An Emergent Theory of Structure and Outcomes in Small-Firm Strategic Manufacturing

Networks
Author(s): Sherrie E. Human and Keith G. Provan
Source: The Academy of Management Journal, Vol. 40, No. 2, Special Research Forum on
Alliances and Networks (Apr., 1997), pp. 368-403
Published by: Academy of Management
Stable URL: http://www.jstor.org/stable/256887
Accessed: 01-10-2015 12:44 UTC

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.

Academy of Management is collaborating with JSTOR to digitize, preserve and extend access to The Academy of Management
Journal.

http://www.jstor.org

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
? Academy of Management Journal
1997, Vol. 40, No. 2, 368-403.

AN EMERGENT THEORY OF STRUCTURE AND

OUTCOMES IN SMALL-FIRM STRATEGIC
MANUFACTURING NETWORKS
SHERRIE E. HUMAN
Texas Christian University
KEITH G. PROVAN
University of Arizona

This article presents the results of a comparative study of two networks

of small- to medium-sized manufacturing enterprises in the U.S. wood
products industry. Qualitative and survey data were collected from a
sample of active participants in each network as well as from a sample
of firms that were not network members (market firms) in the same
industry. Using both qualitative and network analysis methodologies,
we developed a model and testable propositions concerning key aspects
of network structure and outcomes.

In 1992, four member firms of the Kentucky Wood Manufacturers' Net-

work won a $2.5 million contract with Disney World in Florida; in that same
year, ten electronics manufacturers, members of Team Nashua Network in
southern New Hampshire, obtained a $1.2 million contract with Compaq
Computers (Lipnack & Stamps, 1993; Selz, 1992). Despite a historically com-
petitive and individualistic business environment in the United States,
small- and medium-sized enterprises (SMEs) have begun to participate in
ongoing cooperative, multifirm interorganizational relationships modeled
after northern Italian industrial districts and Danish networks (Inzerilli,
1990; Nielsen, 1993). Like their European counterparts, U.S. small-to-
medium-sized-enterprise (SME) networks do not replace competitive rela-
tionships, but operate alongside those relationships, allowing designated
collaborators to cooperate on certain projects and to compete on others (Bos-
worth & Rosenfeld, 1992; Perrow, 1992).
Small-and-medium-sized-enterprise networks have many similarities to
their European counterparts but are conceptually distinct from other types of
interorganizational arrangements commonly found in the United States, like
joint ventures, federations, and trade associations. SME networks generally
differ from these forms along three identifiable dimensions: purpose, inter-
dependence, and membership criteria. First, these networks are generally

This research was funded by a grant (number 95-215) from the Ewing Marion Kauffman
Foundation Center for Entrepreneurial Leadership and by grants from the University of Ken-
tucky Small Business Development Center and the European Science Foundation. We thank the
anonymous reviewers for their insightful and constructive comments.

