
tus (Higher-Status Mobility Out-Degree 𝛽=-0.57,
SE =0.28, p=0.04, CI [-1.12; -0.03]). Hypothesis
3 seems to be supported: there is a positive effect of
losing designers to many lower-status competitors,
and this effect accelerates at an increasing rate
as evidenced by the positive and signicant term
of Lower-Status Mobility Out-Degree Squared
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1247
(𝛽=0.07, SE =0.03, p=0.05, CI [0.00; 0.13]).
Even though the linear effect of this variable is neg-
ative in Model 7 (𝛽=-0.10, SE =0.14, p=0.46, CI
[-0.37; 0.17]), the wide condence interval suggests
that there is no systematic reduction of creative
performance when the fashion house starts to lose
key employees to lower-status competitors. While
we expected no effect of same-status mobility on
creative performance, Model 7 shows that this
effect is actually negative (Same-Status Mobility
Out-Degree 𝛽=-0.23, SE =0.12, p=0.04, CI
[-0.46; -0.01]).
Online Figure 2 in File S1 plots the effects across
the observable range of out-degree on creative per-
formance based on destination of personnel mobil-
ity. It is available in Appendix 3 in File S1.
We also examined the economic effects of
changes in each variable by examining their impact
on the number of creativity points awarded to
each house when the variable changed from its
mean to two standard deviations above its mean.
Such change in the Foreign Mobility Out-Degree
increases the average number of creativity points
awarded by any buyer to the house’s collection by
0.27. Given that the average buyer gives 1.24 points
to an average collection, 0.27 points represents
a 21.5 percent increase in a buyer’s evaluation.
The same change in the Higher-Status Mobility
Out-Degree leads to a loss of 0.47 points, which
represents a 37.69 percent reduction from the
average. The same change in Same-Status Mobility
Out-Degree leads to a loss of 0.26 points, which
represents a 21.15 percent reduction from the aver-
age. Finally, a two standard deviation increase in
Lower-Status Mobility Out-Degree translates into a
gain of 0.28 creativity points, which translates to a
22.19 percent increase relative to average number
of points.
In Model 8, we dropped nonsignicant controls
from Model 7 to see whether they affected our
results. This check is helpful because a high
number of nonsignicant control variables also
means a high number of instruments in a system
GMM AB/BB model, which beyond a certain point
may prevent the model from expunging endo-
geneity. The effects are similar to those reported
in Model 7.
Finally, in Model 9 we show the results without
deploying the AB/BB estimator. This model has
rm, country, and season xed effects implemented
using OLS. This model shows biased estimates.
From Model 9 one would conclude, for example,
that losing employees to higher-status competitors
has a positive (although not signicant) effect
and that personnel loss to lower-status competi-
tors seems to have a signicant negative effect.
However, OLS does not allow us to control for
reverse causality, among other things. That is, how
well a rm performs or is expected to perform
creatively may impact its employees’ mobility to
competitors of different status by affecting how
attractive they are to these competitors. If a rm
is known to perform well or is expected to do well
in the near future, many higher-status competitors
may poach its employees. Without accounting for
reverse causality, we are likely to observe a positive
association between performance and mobility to
higher-status competitors, as in Model 9, that hides
the true (negative) effect of this type of mobility
on the focal rm’s performance. The reverse is also
true. That is, a rm’s poor performance (or expected
performance) may both create a need to let go of
some employees and limit job opportunities for
these employees: those from a poorly performing
rm are likely to end up working for lower-status
competitors. This is likely to create an articial
negative association between performance and
mobility to lower-status competitors, as in Model
9, which hides the true (positive) effect of this type
of mobility.
DISCUSSION AND CONCLUSIONS
While we already had evidence that employee
mobility builds potentially valuable boundary-
spanning ties across organizations long after
employees have left, there was little evidence on
the conditions under which companies can benet
from employee departures. Available research
acknowledged that these departures can potentially
offer rms access to external knowledge, yet it
had assumed that the knowledge accessed did
not depend on the type of boundary spanned by
departed employees (Corredoira and Rosenkopf,
2010; Dokko and Rosenkopf, 2010; Godart et al.,
2014).
