Uzzi 1997 Social Structure & Embeddedness

Social Structure and Competition in Interfirm Networks: The Paradox of Embeddedness
Author(s): Brian Uzzi

Source: Administrative Science Quarterly, Vol. 42, No. 1 (Mar., 1997), pp. 35-67
Published by: Sage Publications, Inc. on behalf of the Johnson Graduate School of Management,
Cornell University
Stable URL: http://www.jstor.org/stable/2393808 .
Accessed: 02/02/2015 12:05
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
Sage Publications, Inc. and Johnson Graduate School of Management, Cornell University are collaborating
with JSTOR to digitize, preserve and extend access to Administrative Science Quarterly.
http://www.jstor.org
This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

All use subject to JSTOR Terms and Conditions
Social Structureand The purpose of this work is to develop a systematic
Competitionin Interfirm understanding of embeddedness and organization
Networks: The Paradox networks. Drawing on ethnographic fieldwork conducted
at 23 entrepreneurial firms, I identify the components of
of Embeddedness embedded relationships and explicate the devices by
which embeddedness shapes organizational and eco-
Brian Uzzi nomic outcomes. The findings suggest that embedded-
Northwestern University ness is a logic of exchange that promotes economies of
time, integrative agreements, Pareto improvements in
allocative efficiency, and complex adaptation. These
positive effects rise up to a threshold, however, after
which embeddedness can derail economic performance
by making firms vulnerable to exogenous shocks or
insulating them from information that exists beyond their
network. A framework is proposed that explains how
these properties vary with the quality of social ties, the
structure of the organization network, and an organiza-
tion's structural position in the network.'
Research on embeddedness is an exciting area in sociology
and economics because it advances our understandingof
how social structureaffects economic life. Polanyi(1957)
used the concept of embeddedness to describe the social
structureof modern markets, while Schumpeter (1950) and
Granovetter(1985) revealed its robust effect on economic
action, particularlyin the context of interfirmnetworks,
stimulatingresearch on industrialdistricts (Leung, 1993;
Lazerson,1995), marketingchannels (Moorman,Zaltman,
and Deshponde, 1992), immigrantenterprise (Portes and
Sensenbrenner, 1993), entrepreneurship(Larson,1992),
lending relationships(Podolny,1994; Sterns and Mizruchi,
1993; Abolafia,1996), locationdecisions (Romo and
Schwartz, 1995), acquisitions(Palmeret al., 1995), and or-
ganizationaladaptation(Baumand Oliver,1992; Uzzi, 1996).
The notion that economic action is embedded in social
structurehas revived debates about the positive and
negative effects of social relationson economic behavior.
While most organizationtheorists hold that social structure
plays a significantrole in economic behavior,many economic
theorists maintainthat social relationsminimallyaffect
economic transactingor create inefficiencies by shieldingthe
transactionfrom the market(Peterson and Rajan,1994).
? 1997 by Cornell University. These conflictingviews indicatea need for more research on
0001-8392/97/4201-0035/$1 .00.
how social structurefacilitates or derails economic action. In
0 this regard,Granovetter's(1985) embeddedness argument
Financial assistance from the NSF (Grants has emerged as a potentialtheory for joiningeconomic and
SES-9200960 and SES-9348848), the
Sigma Xi Scientific Research Society, and
sociological approaches to organizationtheory. As presently
the Institute of Social Analysis-SUNY at developed, however, Granovetter'sargument usefully
Stony Brook made this research possible. explicates the differences between economic and sociologi-
Unpublished portions of this paper have
been awarded the 1991 American cal schemes of economic behaviorbut lacks its own con-
Sociological Association's James D. crete account of how social relationsaffect economic
Thompson Award, the 1993 Society for
the Advancement of Socio-Economics
exchange. The fundamentalstatement that economic action
Best Paper Prize, and the 1994 Academy is embedded in ongoing social ties that at times facilitate
of Management's Louis Pondy Disserta- and at times derailexchange suffers from a theoretical
tion Prize. I thank Jerry Davis, James
Gillespie, Mark Lazerson, Marika Lind-
indefiniteness. Thus, althoughembeddedness purportsto
holm, Willie Ocasio, Michael Schwartz, explainsome forms of economic action better than do pure
Frank Romo, Marc Ventresca, the ASQ economic accounts, its implicationsare indeterminate
editors and anonymous reviewers, and
especially Roberto Fernandez for helpful because of the imbalancebetween the relativelyspecific
comments on earlier drafts of this paper. propositionsof economic theories and the broadstatements
35/AdministrativeScience Quarterly,42 (1997): 35-67

about how social ties shape economic and collective action.
This work aims to develop one of perhaps multiplespecifica-
tions of embeddedness, a concept that has been used to
refer broadlyto the contingent natureof economic action
with respect to cognition, social structure,institutions,and
culture.Zukinand DiMaggio(1990) classified embeddedness
into four forms: structural,cognitive, political,and cultural.
The last three domains of embeddedness primarilyreflect
social constructionistperspectives on embeddedness,
whereas structuralembeddedness is principallyconcerned
with how the qualityand network architectureof material
exchange relationshipsinfluence economic activity. Inthis
paper, I limitmy analysis to the concept of structural
embeddedness.
THEPROBLEMOF EMBEDDEDNESSAND ECONOMIC
ACTION
Powell's (1990) analysis of the sociological and economic
literatureson exchange suggests that transactionscan take
place through loose collections of individualswho maintain
impersonaland constantly shifting exchange ties, as in
markets, or throughstable networks of exchange partners
who maintainclose social relationships.The key distinction
between these systems is the structureand qualityof
exchange ties, because these factors shape expectations
and opportunities.
The neoclassical formulationis often taken as the baseline
theory for the study of interfirmrelationshipsbecause it
embodies the core principlesof most economic approaches
(Wilson, 1989). Inthe ideal-typeatomistic market,exchange
partnersare linkedby arm's-lengthties. Self-interest moti-
vates action, and actors regularlyswitch to new buyers and
sellers to take advantage of new entrants or avoid depen-
dence. The exchange itself is limitedto price data, which
supposedly distillall the informationneeded to make
efficient decisions, especially when there are many buyers
and sellers or transactionsare nonspecific. Personal relation-
ships are cool and atomistic; if ongoing ties or implicit
contracts exist between parties, it is believed to be more a
matter of self-interested, profit-seekingbehaviorthan willful
commitment or altruisticattachment (Macneil,1978).
Accordingly,arm's-lengthties facilitateperformancebecause
firms disperse their business among many competitors,
widely sampling prices and avoidingsmall-numbersbargain-
ing situations that can entrapthem in inefficientrelationships
(Hirschman,1970). Althoughsome economists have recog-
nized that the conclusion that markets are efficient becomes
suspect when the idealizationof theoreticalcases is aban-
doned, they nonetheless have tended to regardthe idealized
model as giving a basicallycorrect view and have paid scant
attention to instances that diverge from the ideal (Krugman,
1986).
At the other end of the exchange continuumare embedded
relationships,and here a well-definedtheory of embedded-
ness and interfirmnetworks has yet to emerge. Instead,
findings from numerous empiricalstudies suggest that
embedded exchanges have several distinctivefeatures.
Research has shown that network relationshipsin the
36/ASQ, March 1997

Paradox of Embeddedness
Japanese auto and Italianknitwearindustriesare character-

ized by trust and personal ties, ratherthan explicitcontracts,
and that these features make expectations more predictable
and reduce monitoringcosts (Dore, 1983; Asanuma, 1985;
Smitka, 1991; Gerlach,1992). Helper(1990) found that close
supplier-manufacturer relationshipsin the auto industryare
distinctivefor their "thick"informationexchange of tacit and
proprietaryknow-how, while Larson(1992) and Lazerson
(1995) found that successful entrepreneurialbusiness
networks are typified by coordinationdevices that promote
knowledge transferand learning.Romo and Schwartz's
(1995) and Dore's (1983) findings concerningthe embedded-
ness of firms in regionalproductionnetworks suggest that
embedded actors satisfice ratherthan maximizeon price and
shift their focus from the narroweconomicallyrationalgoal
of winning immediate gain and exploitingdependency to
cultivatinglong-term,cooperativeties. The basic conjecture
of this literatureis that embeddedness creates economic
opportunitiesthat are difficultto replicatevia markets,
contracts, or verticalintegration.
To a limiteddegree, revisionisteconomic frameworkshave
attempted to explainthe above outcomes by redefining
embeddedness in terms of transactioncost, agency, or
game theory concepts. Liketheir neoclassical parent,
however, these schemes do not explicitlyrecognize or
model social structurebut, rather,apply conventional
economic constructs to organizationalbehavior,bypassing
the issues centralto organizationtheorists.1 Transactioncost
economics, for example, has usefully revised our under-
standing of when nonmarkettransactionswill arise, yet
because its focus is on dyadic relations,network dynamics
"are given short shrift"(Williamson,1994: 85). Transaction
cost economics also displays a bias toward describing
opportunisticratherthan cooperative relationsin its assump-
tion that, irrespectiveof the social relationshipbetween a
buyer and seller, if the transactiondegenerates into a
small-numbersbargainingsituation,then the buyer or seller
will opportunisticallysqueeze above-marketrents or shirk,
whichever is in his or her self-interest (Ghoshaland Moran,
1996).
Agency theory also focuses mainlyon self-interested human
nature,dyadic principal-agentties, and the use of formal
controls to explainexchange, ratherthan on an account of
embeddedness. Forexample, Larson's(1992) study of
interfirmexchange relationshipsrevealed agency theory's
limitedabilityto explainnetwork forms of organizationwhen
she showed that there is a lack of controland monitoring
devices between firms, that the roles of principaland agent
blurand shift, and that incentives are jointlyset. Similarly,
team theory is pressed to explain interfirmexchange
1 relationsbecause of its assumption that group members
My intent is not to critiquerevisionist have identicalinterests, an unrealisticassumption when
economic approachesor contrastall their formal rule structures (a hierarchy)do not exist or group
similaritiesand differenceswith network
perspectives (see Burt,1992; Zajacand members both cooperate and compete for resources, as in
Olsen, 1993; Ghoshaland Moran,1996). the case of manufacturer-supplier networks (Cyertand
Rather,I reviewthe mainpoints of
existingcritiquesto show that the March, 1992).
features of embeddedness cannot be
adequatelyexplainedby these ap- Game theory can accommodate N-person,network-like
proaches. structures, yet the core argument-that selfish playerswill
37/ASQ, March 1997

defect from cooperationwhen the endgame ensues even if
they have had on-going social ties and like each other well
(Jacksonand Wolinsky,1996)-fits poorlywith the empirical
regularitiesof networks. Padgett and Ansell (1993: 1308)
found in their network analysis of fifteenth-centuryMedici
tradingcompanies that "cleargoals of self-interest . . . are
not really features of people; they are . . . varying structures
of games." In cases in which game theory concedes
outcomes to social structure,it tends to do so after the fact,
to align predictionsand empiricalresults, but continues to
ignore sociological questions on the originof expectations,
why people interpretrules similarly,or why actors cooperate
when it contradictsself-interest (Kreps,1990).
Thus, while revisionisteconomic schemes advance our
understandingof the economic details of transacting,they
faintlyrecognize the influence of social structureon eco-
nomic life. Similarly,theory about the propertiesand process
by which embeddedness affects economic action remains
nascent in the organizationsliterature.Below, I reportresults
and formulatearguments that attempt to flesh out the
concept of embeddedness and its implicationsfor the
competitive advantage of network organizations.
RESEARCHMETHODOLOGY
I conducted field and ethnographicanalysis at 23 women's
better-dress firms in the New YorkCityapparelindustry,a
model competitive marketwith intense internationalcompe-
tition, thousands of local shops, and low barriersto entry,
start-upcosts, and search costs. Inthis type of industrial
setting economic theory makes strong predictionsthat social
ties should play a minimalrole in economic performance
(Hirschman,1970), and this is thus a conservative setting in
which to examine conjectures about embeddedness. Field
methods are advantageous here because they providedrich
data for theorizingand conductinga detailed analysis of the
dynamics of interfirmties, even though the 23 cases
examined here can have but moderate generalizability.
I interviewedthe chief executive officers (CEOs)and
selected staff of 23 apparelorganizationswith sales ranging
from $500,000 to $1,000,000,000. An advantage of studying
firms of this type is that the senior managers are involvedin
all key aspects of the business and consequently have
firsthandknowledge of the firm's strategy and administrative
activities. I selected firms that variedin age, sales, employ-
ment, location,type, and the CEO'sgender and ethnicityto
insure properindustryrepresentationand to minimizethe
likelihoodthat interfirmcooperationcould be attributedto
ethnic homogeneity or size (Portes and Sensenbrenner,
1993). The sample was drawnfrom a register that listed all
the firms operatingin the better-dress sector of the New
Yorkapparelindustry.Table 1 provides a descriptivesum-
maryof the sample. This register and other data on firm
attributescame from the InternationalLadies'Garment
Workers'Union (ILGWU,now called UNITE),which orga-
nizes 87 percent of the industry(Waldinger,1989) and which
helped me identifyrepresentativefirms from their data base.
Union records indicatethat there were 89 unionizedmanu-
facturersand 484 unionizedcontractorsin the better-dress
38/ASQ, March 1997