368

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 369

created to provide a forum for direct, even joint, business activities among
multiple network members, in addition to indirect member services. For
instance, Florida's TeCMEN network has linked over 30 firms for joint prod-
uct development, many of which were competitors in that state's electronics
industry. The members remain independent yet work directly together for
mutual objectives. In contrast, the primary purpose of trade associations and
federations is most often to provide indirect services for their participating
firms (e.g., lobbying, promotion). Joint ventures typically pursue the objec-
tives of two organizations through creation of a separately managed venture,
but SME networks pursue organizational objectives through coordinated in-
teractions of perhaps 10, 20, or more individual firms, collectively compris-
ing an interorganizational network.
Second, recent U.S. small-to-medium-sized-enterprise networks pro-
mote complex, reciprocal interdependencies (Thompson, 1967) in which
members provide inputs and receive outputs from each other. Interdepen-
dencies are usually coordinated by the network firms themselves and a
network-level coordinator or administrative organization. A major goal of
this coordinating structure is to facilitate interaction among network mem-
ber firms. Practitioners have identified network firm interdependencies
ranging from "jointly developing solutions to common problems" to "jointly
manufacturing components, assemblies, or finished goods" (Bosworth,
1995: 2). In contrast, federations and joint ventures typically engage in
pooled interdependencies (Thompson, 1967) coordinated through adminis-
trative structures designed to manage, in standardized ways, the pooled
activities of member firms. One network promoter differentiated networks
from trade associations, stating that the latter should change their mission
and "develop real services and do more to facilitate interactions among
members" (Coffey, 1995: 1).
Third, SME network membership criteria emphasize geographically
proximate core competencies among member firms that combine to allow
the firms to accomplish specific organizational objectives that no one mem-
ber could have accomplished individually. Members of the Kentucky Wood
Manufacturers' Network are required to have "production facilities located
in the state" so that members can work on joint production projects (Lich-
tenstein, 1992: 32). In contrast, joint ventures often combine core competen-
cies in large-firm dyadic linkages that are not necessarily geographically
proximate. Trade associations commonly focus on industry participation
rather than on developing organizations' competencies.
Although small firms may benefit substantially from network involve-
ment, much of the research on networks has been on large-firm strategic
alliances or on nonprofit networks, with some exceptions. For instance,
Inzerilli (1990) used a transaction cost perspective to describe how a trust-
based social context reduced opportunism and facilitated success among
small firms in northern Italian industrial districts. Brusco and Righi (1989)
and Lorenzoni and Ornati (1988) confirmed the importance of environmen-
tal factors for small firm growth through networks. Saxenian (1994) de-

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
370 Academy of Management Journal April

scribed the emergence of infrastructures in the United States to support

European-style cooperative systems. These studies helped provide a de-
scriptive foundation for our project while suggesting a link between network
structures and outcomes. For instance, Saxenian (1990: 105) described how
the lack of "administrative coordination" in California's Silicon Valley, "an
American variant of the industrial districts of Europe (1990: 91), made the
Valley's networks more vulnerable to environmental changes over time.
Lorenzoni and Ornati (1988) argued that exchange structures within constel-
lations of complementary firms influenced organizational growth.
Despite some work on the topic, the relationship between key aspects of
network structure and organizational outcomes has often been assumed,
rather than studied directly, or studied only regarding one type of outcome.
Broader organizational theory, like transaction cost economics (Williamson,
1985) and resource dependence (Pfeffer & Salancik, 1978) has often been
used to explain aspects of networks, such as governance or interorganiza-
tional power and resource acquisition, but these theories focus primarily on
interorganizational relations, not networks. Thus, what is known about
which network structures work best and what kinds of outcomes might be
expected from network membership is quite limited, especially in an SME
context. Because of the lack of adequate theory to guide research on small-
to-medium-sized-enterprise networks or to advance network practice in
this important and emerging area, we decided to undertake a study to ex-
plore the structure of U.S. small-to-medium-sized-enterprise networks and
the outcomes member firms obtained from network participation.

RESEARCH QUESTIONS

We organized our research around three questions. The first question

was, What outcomes do firms achieve from network participation? As men-
tioned above, most case studies on small-to-medium-sized-enterprise net-
works have either examined why and how networks have emerged among
small or new private-sector firms or have examined the relevance of the
network strategy for regional economic growth, focusing on a single type of
organizational outcome or on aggregate outcomes such as regional employ-
ment growth. Past research on network outcomes has also tended to focus
heavily on profits and performance (Smith, Carroll, & Ashford, 1995). Al-
though objective measures of profit can sometimes be used with success
(Shortell & Zajac, 1988), small firms often do not have the information sys-
tems needed to collect reliable outcome data (Dess & Robinson, 1984). Fur-
ther, financial benefits either may not accrue to small and medium-sized
firms using cooperative strategies (Baird, Lyles, & Orris, 1992), or they may
lag network formation by several years. We examined both financial and
nonfinancial, or process, consequences (Baird, Lyles, & Reger, 1993) of net-
work involvement, first to address the fundamental question of whether
small and midsize manufacturers benefit from network involvement, and
second, to learn which types of outcomes are most likely to occur.