We knew that when employees span formal orga-
nizational boundaries, they may provide their for-
mer employer with access to external knowledge at
the risk of idea leakage to competition. However,
considering only formal boundary spanning is not
sufcient: the current study suggests that the cre-
ative performance of organizations is also affected
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1248 A. Shipilov, F. C. Godart, and J. Clement
by the extent to which employee departures span
geographic or status boundaries. By investigating
these boundaries, we found that the aggregate
relationship of creative performance with outward
mobility consists of several linear and nonmono-
tonic relationships with the number of employees
who spanned status or geographic boundaries when
leaving their rm. The results show that our focus on
geographic and status boundaries is justied by the
fact that employee mobility across or within these
boundaries differentially affects the balance of costs
and benets of outward mobility for the rm’s cre-
ative performance.
Even though we originally expected that high
employee loss to different foreign competi-
tors would be detrimental to the organization’s
performance, due to opportunities that foreign
competitors have to learn from the rm’s former
collaborators, our results suggest that this is not the
case. Ultimately, even when foreign competitors
learn from the focal house’s alumni and imitate
its practices, the resulting competing products
do not undermine the perception of creativity of
the products made by the focal fashion house. In
other words, following Italian designers’ moves
to Japan or France, buyers will not downgrade
their perceptions of the creativity of Italian fashion
collections even when Japanese or French design-
ers incorporate Italian ideas into their respective
collections. By contrast, we saw neither positive
nor negative effects of employee mobility to
domestic competitors: while such mobility did not
yield signicant benets as it was not associated
with spanning a geographic boundary, there were
no signicant costs of imitation of the focal rm’s
products by the same country competitors’ either.
The most counterintuitive results came from
the analysis of mobility across status groups.
Unexpectedly, we found that companies seemed
to suffer when their employees move to work for
higher-status competitors. Since we controlled for
the quality of mobile individuals, these negative
effects seem to be driven by the higher-status
competitors taking credit for the organization’s
ideas that are carried by its former employees. We
saw similar effects involving the loss of individ-
uals to same-status competitors. Thus, imitation
of the organization’s ideas by even same-status
competitors seems detrimental to its creative perfor-
mance, as the audience becomes unsure as to which
organization these ideas should be attributed to.
This means that organizations should do their best to
limit the loss of employees to same or higher-status
competitors. At the same time, they probably should
not be concerned when their employees move to
many lower-status competitors: the benets from
this type of mobility actually increase at an increas-
ing rate as the higher-status organization attracts
more attention from the audiences to nonconformist
ideas borrowed from lower-status competition.
Our study makes several contributions. First,
we enrich the growing literature that examines
how social networks resulting from the external
boundary-spanning ties of employees affect organi-
zational creativity (Mawdsley and Somaya, 2016).
Godart et al. (2014) assumed that organizations
benet from the loss of personnel because the
resulting bridging ties expose them to diverse
knowledge. In this article we explicitly challenge
this assumption and show that companies are most
likely to benet from diverse knowledge received
from competitors in foreign countries and those
of lower status. Unexpectedly, we also nd that
the benets of receiving diverse knowledge from
competitors of same or higher status are actually
lower than the costs. Thus, it is not just the mere
diversity of knowledge resulting from outward
mobility that matters for the creative performance
of organizations, but rather what the sources of
this diverse knowledge are (i.e., do the ideas come
from domestic or foreign competitors? Do they
come from higher- or lower-status competitors?)
and whether the costs of access to this diverse
knowledge are lower than the benets. While
research has already shown the benets of individ-
ual diversity, our article (and especially Hypothesis
1) suggests that the benets of exposure to diversity
also occur at a different level of aggregation. That
is, companies as a whole seem to benet from the
diverse experiences of their departed employees.
Second, we contribute to research on external
boundary spanning more generally. These studies
can especially benet from the notion of status
boundaries. For example, Rosenkopf and Nerkar
(2001) have found that exploration efforts that
span organizational boundaries produce more
impactful innovations if they also involve spanning
technological boundaries. Furthermore, research
on technological boundary spanning has demon-
strated that geography affects the localization of
knowledge (Almeida and Kogut, 1999) and that
employee mobility facilitates knowledge ows
across geographies (Oettl and Agrawal, 2008).
Our ndings suggest that in some industries,
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1249
the relevant boundaries may not only be based
on technology or geography, but also on status.