Table 1
Summary of Ethnographic Interviews and Organizational Characteristics of the Sample*
Firm's HQ or Number of
birth Number of factory CEO Number of interview
Type of firm year Size employees location demographics interviews hours
Converter 1962 Medium 22 Midtown Jewish female 1 2
Designer 1986 Small 3 Midtown Jewish female 2 4
Designer 1980 Small 3 Midtown Swedish male 2 6
Manufacturer 1951 Large 182 Midtown Jewish male 1 2
Manufacturer 1986 Large 6 Midtown Anglo male 1 3
Manufacturer 1974 Large 153 Brooklyn Anglo male 2 2
Manufacturer(Pilotstudy) 1954 Large 7 Denver Jewish male 1 2
Manufacturer 1941 Medium 51 Midtown Arabmale 1 2
Manufacturer 1939 Medium 75 Midtown Jewish female 1 3
Manufacturer 1977 Medium 10 Midtown Jewish female 7 35
Manufacturer(Pilotstudy) 1970 Small 2 Midtown Jewish male 1 3
Manufacturer 1930 Small 7 Midtown Jewish male 1 2
Manufacturer 1989 Small 3 Midtown Jewish male 1 2
Manufacturer 1973 Small 4 Midtown Anglofemale 2 2
Contractor-Cutting 1962 Large 40 Midtown Jewish male 1 2
Contractor-Sewing 1976 Large 72 Chinatown Chinese female 1 4
Contractor-Sewing 1982 Large 150 Chinatown Chinese male 1 6
Contractor-Sewing 1989 Medium 85 Chinatown Chinese female 2 2
Contractor-Pleating 1972 Small 31 Midtown Hispanicmale 2 2
Contractor-Sewing 1986 Small 46 Chinatown Chinese female 4 8
Truckingcompany 1956 Small 45 Brooklyn Italianmale 3 2
Total 42 113
* Size in sales: small = $500,000-$3 million;medium= $3-10 million;large= $10-35 million.(One largefirmhadsales
of $1 billion.)Mean numberof ties/contractor= 4.33; embedded ties = 1-2, or 61-76% of total business. Mean
numberof ties/manufacturer= 12; embedded ties = 2, or 42% of total business. Source is ILGWUrecords.Sample
and populationmeans do not significantlydiffer.
sector at the time of the study. The unit of analysis was the
interfirmrelationship.
My analysis focused on the women's better-dress sector to
controlfor the differences that exist across industrysectors
(other sectors include menswear, fantasywear, etc.). Better
dresswear is a midscale market (retailsfor $80-$180),
comprises off-the-rackdresses, skirts, and jackets, typically
sells in departmentstores and chains, and tends to be price,
quality,and fashion sensitive.
Figure 1 depicts a typicalorganizationalnetwork in this
sector. Productionrevolves aroundmanufacturers(called
"jobbers")that normallyfabricateno partof the garment;
instead, they design and marketit. The first step in the
productionprocess of a garment entails a manufacturer
makinga "collection"of sample garment designs in-house
or with freelance designers and then showing its collection
to retailbuyers, who place orders. The jobberthen "manu-
factures" the designs selected by the retailbuyers by
managinga network of grading,cutting, and sewing con-
tractingfirms that produce in volume the selected designs in
their respective shops. Jobbers also linkto textile mills that
take raw materialssuch as cottons and plant linens and
make them into griege goods-cloth that has no texture,
color, or patterns. Convertersbuy griege goods from textile
mills and transformthem into fabrics (cloththat has color
and patterns).The fabricis then sold to jobbers who use it in
their clothingdesigns.
39/ASQ, March 1997

Figure 1. Typical interfirm network in the apparel industry's better-
dress sector.
er Makes griege goods Ti)
Dyes and texturizers

griege goods
into fabric
Sells Retailer
Fbric
Places Delivers
order garments
Jobber's Jobber Jobber's

Showroom |Manufacturer' Warehouse
tMakes
Makes / \ \ Sews garment
and delivers to
sample Creates Sends manufacturer
design trim
_ At / Sets Sends
sizes fabric Sewing
(9 Design
Studio 0Contracto
Makes Cuts
pattern fabric
ratn Sizes pattern uatt9
Data collection and analysis followed groundedtheory

buildingtechniques (Glaserand Strauss, 1967; Miles and
Huberman,1984). I contacted each CEOby phone and
introducedmyself as a student doing a doctoraldissertation
on the management practices of garment firms. In-depth
interviews were open-ended, lasted two to six hours, and
were carriedout over a five-monthperiod. In eight cases I
was invitedto tour the firm and interviewand observe
employees freely, and in fourteen cases I was invitedfor a
follow-upvisit. At three firms I passed several days inter-
viewing and observing personnel. In these cases and others,
I accompanied productionmanagers when they visited their
network contacts. These trips'enabledme to gather first-
hand ethnographicdata on exchange dynamics and to
compare actors' declared motives and accounts with direct
observations. I recorded interviews and field observations in
a hand-sizespiralnotebook, creatinga recordfor each firm. I
augmented these data with company and ILGWUdata on
the characteristicsof the sampled firms.
I conducted the study in four phases. A pre-studyphase
consisted of two pilot interviews that I used to learn how
the interview materials,my self-presentation,and the
frequency or salience of an event such as price negotiation,
tie formation,or problemsolving affected the accuracyof
reporting.Phase one involvedopen-ended, moderately
directive interviews, and direct field observations. I con-
40/ASQ, March 1997

ducted interviews carefullyso that economic explanations

were adequatelyexamined duringdiscussions. If an inter-
viewee spoke only of the relationshipbetween trust and
opportunism,I asked how she or he differentiatedtrust and
riskor why hostage takingor informationasymmetry could
not explainan action he or she attributedto social ties. I
stressed accuracyin reportingand used nondirectiveitems
to probe sensitive issues, for example, "Canyou tell me
more about that," "Is there anythingelse," or "I am inter-
ested in details like that." The Appendixlists the interview
items. In phase two, I formed an organizedinterpretationof
the data. I first developed a workingframeworkbased on
extant theory and then traveled back and forth between the
data and my workingframework.As evidence amassed,
expectations from the literaturewere retained,revised,
removed, or added to my framework.In this stage, I also did
a formalanalysis of the data using a "cross-site display,"
shown in Table 2 below, that indicates the frequency and
weighting of data across cases and how well my framework
was rooted in each data source (Miles and Huberman,1984).
Likeall data reductionmethods, however, it cannot display
the full richness of the data, just as statistical routines don't
explainall the variance.Phase three focused on gaining
construct validityby conferringwith over a half-dozen
industryexperts at the ILGWU,the Fashion Instituteof New
York,and the GarmentIndustryDevelopment Corporation.
These discussions revealed few demand characteristicsor
recordingerrorsin my data. Thus I believe the chance of
response bias is low, given the sample's breadth,the
cross-checkingof interviewand archivaldata, and the
formalizationof the analysis.
FEATURESAND FUNCTIONSOF EMBEDDEDTIES
Table 2 summarizes the evidence for the features and
functions of embedded ties. One importantinitialfindingis
that the differentaccounts of transactingcan be accurately
summarizedby two forms of exchange: arm's-lengthties,
referredto by interviewees as "marketrelationships,"and
embedded ties, which they called "close or special relation-
ships." These data and the literatureon organizationnet-
works form the basis for my analysis and the framework
developed in this paper.
I found that marketties conformed closely to the concept of
an arm's-lengthrelationshipas commonly specified in the
economic literature.These relationshipswere described in
the sharp, detached language that reflected the natureof the
transaction.Typicalcharacterizationsfocused on the lack of
reciprocitybetween exchange partners,the non-repeated
natureof the interaction,and narroweconomic matters: "It's
the opposite [of a close tie], one hand doesn't wash the
other." "They'rethe one-shot deals." "A deal in which costs
are everything."Other interviews also focused on the lack of
social content in these relationships:"They'rerelationships
that are like far away. They don't consider the feeling for the
human being." "Youdiscuss only money."
An examinationof close relationshipssuggested that they
reflected the concept of embeddedness (Granovetter,1985).
These relationshipswere distinguishedby the personal
41/ASQ, March 1997

Table2
Summary of Cross-Site Ethnographic Evidence for Features and Functions of Embeddedness in 21 Firms*
Source of Evidence
Arm's-length Ties Embedded Ties
Product Direct Product Direct
Features and Functions of Exchange CEO manager observation CEO manager observation
Uses writtencontracts 2 5 2
Personalrelationshipwith partnermatters 2 2 18 7 11
Trustis majoraspect of relationship 14 7 5
Reputationof a potentialpartnermatters 1 2 2 2 2
Reciprocityand favors are important 4 2 10 4 3
Small-numbersbargainingis risky 11 5 4 2 1
Monitorpartnerfor opportunism 13 7 4 3 1
Thickinformationsharing 17 7 4
Use exit to solve problems 13 5
Joint problemsolving 15 5 3
Concentratedexchange with partnermatters 1 1 10 7
Push for lowest price possible 7 4 2 2
Promotes shared investment 9 4
Shortens response time to market 3 1 8 5 2
Promotes innovation 2 4 5 2
Strong incentives for quality 2 10 5 1
Increases fit with marketdemand 4 1 7 4
Source of novel ideas 5 3 1 1 2
* Numbersin cells representfrequencyof responses by interviewees aggregated across person-cases. Emptycells
indicatethat no responses were made by interviewees in that category. Multipleand unambiguousexamples across
cases and sources constitute strong evidence for an element of the framework.An unambiguousexample across a
single case constitutes modest evidence.
natureof the business relationshipand their effect on

economic process. One CEOdistinguishedclose ties from
arm's-lengthties by their socially constructed character:"It
is hardto see for an outsider that you become friends with
these people-business friends. You trust them and their
work. You have an interest in what they're doing outside of
business." Anotherinterviewee said, "They know that
they're like partof the company. They're partof the family."
All interviewees described dealings with arm's-lengthties

and reportedusing them regularly.Most of their interfirm
relationshipswere arm's-lengthties, but "special relations,"
which were fewer in number,characterizedcriticalex-
changes (Uzzi, 1996). This suggested that (a) arm's-length
ties may be greater in frequency but of lesser significance
than close ties in terms of company success and overall
business volume and that (b) stringent assumptions about
individualsbeing either innatelyself-interested or cooperative
are too simplistic, because the same individualssimulta-
neously acted "selfishly"and cooperativelywith different
actors in their network-an orientationthat was shown to be
an emergent propertyof the qualityof the social tie and the
structureof the network in which the actors were embed-
ded. Fineranalyses showed that embedded relationships
have three main components that regulatethe expectations
and behaviorsof exchange partners:trust, fine-grained
informationtransfer,and joint problem-solvingarrangements.
The components are conceptuallyindependent,though
related because they are all elements of social structure.
42/ASQ, March 1997

Trust
Respondents viewed trust as an explicitand primaryfeature
of their embedded ties. It was expressed as the belief that
an exchange partnerwould not act in self-interest at anoth-
er's expense and appearedto operate not like calculated risk
but like a heuristic-a predilectionto assume the best when
interpretinganother's motives and actions. This heuristic
qualityis importantbecause it speeds up decision making
and conserves cognitive resources, a point I returnto below.
Typicalstatements about trust were, "Trustis the distin-
guishing characteristicof a personal relationship";"It's a
personalfeeling"; and "Trustmeans he's not going to find a
way to take advantage of me. You are not selfish for your
own self. The partnership[between firms]comes first."
Trustdeveloped when extra effort was voluntarilygiven and
reciprocated.These efforts, often called "favors,"might
entail giving an exchange partnerpreferredtreatment in a
job queue, offering overtime on a last-minuterush job, or
placingan order before it was needed to help a network
partnerthrougha slow period.These exchanges are note-
worthy because no formaldevices were used to enforce
reciprocation(e.g., contracts, fines, overt sanctions), and
there was no clear metric of conversion to the measuring
rod of money. The primaryoutcome of governance by trust
was that it promoted access to privilegedand difficult-to-
price resources that enhance competitiveness but that are
difficultto exchange in arm's-lengthties. One contractor
explained it this way, "Withpeople you trust, you know that
if they have a problemwith a fabricthey're just not going to
say, 'I won't pay' or 'take it back'. If they did then we would
have to pay for the loss. This way maybe the manufacturer
will say, 'OKso I'llmake a dress out of it or I can cut it and
make a short jacket instead of a long jacket'." In contrast,
these types of voluntaryand mutuallybeneficialexchanges
were unlikelyin arm's-lengthrelationships.A production
manager said, "They [arm's-lengthties] go only by the letter
and don't recognize my extra effort. I may come down to
their factory on Saturdayor Sunday if there is a problem ...
I don't mean recognize with money. I mean with working
things out to both our satisfaction."Trust promotedthe
exchange of a range of assets that were difficultto put a
price on but that enriched the organization'sabilityto
compete and overcome problems, especially when firms
cooperativelytraded resources that produced integrative
agreements.
An analysis of the distinctionbetween trust and risk is useful
in explicatingthe natureof trust in embedded ties (William-
son, 1994). I found that trust in embedded ties is unlikethe
calculated riskof arm's-lengthtransactingin two ways. First,
the distributionalinformationneeded to compute the risk
(i.e., the expected value) of an action was not culled by
trusting parties. Rather,in embedded ties, there was an
absence of monitoringdevices designed to catch a thief.
Second, the decision-makingpsychology of trust appearedto
conform more closely to heuristic-basedprocessing than to
the calculativeness that underliesrisk-baseddecision making
(Williamson,1994). Interviewees reportedthat among
embedded ties the informationneeded to make risk-based
43/ASQ, March 1997

decisions was not systematicallycompiled, nor were base
rates closely attended to, underscoringthe heuristicprocess-
ing associated with trust. Moreover,the calculativestance of
risk-basedjudgments, denoted by the skeptical interpretation
of another's motives when credible data are absent, was
replaced by favorableinterpretationsof another's unmoni-
tored activities. One CEOsaid, "Youmay ship fabricfor 500
garments and get only 480 back. So what happened to the
other 20? Twenty may not seem like a lot, but 20 from me
and 20 from another manufacturerand so on and the
contractorhas a nice little business on the side. Of course
you can say to the contractor,'What happened to the 20?'
But he can get out of it if he wants. 'Was it the truckerthat
stole the fabric?'he might ask. He can also say he was
shorted in the originalshipment from us. So, there's no way
of knowingwho's to blame for sure. That's why trust is so
important."This interviewee's statement that he trusts his
exchange partneris also not equivalentto his saying that the
probabilityof my exchange partnerskimmingoff 20 gar-
ments is very small, because that interpretationcannot
explaininterviewees' investments in trust if calculations
using base-rate data on shrinkagecould supply sufficient
motives for action.
These observations are also consistent with the psychology
of heuristics in several other ways. Althoughmy intention
here is not to explainsocial structuraloutcomes via psycho-
logical reductionism,I mention these links because they help
distinguishthe psychology of embeddedness from that of
atomistic transacting.By the term "heuristic,"I refer to the
decision-makingprocesses that economize on cognitive
resources, time, and attention processes but do not neces-
sarilyjeopardizethe qualityof decisions (Aumannand Sorin,
1989). In makingthis argument, I draw on the literaturethat
shows that heuristics can help people make quickdecisions
and process more complex informationthan would be
possible without heuristics, especially when uncertaintyis
high and decision cues are socially defined. In such contexts,
heuristics have been shown to produce qualitydecisions that
have cognitive economy, speed, and accuracy(Messick,
1993). Thus, the research that shows that heuristicprocess-
ing is most likelywhen the problemis unique or decision-
makingspeed is beneficial(Kahnemanand Tversky,1982) is
consistent with how embedded ties particularizethe fea-
tures of the exchange relationship,how informationis
attended to, gathered, and processed, and my findingthat
decision-makingspeed is advantageous among network
partners.
The heuristiccharacterof trust also permits actors to be
responsive to stimuli. If it didn't,actors relyingon trust
would be injuredsystematicallyby exchange partnersthat
feign trust and then defect before reciprocating(Burt,Knez,
and Powell, 1997). I found that trust can breakdown after
repeated abuses, because its heuristicqualityenables actors
to continue to recognize nontrivialmistreatments that can
change trust to mistrust over time, a findingconsonant with
research on keiretsuties (Smitka,1991). Two CEOs de-
scribed how repeated abuse of trust can corrodea close tie:
"Sometimes they ask a favor for a lower price and I'lldo it.
44/ASQ, March 1997

But if they always do that, they're rippingme off." "If the

other firm's busy he'll stay with us and vice versa. If he
switches to a new contractorthen I won't work with that
manufactureragain."
Unlikegovernance structures in atomistic markets, which are

manifested in intense calculativeness, monitoringdevices,
and impersonalcontractualties, trust is a governance
structurethat resides in the social relationshipbetween and
among individualsand cognitivelyis based on heuristicrather
than calculativeprocessing. In this sense, trust is fundamen-
tally a social process, since these psychologicalmechanisms
and expectations are emergent features of a social structure
that creates and reproducesthem throughtime. This
component of the exchange relationshipis important
because it enriches the firm's opportunities,access to
resources, and flexibilityin ways that are difficultto emulate
using arm's-lengthties.
Fine-grained Information Transfer
I found that informationexchange in embedded relationships

was more proprietaryand tacit than the price and quantity
data that were traded in arm's-lengthties. Consistent with
Larson's(1992) findings, it includes informationon strategy
and profitmargins,as well as tacit informationacquired
through learningby doing. The CEOof a pleatingfirm
described how exchange of nonpriceand proprietary
informationis a main feature of his embedded ties: "Con-
stant communicationis the difference. It's just something
you know. It's like havinga friend.The small details really
help in a crunch.They know we're thinkingabout them. And
I feel free to ask, 'How are things going on your end, when
will you have work for us?"'
Relativeto price data, fine-grainedinformationtransferis not

only more detailed and tacit but has a holistic ratherthan a
divisiblestructurethat is difficultto communicate through
marketties. Inthe context of the fashion industry,I found
that this informationstructureis manifested as a particular
"style," which is the fusion of components from different
fashions, materials,nomenclatures,and productiontech-
niques. Because a style tends to be forbiddingand time
consuming even for experts to articulateand separate into
discrete component parts, it was difficultto codify into a
patternor to convey via arm's-lengthties without the loss of
information.Forexample, a designer showed me a defective
pleated skirtand described how only his embedded ties
would be likelyto catch the problem. His demonstrationof
how differentfabrics are meant to "fall,""run,""catch
light,"and "forgivestitching" made it clear that information
transferwith his close ties is a composite of "chunks"of
informationthat are not only more detailed than price data
but more impliedthan overtly expressed in conversation. It
also appearedthat the transferof fine-grainedinformation
between embedded ties is consistent with HerbertSimon's
notions of chunkingand expert rationality,in that even
though the informationexchanged is more intricatethan
price data, it is at the same time more fully understood
45/ASQ, March 1997

because it is processed as composite chunks of information
(a style) ratherthan as sequential pieces of dissimilardata.2
A designer explained how these factors improvea firm's
abilityto bringproductsto marketquicklyand to reduce
errors:"If we have a factorythat is used to makingour
stuff, they know how it's supposed to look. They know a
particularstyle. It is not always easy to make a garmentjust
from the pattern. Especiallyif we rushed the pattern. But a
factorythat we have a relationshipwith will see the problem
when the garment starts to go together. They will know
how to work the fabricto make it look the way we intended.
A factorythat is new will just go ahead and make it. They
won't know any better."
Fine-grainedinformationtransfer benefits networked firms
by increasingthe breadthand orderingof their behavioral
options and the accuracyof their long-runforecasts. A
typicalexample of how this occurs was described by a
manufacturerwho stated that he passes on criticalinforma-
tion about "hot selling items" to his embedded ties before
the other firms in the marketknow about it, giving his close
ties an advantage in meeting the future demand: "I get on
the phone and say to a buyer, 'this group's on fire' [i.e.,
many orders are being placed on it by retailbuyers]. But
she'll buy it only as long as she believes me. Other manufac-
turers can say, 'It's hot as a pistol,' but she knows me. If
she wants it she can come down and get it. The feedback
gives her an advantage."
These cases demonstrate that fine-grainedinformation
transfer is also more than a matter of asset-specific know-
how or reducinginformationasymmetry between parties,
because the social relationshipimbues informationwith
veracityand meaning beyond its face value. An illustrative
case involveda manufacturerwho explained how social ties
are criticalfor evaluatinginformationeven when one has
access to an exchange partner'sconfidentialdata. In such a
case, one would imagine that this access would make the
qualityof the social relationshipunimportantbecause the
informationasymmetry that existed between the buyer and
seller has been overcome. This interviewee argued, how-
ever, that while he could demand that the accounting
records of a contractorbe made availableto him so that he
2
I owe the insightfulobservationabout
might check how the contractorarrivedat a price, the
chunkingand expert rationality to an records would be difficultto agree upon in the absence of a
anonymousreviewer,who helped me relationshipthat takes for grantedthe integrityof the source.
fine-tunemy analysisand who also The manufacturersaid, "Ifwe don't like the price a contrac-
suggested a more radicalimplicationof
my findings-that embeddedness can tor gives us, I say, 'So let's sit down and discuss the costing
overcome boundedrationality alto- numbers.' But there are all these 'funny numbers' in the
gether-an interpretationI am more
reluctantto endorse. Althoughembedded contractor'sbooks and so we argue over what they mean.
ties appearto reduce boundedrationality We disagree . . . and in the end the contractor says, 'We
by expandingthe rangeof data attended don't have a markup,'and then he looks at you like you have
to and the speed of processing,I would
arguethat this expansiondoes not three heads for asking . . . because he knows we don't
constitutefull rationality.I am more know each other well enough to agree on the numbers in
confidentin concludingat this pointthat the first place."
embedded ties reflect "expertrational-
ity,"a thirdkindof rationality that exists
between pureand boundedrationality Thus, informationexchange in embedded ties is more tacit
(Prietulaand Simon, 1989). Manypoints and holistic in naturethan the price and quantitydata
in Prietulaand Simon's (1989)discussion exchanged in arm's-lengthties. The valuationof this informa-
of the organizationalimplicationsof
expert rationality
supportand extend the tion has its basis in the social identities of the exchange
argumentsmade here. partnersand in the manner in which it is processed, via
46/ASQ, March 1997