Specically, membership in status groups can
become a source of novel ideas if an organization’s
employees can span boundaries to lower-status
competitors and a source of signicant risks if an
organization’s employees span the boundaries to
higher-status competitors.
Third, we contribute to the literature examining
the link between foreign experiences and creativ-
ity (e.g., Leung and Chiu, 2010; Leung et al., 2008;
Maddux and Galinsky, 2009; Tadmor, Galinsky, and
Maddux, 2012). This research has demonstrated
that individuals become more creative when they
are exposed to high requisite variety through con-
tact with foreign cultures and that companies can
enhance their creative performance by hiring such
individuals (Godart et al., 2015). The current study
shows that the foreign experience of employees may
become an organization-level resource and can also
help their former as opposed to current employ-
ers. This is important for organizations because they
can be exposed to requisite variety as a result of
their former employees’ moving abroad. To benet
from such forms of variety, organizations may bene-
t from instituting programs that allow alumni who
leave to work abroad to stay in touch with the orga-
nization’s current employees and feed insights back
into their creative processes. However, our ndings
also suggest that organizations do not need to stay
in touch with all departed employees; for instance,
those who went to work for competitors in the same
country are unlikely to expose an organization’s cur-
rent employees to signicantly novel ideas. This
nding conrms the arguments advanced in ear-
lier work on requisite variety: professional mobility
within the same country is not benecial to one’s
creativity (e.g., Leung et al., 2008; Maddux and
Galinsky, 2009; Tadmor et al., 2012).
The buyers who evaluated collections had a low
variance in their assessments, as evidenced by a
high Cronbach’s alpha. As they responded to the
question, it is highly probable that they evaluated
the collections through the prisms of novelty (“Does
the collection contain novel combinations?”) and
usefulness (“Will it inspire commercially successful
collections?”). The high level of agreement about
what is novel and useful can be attributed to the fact
that the world of high fashion focuses on just a hand-
ful of trends that change regularly, with each season.
Trends are seen by all market participants, including
designers and buyers. For example, gothic elements
in fashion were highly popular following the suc-
cess of the 1999 movie The Matrix, yet they became
relatively less popular afterwards. This has con-
sequences for the generalizability of this measure
to other contexts. For example, we do not believe
that creative performance in the hard sciences (e.g.,
physics, chemistry) is so massively focused on just
a handful of trends. Yet the evaluation of novelty
in science could also be affected by the competi-
tion between some evaluators and authors of ideas.
For example, Boudreau et al. (2016) nd that sci-
entists tend to discount the novelty of ideas when
these ideas are close to their domain of expertise.
In our case, buyers do not compete with designers
for ideas. This arguably makes the buyers’ eval-
uations of novelty and usefulness more impartial
than the evaluations of scientic ideas by scientists
working in the same domain. This too could be a
source of the high inter-rater reliability of our cre-
ativity scores. Ultimately, the more a particular eld
is subject to fads that lead most of the eld members
to reward a particular combination of creative ele-
ments, and the lower the noise associated with some
evaluators’ competing for novelty with those pro-
ducers whose work they are supposed to evaluate,
the more the creativity evaluations in that eld are
likely to exhibit high inter-rater reliability similar to
that reected in our dependent variable.
If supported by further research, our study has an
important practical implication. Even though com-
panies cannot control where their employees go,
they can decide when to make efforts to retain
employees, and with whom to keep in touch when
they do leave. Our results suggest that compa-
nies should not be overly concerned when some of
their employees leave for foreign or lower-status
competitors; such mobility may help the company
access external ideas and may come at little cost.
However, companies should help their remaining
employees stay in touch with these former cowork-
ers in order to get the maximum benet from
employee mobility. Our results also suggest that
companies should make efforts to keep employees
from moving to higher- or same-status competitors.
Specically, fashion houses can offer more artistic
freedom to designers who have offers from higher-
or same-status competitors. Some houses go as far
as allowing these designers to establish their own
labels within the focal house.
Several boundary conditions apply to our theory.