chunking,even though it is intricateand detailed. These

features help to convey the preferences and range of
strategic options availableto exchange partners,increasing
effective interfirmcoordination.
Joint Problem-solving Arrangements
The use of social arrangementsto coordinatemarket
transactionsis supposedly inefficientbecause the price
system most efficientlycoordinatestransactions,except
under conditions of bilateralmonopoly or marketimperfec-
tion (Hirschman,1982: 1473). In contrast, I found that
embedded ties entail problem-solvingmechanisms that
enable actors to coordinatefunctions and work out problems
"on the fly." These arrangementstypicallyconsist of
routines of negotiationand mutualadjustmentthat flexibly
resolve problems (see also Larson,1992). Forexample, a
contractorshowed me a dress that he had to cut to different
sizes depending on the dye color used because the dye
color affected the fabric'sstretching. The manufacturerwho
put in the order didn't know that the dress sizes had to be
cut differentlyto compensate for the dyeing. If the contrac-
tor had not taken the initiativeto research the fabric's
qualities, he would have cut all the dresses the same
way-a costly mistake for the manufacturerand one for
which the contractorcould not be held responsible. Both the
manufacturerand the contractorreportedthat this type of
integrationexisted only in their embedded ties, because
their work routinesfacilitatedtroubleshootingand their
"business friendship"motivated expectations of doing more
than the letter of a "contract."The manufacturerexplained:
"When you deal with a guy you don't have a close relation-
ship with, it can be a big problem.Things go wrong and
there's no telling what will happen. With my guys [his key
contractors],if something goes wrong, I know we'll be able
to work it out. I know his business and he knows mine."
These arrangementsare special, relativeto market-based
mechanisms of alignment,such as exit (Hirschman,1970),
because learningis explicit ratherthan extrapolatedfrom
anotherfirm's actions. Hirschman(1970) showed that a firm
receives no direct feedback if it loses a customer through
exit; the reasons must be inferred.In embedded relation-
ships, firms work through problems and get direct feedback,
increasinglearningand the discovery of new combinations,
as Helper(1990) showed in her study of automaker-supplier
relationships.In contrast, one informantsaid about market
ties, "Theydon't want to work with the problem.They just
want to say, 'Thisis how it must be.' Then they switch [to a
new firm]again and again." Inthis way, joint problem-
solving arrangementsimproveorganizationresponses by
reducingproductionerrorsand the numberof development
cycles. Joint problem-solvingarrangementsare mechanisms
of voice. They replace the simplistic exit-or-stayresponse of
the marketand enrichthe network, because working
through problems promotes learningand innovation.
A Note on the Formation of Embedded Ties and
Networks
Althougha full discussion of network formationexceeds this
paper's scope, I can summarizemy findings on this process
47/ASQ, March 1997

to establish the linkbetween embedded ties and the
structureof organizationnetworks (Uzzi,1996). I found that
embedded ties primarilydevelop out of third-partyreferral
networks and previous personal relations. In these cases,
one actor with an embedded tie to two unconnected actors
acts as their "go-between." The go-between performstwo
functions: He or she rolls over expectations of behaviorfrom
the existing embedded relationshipto the newly matched
firms and "calls on" the reciprocityowed him or her by one
exchange partnerand transfers it to the other. In essence,
the go-between transfers the expectations and opportunities
of an existing embedded social structureto a newly formed
one, furnishinga basis for trust and subsequent commit-
ments to be offered and discharged.As exchange is recipro-
cated, trust forms, and a basis for fine-grainedinformation
transferand joint problemsolving is set in place (Larson,
1992). This formationprocess exposes network partnersto
aspects of their social and economic lives that are outside
the narroweconomic concerns of the exchange but that
provideadaptive resources, embedding the economic
exchange in a multiplexrelationshipmade up of economic
investments, friendship,and altruisticattachments.
The significantstructuralconsequence of the formationof
dyadic embedded ties is that the originalmarketof imper-
sonal transactionsbecomes concentrated and exclusive in
partnerdyads. Since an exchange between dyads has
repercussions for the other network members through
transitivity,the embedded ties assemble into extended
networks of such relations.The ties of each firm, as well as
the ties of their ties, generate a network of organizations
that becomes a repositoryfor the accumulatedbenefits of
embedded exchanges. Thus the level of embeddedness in a
network increases with the density of embedded ties.
Conversely,networks with a high density of arm's-length
ties have low embeddedness and resemble an atomistic
market.The extended network of ties has a profoundeffect
on a firm's performance,even though the extended network
may be unknownor beyond the firm's control (Uzzi, 1996).
EMBEDDEDNESS,INTERFIRM NETWORKS,AND
PERFORMANCE
Embeddedness is of slight theoreticaland practicalvalue if
more parsimoniousaccounts of exchange can explainas
much. As Friedman(1953) argued, it doesn't matter if reality
is not as the economic model purportsso long as the
model's forecasts agree with empiricalobservation. In
response to this argumentand the need to specify the
mechanisms of embeddedness, I show in this section how
embeddedness advances our understandingof key eco-
nomic and social outcomes. Foreach outcome, I specify
propositionsabout the operationand outcomes of interfirm
networks that are guided implicitlyby ceteris paribus
assumptions. My goal is not to model a specific outcome,
such as profitability,but to show how social structure
governs the interveningprocesses that regulate key perfor-
mance outcomes, both positive and negative.
Economies of Time and Allocative Efficiency
Economists have argued that people's time is the scarcest
resource in the economy and that how it is allocated has a
48/ASQ, March 1997

profoundeconomic effect (Juster and Stafford,1991). I

found that embeddedness promotes economies of time (the
abilityto capitalizequicklyon marketopportunities),because
the transactionaldetails normallyworked out to protect
against opportunism(contracts,price negotiations, schedul-
ing) in arm's-lengthrelationshipspriorto productionare
negotiated on the fly or after productionis completed.
Contractingcosts are avoided, because firms trust that
payoffs will be dividedequitably,even when comparative
markettransactionsdo not exist. In addition,fine-grained
informationtransferspeeds data exchange and helps firms
understandeach other's productionmethods so that
decision makingcan be quickened.Joint problem-solving
arrangementsalso increase the speed at which productsare
broughtto marketby resolving problems in real time during
production."Bud,"the CEOof a large dress firm, explained
how embeddedness economizes on time in a way that is
unachievableusing arm's-lengthcontacts: "We have to go to
marketfast. Bids take too long. He [the contractor]knows
he can trust us because he's partof the 'family.'Sometimes
we get hurt [referringto a contractorthat takes too long to
do a job] and we pay more than we want to. Sometimes we
thinkthe contractorcould have done it quickerand he takes
less than he wants. But everythingis negotiated and it saves
us both from being killedfrom a poor estimate. We do first
and fix price after."
While economies of time due to embeddedness have
obvious benefits for the individualfirm,they also have
importantimplicationsfor allocativeefficiency and the
determinationof prices. This is because embeddedness
helps solve the allocationproblemby enablingfirms to
match productdesigns and productionlevels more closely to
consumer preferences than is possible in an atomized
marketgoverned by the price system. When the price
system operates, there is a lag between the market's
response and producers'adjustments to it. The longer the
lag, the longer the marketis in disequilibrium,and the longer
resources are suboptimallyallocated. Underproduceditems
cause shortages and a rise in prices, while overproduced
items are sold at a discount. This is especially true when
goods are fashion-sensitiveor when long lead times exist
between design and production,because producersare
more likelyto guess inaccuratelythe future demands of the
market.They may devote excess resources to goods that do
not sell as expected and too few resources to goods that
are in higherdemand than expected. Consumers can also
gain increased access to goods that best meet their needs,
while the productionof low-demandgoods is minimized
before prices react.
Consequently,the allocativeefficiency of the market
improves as waste is reduced (fewer products are dis-
counted), and fast-selling items do not run out of stock. In
this way, embedded ties offer an alternativeto the price
system for allocatingresources, especially under conditions
of rapidproductinnovationand mercurialconsumer prefer-
ences. While these findings are not meant to implythat
prices offer no valuableinformationfor makingadjustments,
they do suggest that they are a limiteddevice when adjust-
49/ASQ, March 1997

ment must be timely and coordinated. Under these condi-
tions, as Hirschman (1970) conjectured, both organizational
and interfirm adaptation appears less effectively coordinated
by prices than by embeddedness. These observations can
be summarized in the following propositions:
Proposition la: The weaker the abilityof prices to distillinforma-
tion, the more organizationswill form embedded ties.
Proposition lb: The greaterthe level of embeddedness in an
organization'snetwork,the greater its economies of time.
Proposition ic: The greaterthe competitive advantageof achieving
real-timechange to environmentalshifts or fashion-sensitive
markets,the more networkforms of organizationwill dominate
competitive processes and produceallocativeefficiencies relativeto
other forms of organization.
Search and Integrative Agreements
In the neoclassical model, efficiency and profitmaximization
depend on individualsearch behavior.Search is needed to
identifya set of alternativesthat are then rankedaccording
to a preference function. If there is no search behavior,there
can be no rankingof alternativesand therefore no maximiza-
tion. This suggests that search procedures are a primary
buildingblock of economic effectiveness and therefore are
of great theoreticaland practicalimportanceto the study of
the competitiveness of organizations.
In the neoclassical model, search ends when the marginal
cost of search and the expected marginalgain of a set of
alternativesis equal to zero. "Ina satisficing model search
terminates when the best offer exceeds an aspirationlevel
that itself adjusts graduallyto the value of the offers re-
ceived so far" (Simon, 1978: 10). The above statements by
"Bud"that "everythingis negotiated" and that "Sometimes
... we pay more than we want to," or "Sometimes we
thinkthe contractorcould have done it quickerand he takes
less than he wants," suggest that each firm satisfices rather
than maximizes on price in embedded relationships.More-
over, Bud's statements that "we need to go to marketfast"
and "We do first and fix price after" demonstrate that in
contrast to arm's-lengthmarketexchange, firms linked
throughembedded ties routinelydo not search for competi-
tive prices first but, rather,negotiate key agreements
afterwards.
To Simon's (1978) model of search I add the following
qualification:search procedures depend on the types of
social ties maintainedby the actor, not just the cognitive
limits of the decision maker. I found that embedded ties
shape expectations of fairness and aspirationlevels, such
that actors search "deeply" for solutions within a relation-
ship ratherthan "widely"for solutions across relationships.
A reasonablefirst conjectureof how this network phenom-
enon operates is that multiplexlinksamong actors enable
assets and interests that are not easily communicated
across marketties to enter negotiations, increasingthe
likelihoodof integrativeagreements that pool resources and
promote mutuallybeneficialsolutions, ratherthan distributive
agreements that aim for zero-sum solutions. Solutions are
resolved within the relationship,on integrativeratherthan
distributed grounds, where integrative agreements are
50/ASQ, March 1997

themselves made possible because multiplexties among

network partners(e.g., supplier,friend,community member)
reveal interests and enlarge the pie of negotiable outcomes
(Bazermanand Neale, 1992). For example, when the
above-describedcontractorincorrectlycut a jobber's gar-
ment, the jobber searched for a solution within the relation-
ship (i.e., makinga short jacket instead of the planned long
jacket) based on the expectation that the contractorwould
voluntarilyreciprocatein the future and preferto solve the
problemwithin the relationshipratherthan throughexit. An
interviewee explainedthis logic: "I'd ratherbusiness go to a
friend, not an enemy. My theory is it is not competition.
Problemsare always happeningin production.I always tell
the manufacturerthat 'it's not my problem,it's not his.' I call
to always solve the problem, not to get out of fixingthe
problem.We are all in the same boat." Anothersaid suc-
cinctly, "Win-winsituations definitelyhelp firms survive. The
contractorsknow that they will not lose."
My findings also suggest that embeddedness operates
under microbehavioraldecision processes that promote a
qualitativeanalysis of discrete categories (highvs. low
quality),ratherthan continuous amounts (quantitiesand
prices), as in the neoclassical approach.This point is illus-
trated by a CEO's rankingof embedded ties as more
effective enablers of qualityproductionthan arm's-length
ties: "Anyfirm, any good firmthat's been around,whether
it's Italian,Japanese, or German,and I've been there
because we've been aroundfor four generations, does
business like we do. I have a guy who has been with me 22
years. We all keep long-termrelationshipswith our contrac-
tors. That's the only way you become importantto them.
And if you're not important,you won't get quality."Embed-
ded ties promote each party'scommitment to exceed
willinglythe letter of a contract,to contributemore to the
relationshipthan is specified, and solve problems such that
categoricallimits are sufficient to motivate a high level of
qualityin production.In arm's-lengthties, by contrast, target
outcomes must be contractuallydetailed at the outset
because there are no incentives to motivate positive contri-
butions afterwards,a conditionthat also limits the search for
and recognitionof potentialproblems.
These findings suggest that it is of theoreticaland practical
importto assess the economic development potentialof
differentsearch procedures. Hence, I offer the following
propositions:
Proposition 2a: Search proceduresdepend on the type of ex-
change tie. The width of search across relationshipsincreases with
the numberof arm's-lengthties in the networkand decreases with
the numberof embedded ties in the network.The depth of search
within a relationshipincreases with the strength of the embedded
tie.
Proposition 2b: The greaterthe level of embeddedness in a
network,the more likelyit is that integrativeratherthan distributive
agreements will be reached.
Proposition 2c: The more competitiveadvantagedepends on
reachingpositive-sumsolutionsto interfirm
coordination
problems,
the moreorganizationnetworks,ratherthanotherformsof organi-
zation,willdominatecompetitiveprocesses.
51/ASQ, March 1997

Risk Taking and Investment
The level of investment in an economy promotes positive
changes in productivity,standardsof living,mobility,and
wealth generation. Economictheory credits investment
activityprimarilyto tax and interest-ratepolicies that influ-
ence the level, pattern,and timing of investments. I found
that in the apparelindustryembeddedness enables invest-
ments beyond the level that would be generated alone by
the modern capitaland factor markets. Embeddedness
creates economic opportunitiesbecause it exists priorto the
individualswho occupy competitive positions in a network of
exchange and defines how traits that signal reliabilityand
competence are interpretedby potentialexchange partners,
for three reasons. First,it increases expectations that
noncontractual,nonbindingexchanges will be reciprocated
(Portes and Sensenbrenner, 1993). Second, social networks
reduce the complexityof risktaking by providinga structure
that matches known investors. Third,networkties linkactors
in multipleways (as business partners,friends, agents,
mentors), providinga means by which resources from one
relationshipcan be engaged for another. In riskyinvestment
situations, these factors increase an actor's capacityto
access resources, adjust to unforeseen events, and take
risks.
Interviewees argued that the unique expectations of reci-
procityand cooperative resource sharingof embedded ties
generate investments that cannot be achieved through
arm's-lengthties that are based on immediate gain. The
importanceof these consequences of embeddedness
seemed particularlymeaningfulfor investments in intellec-
tual propertyor culturalproducts (i.e., an originalstyle),
which are difficultto value by conventionalmeans but
importantfor economic development in informationecono-
mies. On the natureof this process of valuationthrough
embedded ties, a characteristicresponse was, "If someone
needs advertisingmoney, or returns,or a special style for
windows-it will be like any relationship.You'lldo things for
friends. You'llgo to the bankon their orders. The idea that
'they buy and we sell' is no good. Friendswill be there with
you throughthe bad times and good."
The role of embeddedness in matchinginvestors and
investment opportunitieswas exemplified by the CEOof a
truckingand manufacturingfirm who explainedthe condi-
tions underwhich his firm helps contractorswho need
capitalto expand. Consistent with Macaulay's(1963)
findings, both parties independentlysaid that they signed no
contracts because of their expectations of long-termfair
play. The CEOstated, "We never make gifts [i.e., sewing
machines, hangers, racks, new lighting]to potentialstartups
unless there is a historyof personal contact. Never for a
stranger. Onlyfor people we have a rapportwith. So, if
Elaine[CEOof a contractingfirm]wanted to start her own
shop I would make her a gift. But for some stranger-never.
Why should I invest my money on a guy I may never see
again?" In contrast, interviewees believed that few firms use
arm's-lengthties to find investment partners.On this point,
a CEOsaid, "I will give a firm a chance based on Dun and
Bradstreetdata. I call the bankand get a financialreporton
52/ASQ, March1997

the firm's size. I know this is 'marketing'but most contrac-

tors don't do marketing[they mainlyuse firms they know]."
I observed a similarpatternfor investment in special-purpose
technology. The added riskof special-purposetechnology,
however, meant that firms wanted assurances that usually
consisted of a joint-equitystake in the technology. Interest-
ingly,the demand for shared equity was not viewed as
distrust but as a deepening of trust and a symbol of risk
sharing.This was demonstrated by the fact that CEOs most
often approachedanother firm about joint investments when
a close tie existed priorto the planned investment. The
president of a dress company described how priorsocial
relationsshape investment behaviorin specialized technol-
ogy, in this case a $20,000 stitching machine: "Say we want
to do a special stitch. So we go to the contractorand ask
him to buy the machine to do the job. But he says that he
wants us to buy it. But, he has money for this machine like
you have money for bubble gum. He's been with us for 25
years. You see, we might not like the way the dress looks
with the special stitch. Then we won't use the machine, so
he's stuck. The reason he wants us to buy it is that he
wants to know that we're not committed to bullshit-we're
committed to using it."
This kindof joint equity sharingis only partlyconsistent with
the transactioncost economic notion of credible commit-
ments, since the equity ties symbolize trust, not protection
against perfidy-a findingconsistent with the Japanese
suppliermodel (Smitka,1991; Gerlach,1992). Since both
parties had money for the machine, the co-equity stake was
not a significantenough sum to be a reliablehostage for
either firm. Moreover,since both firms had the money to
buy the machine unilaterallyand auction it to the lowest-
biddingshop if they wanted, the transactioncosts of
monitoringa joint investment and hagglingwith a known
individualcould have been avoided altogether.
My analysis suggests that in these situations, the equity
investment acted primarilyas a backup-a redundant
structuraltie that reinforcedthe firms' attachments to each
other. Just as engineers overbuildstructures such as bridges
to withstand supernormalstress when the cost of a failureis
high but the chance of failureis low, these actors appearto
overbuildthe structures of importantexchanges even though
the riskof failuredue to-opportunismmay be low, perhaps
because the cost of randommishaps is high. The mecha-
nism guidingthis process, like that of integrativebargaining,
appears to be multiplexity.In riskysituations, multiplexity
enables resource poolingand adaptationto randomevents.
This implies that multiplexties may develop because of the
riskiness of exchanges. The action of taking risks, however,
is a consequence of having multiplexties at one's disposal.
Cyertand March(1992: 228) discussed a similarassociation
between risktakingand physicalassets in the form of slack.
The contributionhere is that a portfolioof social ties can
performthe same function slack does in boosting risktaking,
especially when actors are in resource-scarce, competitive
environments.The key implicationis that firms would be
less likelyto make investments and take risks in the ab-
53/ASQ, March 1997

sence of embeddedness. These observations suggest the
following propositions:
Proposition 3a: The greaterthe level of embeddedness in an
organization'snetwork,the greateran organization'sinvestment
activityand risktakingand the lower its level of resource commit-
ment to hostage taking.
Proposition 3b: The more competitiveadvantagedepends on the
abilityto reduce productdevelopment riskor investment uncer-
tainty,the more organizationalnetworks, ratherthan other forms of
organizations,will dominate competitive processes.
Complex Adaptation and Pareto Improvements
Neoclassicists argue that social arrangementsof coordination
among firms are unnecessary because the price system
directs self-interested maximizersto choose optimally
adaptive responses. A related approachheld in game theory,
agency theory, and evolutionaryeconomics predicts that
actors will coordinateonly as long as the expected payoffs
of cooperationexceed those of selfish behavior(Simon,
1991).
Contraryto these arguments, I found that embeddedness
assists adaptationbecause actors can better identifyand
execute coordinatedsolutions to organizationalproblems.
Similarto mechanisms identifiedby Dore (1983) and Lincoln,
Gerlach,and Ahmadjian(1996) on the durationof Japanese
interfirmties, these solutions stem from the willingness of
firms to forego immediate economic gain and the abilityto
pool resources across firms. In embedded relationships,it
was typicalfor exchange partnersto informone another in
advance of future work slowdowns or to contractearly for
services to help out an exchange partnerwhose business
was slow. These actions improveforecasts and adaptation
to marketchanges in ways that cannot be achieved through
prices or the narrowpursuitof self-interest. A production
managerexplainedto me how her firm foregoes immediate
self-gain in embedded relationshipsto benefit the adaptation
of her exchange partners.In this case, she could not predict
if the aided contractorwould regain profitabilityor how long
a recovery might take, but she knew that another contractor
could offer high volume discounts and a better immediate
payoff. She said, "I tell them [key contractors]that in two
weeks I won't have much work. You better start to find
other work. [At other times] when we are not so busy, we
try to find work for that time for our key contractors.We will
put a dress into work to keep the contractorgoing. We'll
then store the dress in the warehouse. Where we put work
all depends on the factory. If it's very busy I'llgo to another
factorythat needs the work to get by in the short run."
In contrast, these behaviorswere virtuallynonexistent in
arm's-lengthties because informationabout the need for
work was used opportunisticallyto drive price down, a
findingconsistent with traditionalU.S. automaker-supplier
relationships(Helper,1990). Moreover,price is too unre-
sponsive and noisy a signal of organizationaleffectiveness to
foster interfirmcoordinationor adaptation(Hirschman,1970).
A contractorillustratedwhy price is a poor signal for organi-
zationaladaptationand how it can be used opportunistically
to mask problems:"In close relationshipswe work together.
54/ASQ, March 1997

I handle their last-minutegarment changes and ship fast and

jobbers help me expand and solve productionproblems....
[Other]jobbers push the price down when the contractor
tells his productionproblems. Eventuallythe contractor
wants to leave the manufacturerbecause he doesn't pay
enough next time [to make up for earlierprice concessions].
But in the time a good contractorneeds to find a new jobber
to replace their business they lose their best workers and
then they go out of business."
The implicationsof these findings are revealingwhen
contrasted with game theoretic predictionsthat rely on
self-interested motives to explaincooperation.A core
predictionof game theory is that playerswill switch from
cooperativeto self-interested behaviorwhen the end game
is revealed-when players know the end of the game is near
and therefore should end cooperative play because it yields
lower payoffs than unilateralself-interest (Murnighan,1994).
Contraryto this prediction,I found that embedded firms
continue to cooperate even after the end of the game is
apparent.An illustrativecase concerned a manufacturerthat
was permanentlymoving all its productionto Asia and thus
had begun its end game with its New Yorkcontractors.As a
result, this manufacturerhad strong incentives not to tell its
contractorsthat it intended to leave. Doing so put it at risk
of receiving low-qualitygoods from contractorswho now
saw the account as temporaryand had to redirecttheir
efforts to new manufacturerswho could replace the lost
business. Yet the CEOof this manufacturerpersonally
notified his embedded ties, because his relationshipswith
them obliged him to help them adapt to the closing of his
business, and his trust in them led him to believe that they
would not shirkon quality.Consistent with his account, one
of his contractorssaid that the jobber's personal visit to his
shop reaffirmedtheir relationship,which he repaidwith
qualitygoods. The same manufacturer,however, did not
informthose contractorswith which it had arm's-lengthties.
These findings thus suggest another importantoutcome of
embedded networks:They generate Pareto improvements,
promotinga reallocationof resources that makes at least
one person better off without makinganyone worse off. In
the above case, the jobber's embedded ties were made
better off by receiving informationthat enabled them to
adapt to the loss of his business. By contrast, in the baseline
system of marketexchange, the jobber's arm's-lengthties
were denied access to criticalinformationand thus found
the manufacturer'sdeparturedebilitating.
This behavioris difficultto explainas rationalreputation
maintenance.The manufacturer'sNew Yorkreputationwas
irrelevantto its future success in Asia. Likewise, it would not
have hurtthe contractors'reputationsto shirk,since the
manufacturerwas "deserting"them, not the reverse. As a
rule, I found that in a large marketlike New York's,general-
ized reputationsare surprisinglyweak controldevices
because firms can easily escape their bad reputations,while
positive informationis often hoardedin the open "market."
One contractorsaid, "Manufacturerscan play hit and runfor
years before their reputationcatches up with them."
Anotheradded, "I hear 'Thisone is very picky'or 'Thisguy
55/ASQ. March 1997

is reallybad trouble.' But firms I do all the business for, I
don't tell a word about to others. I don't want the competi-
tion."
Such acts of nondefection raise an interestingquestion: Why
don't actors defect when it serves their self-interest to do
so? One possibilityis that embedded social structures entail
expectations that either change more slowly than or remain
resistant to changes in the purelyeconomic features of the
exchange. This enables the logic of embedded ties to extend
to subsequent transactions,even those subject to different
incentive structures, at least in the near term. Inthe above
case, the manufacturer'svisit to the contractor'sfactory
continued their reciprocalindebtedness even though the
incentive systems for both firms were radicallytransformed
at the instant the manufacturerrevealed his preference to
migrateoffshore.
Althougha conclusive account of such non-self-interested
behaviorcalls for more than ethnographicanalysis, one
explanationis that, with the blendingof the social and
economic lives of actors, relationshipstake on an existence
of their own that remains after the economic transaction
ends (Granovetter,1993). Collectivesuccesses ("We had a
hit season"), common experiences ("Iwent to her daugh-
ter's wedding"), and shared symbols (plaques of apprecia-
tion from exchange partners)vividlyand enduringlyinfluence
actions to furnishresources for which no future gain can be
expected. These causal mechanisms are buttressed by
ample psychologicalresearch that shows that close personal
ties heighten empathy, which increases altruisticbehavior
(Batson, 1990). One manufacturerexplicitlydisplayedthis
reasoning in discussing the main issues affecting his
decision not to move to Taiwan,"You have a heart and a
soul here with the people you work with. I don't want to
pick up my familyeither. So, I'lltry to make it work here.
Not everythingin business works by the economic model.
You act like a schmuck sometimes." This suggests that the
motivationto cooperate when it is not in an individual's
self-interest occurs because the expectations of embedded
ties lag changes in economic incentives or persist against
them, an outcome that is itself sustained by psychological
processes that are set in motion by embedded ties.
This altruisticbehaviorappears irrationalonly in the narrow
economic sense, in that actors forego purelyself-seeking
behavior(Simon, 1978). These actors are conscious nonethe-
less of the fact that they are in business to make a return
and that big returnsare better than small ones. What
distinguishes this rationalityfrom formaleconomic rationality
is not just satisficing and heuristics, but the fact that
self-interest gives way to altruism:Actors strategically
cooperate and equitablydistributeboth positive and negative
outcomes. Thus, contraryto Adam Smith's quip that indi-
viduals do best for others by doing selfishly for themselves,
the above evidence suggests that firms that act in the
interest of others (andagainst their short-terminterests) may
do more for the collective economy and society than if they
had followed purelyselfish pursuits. Hence, I offer the
following propositions:
56/ASQ, March 1997