First, our arguments should apply in industries that
generate creative output— i.e., products or services
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1250 A. Shipilov, F. C. Godart, and J. Clement
that require recombination of diverse ideas. Second,
another assumption is that departed employees stay
in touch with their former colleagues and pay
attention to their activities in their new place of
employment. This is likely to be true when an indus-
try is characterized by important communities of
practice, so that professionals consider themselves
part of a broad professional community that goes
beyond their own organizations. Belonging to a
professional community tends to make individu-
als behave as though their work for companies is
not a “zero-sum game”; they expect to create more
novel ideas for their companies by sharing insights
with colleagues working for competitors. A con-
text where our arguments would probably not apply
is the mobility of athletes. For instance, Grohsjean
et al. (2016: 394) nd that NHL players who change
teams “experience a conict in their collective iden-
tity as they identify with both [teams] but can only
increase the welfare of one.” In that context, the vic-
tory of one team can only happen at the expense
of another team; hence it is a zero-sum game,
which we would not observe in fashion. Finally,
our theory applies to industries where status groups
and country boundaries generate a heterogeneity of
business-relevant ideas across rms. Yet, our more
general argument—that the effect of key employee
mobility depends on whether the employees cross
boundaries that expose them to new ideas— should
apply regardless of the specic boundary type.
In conclusion, our ndings suggest the central
role played by national borders and status groups
as delimiters of knowledge among different
communities of practice in creative industries.
National borders and status groups are examples
of boundaries that mobile employees could span.
If conrmed by future research, such boundaries
may inuence the ability of organizations to extract
creative performance from external boundary-
spanning ties that emerge through the career mobil-
ity of their employees across these communities.
By examining the variations in costs and benets of
employee outward mobility that involves crossing
status or geographic boundaries, we offer a new
theoretical lens that should help organizations
manage personnel loss in a positive manner.
REFERENCES
Aime F, Johnson S, Ridge JW, Hill AD. 2010. The rou-
tine may be stable but the advantage is not: competitive
implications of key employee mobility. Strategic Man-
agement Journal 31(1): 75–87.
Almeida P. 1996. Knowledge sourcing by foreign
multinationals: patent citation analysis in the US
semiconductor industry. Strategic Management
Journal 17(S2): 155– 165.
Almeida P, Kogut B. 1999. Localization of knowledge
and the mobility of engineers in regional networks.
Management Science 45(7): 905– 917.
Arellano M, Bond S. 1991. Some tests of specication for
panel data: Monte Carlo evidence and an application
to employment equations. Review of Economic Studies
58(2): 277–297.
Arellano M, Bover O. 1995. Another look at the instru-
mental variable estimation of error-components mod-
els. Journal of Econometrics 68(1): 29– 51.
Barkey K, Godart F. 2013. Empires, federated arrange-
ments, and kingdoms: using political models of gov-
ernance to understand rms’ creative performance.
Organization Studies 34(1): 79– 104.
Barkey K, Parikh S. 1991. Comparativeperspectives on the
state. Annual Review of Sociology 17: 523– 549.
Barrett M, Oborn E, Orlikowski WJ, Yates J. 2012.
Reconguring boundary relations: robotic innovations
in pharmacy work. Organization Science 23(5):
1448–1466.
Bell GG, Zaheer A. 2007. Geography, networks,
and knowledge ow. Organization Science 18(6):
955–972.
Bidwell M, Briscoe F. 2010. The dynamics of interor-
ganizational careers. Organization Science 21(5):
1034–1053.
Blundell R, Bond S. 1998. Initial conditions and moment
restrictions in dynamic panel data models. Journal of
Econometrics 87(1): 115–143.
Bothner MS, Kim Y-K, Smith EB. 2012. How does status
affect performance? Status as an asset vs status as
a liability in the PGA and NASCAR. Organization
Science 23(2): 416–433.
Boudreau KJ, Guinan EC, Lakhani KR, Riedl C. 2016.
Looking across and looking beyond the knowledge
frontier: intellectual distance and resource allocation
in science. Management Science 62(10): 2765– 2783.
DOI: 10.1287/mnsc.2015.2285.
Breward C. 2003. Fas hi on. Oxford University Press:
Oxford, UK; New York.
Cattani G, Ferriani S. 2008. A core/periphery perspective
on individual creative performance: social networks
and cinematic achievements in the Hollywood lm
industry. Organization Science 19(6): 824– 844.
Caves RE. 2000. Creative Industries: Contracts Between
Art and Commerce. Harvard University Press: Cam-
bridge, MA; London, UK.