Proposition 4a: The weaker the abilityof generalizedreputationor

prices to providereliableinformationabout productsor exchange
partners'characteristics,the more organizationswill form embed-
ded ratherthan arm's-lengthties.
Proposition 4b: The greaterthe level of embeddedness in an
organization'snetwork,the more likelyare Pareto-improved
solutions to coordinationproblems.
Proposition 4c: The more competitiveadvantagedepends on
complex adaptation,the more networkforms of organization,rather
than other forms, will dominate competition.
PARADOXESOF EMBEDDEDNESS
If a firm becomes too embedded, does adaptationbecome
more difficultas network relationshipsare tuned to specific
tradingpartners,isomorphismwithin the network decreases
diversity,and a concentrated level of exchange with only a
few network partnersreduces nonredundantinformationand
access to new opportunities(Burt,1992)? This question
suggests a paradoxof theoreticalsignificance:The same
processes by which embeddedness creates a requisitefit
with the currentenvironmentcan paradoxicallyreduce an
organization'sabilityto adapt. In this section I explicate three
conditionsthat turn embeddedness into a liability:(1) there is
an unforeseeable exit of a core network player,(2) institu-
tional forces rationalizemarkets, or (3) overembeddedness
characterizesthe network.
The unexpected loss of a network's core organization,or
more generally,a deep and sudden structuralchange in
resource flows can cause embeddedness to shift from an
asset to a liability.Underthese conditions, social processes
that increase integrationcombine with resource dependency
problems to increase the vulnerabilityof networked organiza-
tions. Forexample, a contractormay become highlyskilled
at workingwith a manufacturer'sfabric,productionsched-
ule, and design specifications. If that manufacturercloses
shop or migrates offshore, then the embedded relationship
that had originallybenefited the contractormay now put it at
a higher riskof failurethan if it had diversifiedits ties,
because it is likelyto lackthe resources needed to transition
to a replacement partner(Romo and Schwartz, 1995).
The problemis the opposite of the free-riderproblem:
diligentcommitment, backed by expectations of reciprocity
and social pressure to perform,intensifies an organization's
involvementwith certain network partnerswhile raisingthe
concomitantcosts of keeping ties to extra-networkpartners
that can providea safety net for unexpected or random
fluctuations.Portes and Sensenbrenner (1993: 1340) drew
attentionto this phenomenon in their study of entrepre-
neurs, whose socially embedded relationshipsgave them
access to resources but restrictedtheir actions outside their
network. In the apparelindustry,the unexpected failureof
Leslie Fay, Inc., a manufacturerat the center of a large
network, was most debilitatingfor the primarycontractors
that had benefited from their close tie to Leslie Fay, Inc.
which, before fallingvictim to a few unscrupuloustop
executives, had sheltered them from a glut of low-cost
competitors and downturns in the economy (Uzzi, 1997). The
above arguments suggest the following propositions:
57/ASQ, March 1997

Proposition 5a: The loss of a core organizationin a networkwill
have a large negative effect on the viabilityof the networkas a
whole.
Proposition 5b: The intensityof the effects of the loss of a core
organizationincreases with its size and the level of embeddedness
in the networksuch that, at the limit,an "extinctioneffect" will
occur, as the deleterious effect of destabilizedeconomic transac-
tions ripplesthroughthe networkand causes widespread failureof
even healthyfirms in the network.
Institutionalarrangementsthat "rationalize"markets or
fracturesocial ties can also cause instability.If changes to
the system rupturesocial ties, then the benefits of embed-
dedness generated by the ties can be lost. Ironically,this
can place firms that invested heavilyin networks at a higher
riskof failurethan market-orientedfirms because the social
relationshipsthat created and supported competitive
advantages no longer exist, and the distinctivecompetencies
of managingnetwork relationshipsmay not translatewell to
other modes of exchange. An example of this type of
breakdownoccurred in the apparelretailtrade in the 1980s.
Priorto that time, retailbuyers maintainedembedded
relationshipswith clothing manufacturers.Duringthe '80s
these longstandingties were brokenwhen many of the giant
retailers(e.g., Macy's, Bullocks,A&S)were bought by
corporateconglomerates (e.g., Federated, Inc.)that imposed
a shift from "relationshipbuying"to "numbers buying"
among their retailbuyers. Numbers buyingemphasizes
short-termprofits,one-shot relationships,and whipsaw-like
competitive bidding.A manufacturerwith 30 years of
experience explained how a shift toward impersonalmarket
exchange destabilizedembedded ties and permanently
affected organizationaloutcomes:
A symbolic relationshipdeveloped into a one-sided relationship.The
big stores got accountingrunningthe store. You didn'thave
fashion-sensitivepeople runningthe store any longer.The fashion-
sensitive people had great fashion sense. They couldn'tread a
balance sheet [but]they developed merchantsand buyers in the
stores. If there was a problemyou knew you'd work it out and
they'd help you. There was a personal rapportwith buyers. We'd
say to one another "Let's work it out." It happened in lots of
situations-promotions, jointadvertising,in seeing what's in for
next year. Then everyone became cautious [post buyout].There
was no longer a good dialogue. Ultimatelywe walked away from
the departmentstores. The relationshipbecame very impersonal
and the manufacturersgot squeezed. The corpses [of failed
manufacturers]littered7th Avenue.
These observations suggest another proposition:
Proposition 6: Organizationsthat buildtheir competitiveadvantage
on the use of embedded ties will be at a high riskof failureif
institutionalchanges fundamentallyrationalizethe basis of, or
precludethe formationof, new embedded ties.
The thirdinstabilityresults from overembeddedness, when
all firms in a network are connected throughembedded ties.
This can reduce the flow of new or novel informationinto
the network because redundantties to the same network
partnersmean that there are few or no linksto outside
members who can potentiallycontributeinnovativeideas
(Burt,1992). Underthese conditions, the network becomes
58/ASQ, March 1997

ossified and out of step with the demands of its environ-

ment, ultimatelyleadingto decline.
Overembeddedness can also stifle effective economic action
if the social aspects of exchange supersede the economic
imperatives.Feelings of obligationand friendshipmay be so
great between transactorsthat a firm becomes a "relief
organization"for the other firms in its network. The stronger
firms in the network may dedicate resources to weaker
members at a rate that outpaces their capacityto rejuvenate
their own resources, an argumentthat is consistent with
Portes and Sensenbrenner's (1993: 1339) findingthat
networks of exclusive ties "turnpromisingenterprises into
welfare hotels, checking economic expansion."
Overembedded networks can sometimes release intense
negative emotions of spite and revenge that trap firms in
self-defeating cycles of behavior.Forexample, Axelrod
(1984) found that defection, even in simple tit-for-tatgames,
can cause feuding when it is perceived as illicitor violates an
implicitunderstandingbetween playersto cooperate. I found
that if the strong assumptions of trust and cooperationare
exploited in embedded ties, vendettas and endless feuds
can arise. Over time these actions can prevailagainst
rationalaction and reduce the firm's abilityto meet the
economic demands of the marketplace. A CEOput it
simply, "Ifyou screw a guy like that [a close tie] he'll stay in
business just long enough to get even."
How can these results be reconciledwith the findingthat
embeddedness is an enablingfeature of organizational
efficacy? My argument has been that organizationsgain
access to special opportunitieswhen connected to their
exchange partnersthroughembedded ties, such that the
opportunitylevel is positively relatedto the degree to which
a firm's network partnersuse embedded ties-at least up to
some threshold.These relationshipssuggest that the effect
of embeddedness and network structureon economic action
depends on two variables:(1) how a firm linksto its network
and (2) the composition of the networkthat a firm is linked
to. The best way for an organizationto linkto its network is
by means of embedded ties, which providebetter access to
the benefits circulatingin the networkthan arm's-lengthties.
The optimalnetwork structureto linkto is a mix of arm's-
length and embedded ties, because each type of tie per-
forms differentfunctions: Embeddedties enrichthe net-
work, while arm's-lengthties prevent the complete
insulationof the network from marketdemands and new
possibilities.This suggests two propositions:
Proposition 7a: Organizational performanceincreases with the use
of embedded ties to linkto networkpartners.
Proposition 7b: Networkstructuresthat integratearm's-lengthand
embedded ties optimize an organization'sperformancepotential;
networkstructures comprisingonly arm's-lengthties or embedded
ties decrease organizationalperformancepotential.
In a study of the New Yorkapparelindustry,I found plau-
sible evidence for propositions7a and 7b using data on the
networkties among contractorsand manufacturersin the
better-dress sector of the New Yorkapparelindustryover an
18-month period (Uzzi,1996). I found that contractorshad a
59/ASQ, March 1997

Figure 2. Network structure and embeddedness from a focal firm's
perspective.
Underembedded
Arm's-length Network Integrated Network Overembedded Network
0 /s0 /u
0~~~~0
0 ~ ~ 0 00
A? o
000~~~0
/ # 0
Focal firm's 1st-order Focal firm's 1st-order Focal firm's 1st-order

and 2nd-order network network is composed of and 2nd-order network
is composed of arm's- mostly embedded ties, is composed of
length ties. and its 2nd-order network embedded ties.
is an integration of arm's-
length and embedded ties.
Key:
..g-- e--*. - = Embedded tie 0=Focal Firm
- = Arm's-length tie a = Manufacturers
.-> ;e t o = Contractors
1st-order 2nd-
Network order
network
significantlylower failurerate when linkedby embedded ties

to their network partnersand that being connected to a
network comprisingan integrationof embedded ties and
arm's-lengthties, ratherthan a network comprisingeither
embedded ties or arm's-lengthties, significantlydecreased
the failurerate even further.Figure2 summarizes this
theoreticalargumentand the types of network structures a
firm could be linkedto. For illustrativepurposes, each
network is composed of contractors(circles)and manufac-
turers (squares)and has a first-ordernetwork made up of an
actor's ties to its exchange partnersand a second-order
network made up of an actor's exchange partners'ties to
their tradingpartners.Exchange relationshipsalso vary in
quality-thick lines and thin lines represent embedded ties
and arm's-lengthties, respectively.The underembedded
network structurehas first-and second-ordernetworks that
both comprise arm's-lengthties. In this type of network, all
ties are arm's-length:A contractoruses arm's-lengthties to
linkto its manufacturers,who also use arm's-lengthties to
transact with their contractors.This network has low
embeddedness and approximatesan atomistic market
(Baker,1990). The overembedded network structurehas
first-and second-ordernetworks that comprise embedded
ties: The contractoruses embedded ties to transact with its
manufacturers,who also use embedded ties to transact with
their contractors.The integratednetwork structureexempli-
fies the hypothesized optimal, integratedstructure.The focal
firm's first-ordernetwork comprises embedded ties, and its
second-ordernetwork is an integrationof embedded and
arm's-lengthties. The conjecture drawnfrom these patterns
is that the degree to which embeddedness facilitates
economic action depends on the qualityof interfirmties,
network position, and network architecture.
60/ASQ, March 1997

DISCUSSION
If economic action is embedded in networks of relations
(Granovetter,1985), then a logicalfirst step is to specify the
dimensions of embedded relationshipsand the mechanisms
by which they influence economic action. This undertaking
builds on the work of others who have launchedthe impor-
tant enterprise of reintroducingsocial structureinto the
analysis of economic phenomena. In tryingto demonstrate
the unique organizationaland marketprocesses that follow
from an understandingof social structureand economic
performance,I analyzedthe propertiesof embedded
relationsand how they create competitive advantages for
firms and networks of firms.
While these processes and outcomes are not quickly
summarized,Figure3 diagramsthe propositionsand logic
presented in the preceding sections into their antecedents
and consequences. The figure lays out, from left to right,the
social structuralantecedents of embedded ties, the compo-
nents of embedded ties, the positive outcomes of an
integratednetwork structure,and the negative outcomes of
overembedded networks. The plausibilityof these proposi-
tions has been partlyestablished in the empiricalresearch
analyzed here. Testing and refinement await future research.
The programmaticimplicationof this work is that embedded-
ness is a unique logic of exchange that results from the
distinct social structureof organizationnetworks and the
microbehavioraldecision-makingprocesses they promote. In
an embedded logic of exchange, trust acts as the primary
governance structure.Calculativeriskand monitoring
systems play a secondary role. Informationtransfer is more
fine-grained,tacit, and holistic than the typical price data of
pure marketexchanges, and joint problem-solvingarrange-
ments promote voice ratherthan exit. On a microbehavioral
level, actors follow heuristicand qualitativedecision rules,
ratherthan intensely calculativeones, and they cultivate
long-termcooperativeties ratherthan narrowlypursue
self-interest. These factors furnishan alternativemechanism
for coordinatingadaptation,speeding productsto market,
and matchingconsumer demand to production.
This paper offers an explanationof the links between social
structure,microbehavioraldecision-makingprocesses, and
economic outcomes within the context of organizational
networks. At the same time, it adds complexityto models of
exchange such as game theory, agency theory, and transac-
tion costs economics. Typicalclassificationschemes differ-
entiate these theories accordingto an actor's motivation
(selfish or cooperative)and rationality(pureor bounded)
(Williamson,1985: 50). But I found that in embedded ties, an
actor's motivationand rationalityresist characterization
within these distinctions,thereby demonstratingthe unique
logic of embeddedness. My findings suggest that in net-
works of close ties, motivationis neither purelyselfish or
cooperative but an emergent propertyof the social structure
within which actors are embedded and that rationalityis
neither purelyrationalor boundedlyrational,but expert. Thus
embeddedness broadens the typicaltypology of exchange
theories to include a new categorizationfor motives (emer-
61/ASQ, March1997

-J
C >
D lw
Lu
0~~~~~~
2 U) W
~~~~ )
y 0 > '
OWa IL L o0 Z + C? C
+~~~~~~~~~~~~~~~~~~~M40L
-
U ~
0 U)0
~ 0
+ . ::
U
*. .
E00Uo0 4
LU~~~~~~~~
> O'6 *. *~ + C ~~~~~0
0
; >x * =E E_ E
H **> ~ * * x * *) *->z * E
' : Co : CD
LC
Cn# Cn Cn .
o@~
OZ~
~~~..J 0
W
+. C,) +U) ' 0 +*.
4-0~~~~~~~~~~~~~~~~4 +
'-'LJ~~~co . ..m * +
~~
0
Zz ~C=W~~~
0
C
D
~+CM
CDUW 4-0.
(DU C + + U) 0 C0D
2 U)U)~~~~
(X >V x EU
LU ..
.2 ~~~~~~C0 W~ %Tj L-CU 4
CC o
-' o.
D4
*E L-
(D
U . *_
EC 4*_U)U 4o0 E(
U s
0
0
Z .
.~~~~~~~>
. .C .
.=
W
0>>&
. a. Ua
0 0):f** 2
0
o
0)
0 m +-,
CU*-s-
CO
0 0U
E-og,
NU)U)
o c-
C0 C
0~~~~~~~~~~~~~~~
0 -~~~~~~~~~~~~~~
U. 4- )0 C>C
B-
0 L-CU C ) C
~~o U) Cl 0 * :
{ H uj
CC
_C ia
.3aL CUU) )
U=Um
U) @0'a
a
o~~~~WCJ)L oi Co a)'
0 - 0
-'H 0 0 a m- C
U 0U)X-CD l
0 ~ ~ U)~US ~ 4-'oU)0 U)L.
C
E >
-LLJ
H0
>0
4~~~C A LCO
(D0CU3, AWO E
<U
0~~~~(
0~~~~~0 ~ ~
U))
62/ASQ, March 1997

gent rationality)and for rationality(expert rationality).To-

gether, the emergent characterof motives and the expanded
nature of informationprocessing sets embeddedness apart
from other logics of exchange and situates social structure
as a preconditionto these psychologicalprocesses.
While structuralembeddedness rests on differentbehavioral
assumptions than revisionaryeconomic accounts, it also
shares commonalities and differences with these frame-
works that have implicationsfor future research. A key
difference is that the unit of analysis is the relationship
between and among actors. This unit of analysis shifts the
focus of inquiryfrom the qualitiesof the transactionto the
qualitiesof the relationship.Anotherdifference is how
structurationinfluences economic action. The cooperative
behaviorsthat follow from an organizationnetwork are
significantbecause networks are the sociological analogue of
the type of small-numbersbargainingsituations that in
transactioncost, game, and agency theories are a fundamen-
tal source of opportunismand inefficiency(Harrison,1994).
In contrast, I have shown that firms in networks cooperate
under ostensible small-numbersbargainingsituations and
end-game conditions.These differences suggest that future
research might examine how differentinterfirmrelationships
affect the development of differenttypes of transactions,
invertingthe logic of transactioncosts economics. If embed-
dedness encourages firms to increase their asset specificity
or to engage in repeated transactionsthat have uncertain
future states, then there are significantimplicationsfor how
learningand technology transferare promoted in alliances.
Futureresearch might also examine when embeddedness
can solve coordinationproblems without the need to
integrateverticallyor erect costly monitoringsystems.
My findingthat actors acknowledge differences between
arm's-lengthand embedded ties and designedly use them
has implicationsfor research in network analysis, where it is
often proposed that network structurealone virtually
determines action (Emirbayerand Goodwin, 1994). Burt's
(1992) foundationalwork takes this structuralapproachto its
most naturalconclusion:A network structurerich in struc-
turalholes is virtuallyall that is needed to induce information
and resources to flow throughthe network like electric
currentthrougha circuitboard. Understandingthe difference
between embedded and arm's-lengthties can add to these
theories by specifying how an actor's abilityto access the
opportunitiesof a contact or network strategicallydepends
on the qualityof the relationshipthat connects them and
how it is managed.
This research extends classic statements of embeddedness.
Polanyi(1957: 43-68) argued that the embeddedness of
economic action in preindustrialsocieties was for all intents
and purposes supplanted in modern life by the logic of
efficient markets. Elaboratingon this view, Granovetter
(1985) argued that virtuallyall economic behaviorin modern
life is embedded in networks of social relationsthat condi-
tion economic processes in ways that Polanyiand neoclassi-
cists only faintlyrecognized. My work adds complexityto
both of these views. LikePolanyi,I found that in the apparel
industryat least, atomistic relationsgovern some transac-
63/ASQ, March 1997

tions. Yet, consistent with Granovetter,I found that atomis-
tic relationsoccupy a confined area of economic life and that
the criticaltransactionson which firms depend most are
embedded in networks of social relationshipsthat produce
positive and unique outcomes that are difficultto imitatevia
other means.
Granovetter(1985) also argued that institutionaleconomics
overly emphasizes the selection mechanism of efficiency in
explainingthe existence of organizationalforms. I found that
the functioningof organizationnetworks depends in parton
historicaland institutionalfactors, as well as their abilityto
satisfy non-efficiency-basedselection mechanisms. Like
Schumpeter (1950), I found that organizationperformance
can depend on the abilityto service niche markets, which
requiresthat the firm bringproductsto marketquicklyand
adapt rapidly,ratherthan to optimize cost efficiency. Embed-
ded networks offer competitive advantages of precisely the
type Schumpeter conjectured:entrepreneurialfirms adept at
innovatingand organizinga shifting network of talent,
products, and resources. This implies that future research
might examine how markets function and competitive
dynamics unfoldwhen organizationscompete on the basis
of their abilityto access and reconfigurean external pool of
resources and partnersratherthan firm-basedcompetencies.
Finally,some caveats may be in order, given that my
argument is grounded in observations from one industry.
First,because I focused on fleshing out the concept of
structuralembeddedness, I necessarily neglected other
types of embeddedness that affect organizations,including
cognitive, institutional,and politicalembeddedness (Zukin
and DiMaggio,1990; Oliver,1996). Second, the small
employment size of these firms and the personal natureof
interfirmties in this industrymay providean especially fertile
groundfor embeddedness that might not exist for larger
firms. As firms grow, ties among individualsmay become
insufficientsources of embeddedness, and other social
mechanisms such as interlocksor shared equity may then
be needed. Studies of large Japanese firms show, however,
that the criticalfactor may be organizationalform ratherthan
size (Dore, 1983; Gerlach,1992). If this is true, and the trend
toward smaller, flatter, more connected organizations
continues, networks could become an importantmode of
organizing.Third,the fashion- and quality-sensitivenatureof
better-dress firms may make economies of time, allocative
efficiency, and complex adaptationmore competitive
advantages in this industrythan in others.
Embeddedness is a puzzle that, once understood,can
furnishtools for explicatingnot only organizationalpuzzles
but marketprocesses such as allocativeefficiency, econo-
mies of time, Pareto-improvements,investment, and
complex adaptation.While these processes are sociological,
they have needlessly been the subject of only economics,
both because of a historicaldivisionof laborbetween the
disciplinesand because organizationtheorists have been
reluctantto study problems that are not within the boundary
of the firm (Swedberg, 1994). As economists pay greater
attention to organizationalissues and the field of organization
64/ASQ, March 1997

theory matures, this divisionof laborappears outmoded. The

interconnectedand institutionalcharacterof the world
economy denotes an opportunetime for economic sociology
to contributeto the theory of markets in a way that con-
nects organizationtheory with an understandingof the social
mechanisms that underliemarketallocationprocesses.
REFERENCES
Abolafia, Mitchell Y. Dore, Ronald Hirschman, Albert 0.