Chadwick C, Dabu A. 2009. Human resources, human
resource management, and the competitive advantage
of rms: toward a more comprehensive model of causal
linkages. Organization Science 20(1): 253– 272.
Coff RW.1997. Human assets and management dilemmas:
coping with hazards on the road to resource-based the-
ory. Academy of Management Review 22(2): 374– 402.
Corredoira R, Rosenkopf L. 2010. Should auld acquain-
tance be forgot: the reverse transfer of knowledge
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
Mobility Networks 1251
through mobility ties. Strategic Management Journal
31(2): 159–181.
Crane D. 1997. Globalization, organizational size, and
innovation in the French luxury fashion industry: pro-
duction of culture theory revisited. Poetics 24(6):
393–414.
Dokko G, Rosenkopf L. 2010. Social capital for hire?
Mobility of technical professionals and rm inuence
in wireless standards committees. Organization Science
21(3): 677–695.
Elias N, Jephcott E. 1982. The Civilizing Process, State
Formation and Civilization (Vol. II). Blackwell:
Oxford, UK.
Godart F. 2012. Unveiling Fashion: Business, Culture,
and Identity in the Most Glamorous Industry. Palgrave
Macmillan: Hampshire, UK.
Godart F, Maddux WW, Shipilov AV, Galinsky AD.
2015. Fashion with a foreign air: professional expe-
riences abroad facilitate the creative innovations of
organizations. Academy of Management Journal 58(1):
195–220.
Godart F, Mears A. 2009. How do cultural producers make
creative decisions? Lessons from the catwalk. Social
Forces 88(2): 671–692.
Godart FC, Shipilov AV, Claes K. 2014. Making the most
of the revolving door: the impact of outward personnel
mobility networks on organizational creativity. Organi-
zation Science 25(2): 377–400.
Grant AM, Schwartz B. 2011. Too much of a good
thing: the challenge and opportunity of the inverted U.
Perspectives on Psychological Science 6(1): 61– 76.
Grohsjean T, Kober P, Zucchini L. 2016. Coming back
to Edmonton: competing with former employers and
colleagues. Academy of Management Journal 59(2):
394–413.
Henderson R, Cockburn I. 1994. Measuring competence?
Exploring rm effects in pharmaceutical research.
Strategic Management Journal 15(S1): 63– 84.
Hofstede G. 1980. Culture’s Consequences: International
Differences in Work-Related Values. Sage Publications:
Newbury Park, CA.
James A. 2007. Everyday effects, practices and causal
mechanisms of ‘cultural embeddedness’: learning from
Utah’s high tech regional economy. Geoforum 38(2):
393–413.
Kawamura Y. 2005. Fashion-Ology: An Introduction to
Fashion Studies. Berg: New York.
Kawamura Y. 2011. Doing Research in Fashion and Dress:
An Introduction to Qualitative Methods. Berg: New
York.
Knoke D, Pappi FU, Broadbent J, Tsujinala Y. 1996.
Comparing Policy Networks: Labor Politics in the
US, Germany, and Japan. Cambridge University Press:
Cambridge, UK.
Kogut B, Zander U. 1992. Knowledge of the rm, com-
binative capabilities, and the replication of technology.
Organization Science 3(3): 383– 397.
Leung AK-Y, Chiu C-Y. 2010. Multicultural experiences,
idea receptiveness, and creativity. Journal of Cross
Cultural Psychology 41: 1– 19.
Leung AK-Y, Maddux WW, Galinsky AD, Chiu
C-Y. 2008. Multicultural experience enhances
creativity – the when and how. American Psychologist
63: 169–181.
Maddux WW, Galinsky AD. 2009. Cultural Borders and
mental barriers: the relationship between living abroad
and creativity. Journal of Personality and Social Psy-
chology 96(5): 1047– 1061.
March J. 1991. Exploration and exploitation in organiza-
tional learning. Organization Science 2(1): 71– 81.
Mawdsley JK, Somaya D. 2016. Employee mobility
and organizational outcomes: an integrative conceptual
framework and research agenda. Journal of Manage-
ment 42(1): 85–113.
McEvily B, Soda G, Tortoriello M. 2014. More formally:
rediscovering the missing link between formal orga-
nization and informal social structure. Academy of
Management Annals 8(1): 299– 345.
Mears A. 2011. Pricing Beauty: The Making of a Fashion
Model. University of California Press: Berkeley, CA.
Merton RK. 1968. The Matthew effect in science. Science
159(3810): 56–63.
Milanov H, Shepherd DA. 2013. The importance of the rst
relationship: the ongoing inuence of initial network
on future status. Strategic Management Journal 34(6):
727–750.
Oettl A, Agrawal A. 2008. International labor mobility and
knowledge ow externalities. Journal of International
Business Studies 39(8): 1242–1260.
Perry-Smith JE, Shalley CE. 2003. The social side of cre-
ativity: a static and dynamic social network perspective.
Academy of Management Review 28(1): 89–106.
Phillips DJ. 2002. A genealogical approach to organiza-
tional life chances: the parent-progeny transfer among
Silicon Valley law rms, 1946–1996. Administrative
Science Quarterly 47(3): 474–506.
Phillips DJ, Zuckerman E. 2001. Middle-status confor-
mity: theoretical restatement and empirical demonstra-
tion in two markets. American Journal of Sociology
107(2): 379–429.
Podolny JM. 1993. A status-based model of market compe-
tition. American Journal of Sociology 98(4): 829– 872.
Podolny JM. 2005. Status Signals: A Sociological Study
of Market Competition. Princeton University Press:
Princeton, NJ.
Price Alford H, Stegemeyer A. 2009. Who’s Who in
Fas hio n, (5th edn). New York: Fairchild Publications.
Rao H, Greve HR, Davis GF. 2001. Fool’s gold: social
proof in the initiation and abandonment of coverage by
Wall Street analysts. Administrative Science Quarterly
46(3): 502–526.
Roodman D. 2009. A note on the theme of too many
instruments. Oxford Bulletin of Economics and Statis-
tics 71(1): 135–158.
Rosenkopf L, Metiu A, George VP. 2001. From the bot-
tom up? Technical committee activity and alliance
formation. Administrative Science Quarterly 46(4):
748–772.
Rosenkopf L, Nerkar A. 2001. Beyond local search:
boundary-spanning, exploration, and impact in the opti-
cal disk industry. Strategic Management Journal 22:
287–306.
Saxenian A. 1996. Regional Advantage: Culture and Com-
petition in Silicon Valley and Route 128, (1st Harvard
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
1252 A. Shipilov, F. C. Godart, and J. Clement
University Press pbk. edn). Cambridge, MA: Harvard
University Press.
Schwartz SH. 1994. Are there universal aspects in the
content and structure of values? Journal of Social Issues
50(4): 19–45.
Sgourev SV. 2013. How Paris gave rise to Cubism (and
Picasso): ambiguity and fragmentation in radical inno-
vation. Organization Science 24(6): 1601–1617.
Somaya D, Williamson IO, Lorinkova N. 2008. Gone
but not lost: the different performance impacts of
employee mobility between cooperators versus com-
petitors. Academy of Management Journal 51(5):
936–953.
Suarez FF, Cusumano MA, Kahl SJ. 2013. Services and the
business models of product rms: an empirical analysis
of the software industry. Management Science 59(2):
420–435.
Tadmor CT, Galinsky AD, Maddux WW. 2012. Getting the
most out of living abroad: biculturalism and integrative
complexity as key drivers of creative and professional
success. Journal of Personality and Social Psychology
103(3): 520–542.
Tushman ML, Scanlan TJ. 1981. Boundary spanning indi-
viduals: their role in information transfer and their
antecedents. Academy of Management Journal 24(2):
289–305.
Uzzi B, Spiro J. 2005. Collaboration and creativity: the
small world problem. American Journal of Sociology
111(2): 447–504.
Vergani G. 2010. The Fashion Dictionary, (2nd edn).
Baldini Castoldi Dalai Editore: New York.
Williamson OE. 1981. The economics of organization:
the transaction cost approach. American Journal of
Sociology 87(3): 548– 577.
Windmeijer F. 2005. A nite sample correction for the
variance of linear efcient two-step GMM estimators.
Journal of Econometrics 126(1): 25– 51.
SUPPORTING INFORMATION
Additional supporting information may be found
in the online version of this article:
File S1. Online Supplement.
Copyright © 2016 John Wiley & Sons, Ltd. Strat. Mgmt. J.,38: 1232– 1252 (2017)
DOI: 10.1002/smj