1996 Taming the Market. Chicago: 1983 "Goodwill and the spirit of 1970 Exit, Voice and Loyalty:
University of Chicago Press. market capitalism." British Responses of Decline in
Asanuma, Banri Journal of Sociology, 34: Firms, Organizations, and
1985 "The organization of parts 459-482. States. Cambridge, MA:
purchases in the Japanese Emirbayer, Mustafa, and Jeff Harvard University Press.
automotive industry." Japa- Goodwin 1982 "Rival interpretation of market
nese Economic Studies, 1994 "Network analysis, culture, society: Civilizing, destructive
Summer: 32-53. and the problem of agency." or feeble?" Journal of
American Journal of Sociol- Economic Literature, 20:
Aumann, Robert, and S. Sorin 1463-1484.
1989 "Cooperation and bounded ogy, 99: 1411-1454.
recall." Games and Economic Friedman, Milton Jackson, Matthew O., and Asher
Behavior, 1: 5-39. 1953 Essays in Positive Economics. Wolinsky
Chicago: University of Chicago 1996 "A strategic model of social
Axelrod, Robert and economic networks."
1984 The Evolution of Cooperation. Press.
Journal of Economic Theory,
New York: Basic Books. Gerlach, Michael L. 71: 44-74.
Baker, Wayne E. 1992 Alliance Capitalism: The Social
Organization of Japanese Juster, Thomas F., and Frank
1990 "Market networks and Stafford
Business. Berkeley, CA:
corporate behavior." American 1991 "The allocation of time:
Journal of Sociology, 6: University of California Press.
Empiricalfindings, behavioral
589-625. Ghoshal, Sumantra, and Peter models, and problems of
Batson, Daniel C. Moran measurement." Journal of
1990 "How social an animal: The 1996 "Bad for practice: A critique of Economic Literature, 29:
human capacity for caring." the transaction cost theory." 471-522.
American Psychologist, 45: Academy of Management
Review, 21: 13-47. Kahneman, Daniel, and Amos
336-346. Tversky
Baum, Joel A. C., and Christine Glaser, Barney G., and Anselm 1982 "The psychology of prefer-
Oliver Strauss ences." Scientific American,
1992 "Institutional embeddedness 1967 The Discovery of Grounded 246: 161-173.
and the dynamics of organiza- Theory: Strategies for Qualita-
tive Research. New York: Kreps, David
tional populations." American 1990 Game Theory and Economic
Sociological Review, 57: Aldine.
Modeling. New York: Oxford
540-559. Granovetter, Mark University Press.
Bazerman, Max, and Margaret 1985 "Economic action and social
structure: The problem of Krugman, Paul R.
Neale 1986 Strategic Trade Policy and the
1992 Negotiating Rationally. New embeddedness." American
Journal of Sociology, 91: New International Economics.
York: Free Press. Cambridge, MA: MIT Press.
481-510.
Burt, Ronald 1993 "The nature of economic Larson, Andrea
1992 Structural Holes: The Social relationships." In Richard 1992 "Network dyads in entrepre-
Structure of Competition. Swedberg (ed.), Exploration in neurial settings: A study of
Cambridge, MA: Harvard Economic Sociology: 3-41. the governance of exchange
University Press. New York: Russell Sage. processes." Administrative
Burt, Ronald, Marc Knez, and Science Quarterly, 37:
Harrison, Bennett
Walter W. Powell 1994 Lean and Mean: The Chang- 76-1 04.
1997 "Trust and third-partyties." ing Landscape of Corporate Lazerson, Mark
Unpublished manuscript, Power in the Age of Flexibil- 1995 "A new phoenix: Modern
Graduate School of Business, ity. New York: Basic Books. putting-out in the Modena
University of Chicago. knitwear industry." Adminis-
Helper, Susan
Cyert, Richard, and James G. 1990 "Comparative supplier trative Science Quarterly, 40:
March relations in the U.S. and 34-59.
1992 A Behavioral Theory of the Japanese auto industries: An
Firm, 2d ed. New York: exit voice approach." Busi-
Blackwell. ness Economic History, 19:
153-162.
65/ASQ, March 1997

Leung, Chi Kin Palmer, Donald, Brad Barber, Sterns, Linda Brewster, and
1993 "Personal contacts, subcon- Xueguang Zhou, and Yasemin Mark Mizruchi
tracting linkages, and develop- Soysal 1993 "Corporate financing: Social
ment in the Hong Kong- 1995 "The friendly and predatory and economic determinants."
Zhuiang Delta Region." Annals acquisition of large U.S. In Richard Swedberg (ed.),
of the Association of Ameri- corporations in the 1960s." Explorations in Economic
can Geographers, 83: 272- American Sociological Review, Sociology: 279-308. New
302. 60: 469-500. York: Russell Sage.
Lincoln, James R., Michael L. Peterson, Mitchell, and Raghu Swedberg, Richard
Gerlach, and Christina L. Ahmad- Rajan 1994 "Markets as social struc-
jian 1994 "The benefits of lending tures." In Neil J. Smelser and
1996 "Keiretsu networks and relationships: Evidence from Richard Swedberg (eds.), The
corporate performance in small business data." Journal Handbook of Economic
Japan." American Sociological of Finance, 49: 3-37. Sociology: 255-282. Prince-
Review, 61: 67-88. Podolny, Joel ton, NJ: Princeton University
Macaulay, Stuart 1994 "Market uncertainty and the Press.
1963 "Non-contractual relations in social character of economic Uzzi, Brian
business: A preliminary exchange." Administrative 1996 "The sources and conse-
study." American Sociological Science Quarterly, 39: quences of embeddedness
Review, 28: 53-67. 458-483. for the economic performance
Macneil, Ian R. Polanyi, Karl of organizations." American
1978 "Contracts: Adjustment of 1957 The Great Transformation. Sociological Review, 61:
long-term economic relations Boston: Beacon Press. 674-698.
under classical, neoclassical, 1997 "A network perspective on
Portes, Alejandro, and Julia
and relational contract law." Sensenbrenner organizational decline and
Northwestern University Law 1993 "Embeddedness and immigra- deindustrialization." Interna-
tional Journal of Sociology and
Review, 72: 854-905. tion: Notes on the social
determinants of economic Social Policy (forthcoming).
Messick, David
action." American Journal of Waldinger, Roger D.
1993 "Equality as a decision
heuristic." In BarbaraA. Sociology, 98: 1320-1350. 1989 Through the Eye of the
Mellers and Jonathan Baron Powell, Walter W. Needle: Immigrants and
1990 "Neither market nor hierarchy: Enterprise in New York's
(eds.), Psychological Perspec-
tives on Justice: Theory and Network forms of organiza- Garment Trades. New York:
tion." In Barry Staw and L.L. New York University Press.
Applications: 11-31. New
York: Cambridge University Cummings (eds.), Research in Williamson, Oliver E.
Press. Organizational Behavior, 12: 1985 The Economic Institutions of
295-336. Greenwich, CT: JAI Capitalism. New York: Free
Miles, Matthew, and Michael Press.
Huberman Press.
1984 Qualitative Data Analysis. Prietula, Michael J., and Herbert 1994 "Transaction cost economics
Simon and organization theory." In
Newbury Park, CA: Sage.
1989 "The experts in your midst." Neil J. Smelser and Richard
Moorman, Christine, Gerald Harvard Business Review, Swedberg (eds.), The Hand-
Zaltman, and Rohit Deshponde Jan.-Feb.: 120-124. book of Economic Sociology:
1992 "Relationships between 77-107. Princeton, NJ:
providers and users of market Romo, Frank P., and Michael Princeton University Press.
research: The dynamics of Schwartz
trust within and between 1995 "Structural embeddedness of Wilson, Robert B.
business decisions: A socio- 1989 "Exchange." In John Eatwell,
organizations." Journal of
logical assessment of the Murray Milgate, and Peter
Marketing Research, 29:
migration behavior of plants in Newman (eds.), Allocation,
314-328.
New York State between Information, and Markets, The
Murnighan, Keith J. 1960 and 1985." American New Palgrave: A Dictionary of
1994 "Game theory and organiza- Sociological Review, 60: Economics: 83-93. New York:
tional behavior." Research in 874-907. W. W. Norton.
Organizational Behavior, 16:
83-123. Schumpeter, Joseph A. Zajac, Edward J., and Cyrus Olsen
1950 Capitalism, Socialism and 1993 "From transaction cost to
Oliver, Christine Democracy. New York: Harper transactional value analysis:
1996 "The institutional embedded- Collins. Implications for the study of
ness of economic activity." In interorganizational strategies."
Simon, Herbert A.
Jane Dutton and Joel Baum 1978 "Rationality as process and as Journal of Management
(eds.), The Embeddedness of product of thought." American Studies, 30: 132-145.
Strategy: Advances in Strate- Economic Review, 68: 1-16. Zukin, Sharon, and Paul DiMaggio
gic Management: 163-186. 1991 "Organizations and markets." 1990 Structures of Capital: The
Greenwich, CT: JAI Press. Journal of Economic Perspec- Social Organization of the
Padgett, John F., and Christopher tives, 5: 24-44. Economy. New York: Cam-
K. Ansell Smitka, Michael bridge University Press.
1993 "Robust action and the rise of 1991 Competitive Ties: Subcontract-
the Medici, 1400-1434." ing in the Japanese Automo-
American Journal of Sociol- tive Industry. New York:
ogy, 98: 1259-131 9. Columbia University Press.
66/ASQ, March 1997

APPENDIX:Open-ended Interview Items
Because many of the data were collected ethnographically, these items
summarizethe questions but only partlyconvey the nuances and details of
the lengthy,interactive,and face-to-facediscussions reportedin the text.
Internalorganization:
Is this a proprietorship,
partnership,or corporation?
How manyyears of industryexperience do the principalshave?
Do you produceany productsin-house?
Why do you contractinstead of producein-house?
Do you outsource work that was done in-house?
Is the decision to producein-house primarilyfinancial,organizational,or
historic?
How many firms work for you per year?
How many retailersdo you typicallysell to in a year?
Marketand product characteristics:
What are the characteristicsof your product?
How is your productionorganized?
How sensitive is your product'sdemand to quality,price,and fashion
trends?
How has your marketchanged in the last 5 years?
How has the firmadaptedto these changes?
What does it take to succeed in this business?
Forming interfirmcontacts:
How do you contact new contractors?
When will you use new contractors?
What role does reputationplay?
How does the typicalrelationshipbegin and develop over time?
Are writtencontracts used and when?
Interfirminteraction:
What kindsof relationshipsdo you form with contractors?
Is opportunisma problem?
How do you protectyourself?
How are disagreements resolved?
How do you manage the tradeoffs?
Inwhat ways is power gained in a relationship?
When are you most vulnerablein a relationship?
How do you respondto poor performance?
How do you react to a contractorthat passes on his price increases?
What happens when a new contractoroffers you a lower pricethan your
present contractor(s)?Do you visit your contractor'sshop?
Inwhat way do you rewardgood performance?
Network outcomes:
What benefits do you get from each type of relationship?
What are the downsides?
How do you set prices for goods and services?
What kindof informationis shared in differentrelationships?
Please describe your contractualagreements with regardto setting perfor-
mance and price.
How are new productscreated and test marketed?
How are investments in new equipmentmade?
How do firms borrowmoney or get loans?
How do you increase your abilityto respondto the market?
What promotes innovation?
What events or conditionslead to close business relationships?
What mechanisms are effective in reducingcosts?
Do you attempt to attaina specific mix of relationships?
What prevents you from attainingthe mix you want?
67/ASQ, March 1997


Uzzi 1997 Social Structure &amp; Embeddedness

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Uzzi 1997 Social Structure &amp; Embeddedness

Diunggah oleh

Hak Cipta:

Format Tersedia

Social Structure and Competition in Interfirm Networks: The Paradox of Embeddedness

Author(s): Brian Uzzi

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

Japanese auto and Italianknitwearindustriesare character-

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

er Makes griege goods Ti)

Dyes and texturizers

Jobber's Jobber Jobber's

ratn Sizes pattern uatt9

Data collection and analysis followed groundedtheory

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

ducted interviews carefullyso that economic explanations

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

natureof the business relationshipand their effect on

All interviewees described dealings with arm's-lengthties

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

But if they always do that, they're rippingme off." "If the

Unlikegovernance structures in atomistic markets, which are

Fine-grained Information Transfer

I found that informationexchange in embedded relationships

Relativeto price data, fine-grainedinformationtransferis not

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

chunking,even though it is intricateand detailed. These

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

profoundeconomic effect (Juster and Stafford,1991). I

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

themselves made possible because multiplexties among

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

the firm's size. I know this is 'marketing'but most contrac-

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

I handle their last-minutegarment changes and ship fast and

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

Proposition 4a: The weaker the abilityof generalizedreputationor

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

ossified and out of step with the demands of its environ-

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

Focal firm's 1st-order Focal firm's 1st-order Focal firm's 1st-order

significantlylower failurerate when linkedby embedded ties

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

.2 ~~~~~~C0 W~ %Tj L-CU 4

62/ASQ, March 1997

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

gent rationality)and for rationality(expert rationality).To-

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

theory matures, this divisionof laborappears outmoded. The

Abolafia, Mitchell Y. Dore, Ronald Hirschman, Albert 0.

65/ASQ, March 1997

This content downloaded from 193.255.248.150 on Mon, 2 Feb 2015 12:05:37 PM

Uzzi 1997 Social Structure & Embeddedness

Uzzi 1997 Social Structure & Embeddedness