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 371

Our second research question was, Do SME network firms and their
outcomes differ in discernible ways from their market counterparts and their
outcomes? If the aim of extending research on interorganizational relations
is to explore outcomes of network participation, then it is useful to compare
networks with the alternative mode for organizing, market arrangements
(Jarillo, 1988), thereby highlighting potential network-market differences.
Comparative research has primarily examined network and market ap-
proaches as sets of loosely connected activities, in which researchers list
separate cooperative and competitive strategies, then ask respondents to
indicate the frequency and extensiveness of their use of each activity (Baird
et al., 1992; Brush & Chaganti, 1993). Although these studies provide useful
comparisons of specific competitive and cooperative strategies across com-
panies, they do not address whether a set of market firms can derive benefits
similar to those network firms derive through nonnetwork interorganization-
al linking mechanisms. This is an important question since it underlies the
basic rationale for SME network formation.
Our third research question was, How are small-to-medium-sized-
enterprise networks structured, and to what extent can structural differences
explain organizational outcomes resulting from network participation? De-
spite the fact that sociologists and organizational researchers have examined
network structure for well over two decades (see reviews by Marsden [1990],
Nohria and Eccles [1992], and Wasserman and Galaskiewicz [1994]), struc-
ture has mostly either been the dependent variable or has been used to
predict outcomes related to power and influence within a network (Boje &
Whetten, 1981; Cook, 1977; Cook & Yamagishi, 1992). Researchers have also
used structural properties of organizational relationships to explain other
outcomes, including performance (Galaskiewicz & Marsden, 1978; Shortell
& Zajac, 1988; Stearns, Hoffman, & Heide, 1987), but their focus has mostly
been on dyadic links rather than on multifirm networks, and the context has
mostly been health and human services (Provan & Milward, 1995). Further,
studies of network structure among profit-making firms have often focused
on one or a few interfirm relationships, such as interlocking directorates,
rather than capturing multiple ties among organizations (Barley, Freeman, &
Hybels, 1992). As a result, there is little understanding of small-to-medium-
sized-enterprise network structure and whether different structural forms
facilitate or inhibit the attainment of desired network outcomes. Particularly
at this early stage in SME network evolution in the United States, this in-
formation is critical for network planners and organizers trying to establish
a credible alternative to traditional market-based, competitive relationships.

METHODS

Because there is so little comparative empirical research on SME net-

works and because the research on other types of networks has limited
generalizability to a small-to-medium-sized-enterprise context, we chose

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
372 Academy of Management Journal April

to conduct an inductive study. In keeping with an inductive case study

approach (Eisenhardt, 1989), our project did not start from a particular theo-
retical perspective. Rather, our goal was to build theory from an in-depth
examination of two small-to-medium-sized-enterprise networks and a
matched set of market firms. Guided by relevant gaps in the literature, we
first identified our general research questions and the related constructs,
network structure and organizational outcomes. Next, we collected and ana-
lyzed data using an exploratory (Hartwig & Dearing, 1979) rather than a
confirmatory hypothesis-testing approach; thus, we identified patterns in
the data, then returned to the literature for conceptual clarification of those
patterns. Finally, we developed propositions from our findings. We adopted
Jarillo's (1988) framework, in which strategic networks are contrasted to
markets, to describe interorganizational relationships among independent
organizations, and we adopted his broad definition of strategic networks as
"long-term, purposeful arrangements among distinct but related for-profit
organizations that allow those firms in them to gain or sustain competitive
advantage ... [while remaining] independent along some dimensions"
(Jarillo, 1988: 32).
We narrowed our definition of networks still further, in view of the goals
of our research and how our informants defined them. The definition that
emerged was that networks are intentionally formed groups of small- and
medium-sized profit-oriented companies in which the firms (1) are geo-
graphically proximate, (2) operate within the same industry, potentially
sharing inputs and outputs, and (3) undertake direct interactions with each
other for specific business outcomes (Bosworth & Rosenfeld, 1992). These
interactions may include joint production, new product development, col-
lective marketing, and employee training.1 To examine small-and-medium-
sized enterprises' involvement in multifirm networks, we adopted an organ-
ization-level perspective regarding outcomes that is consistent with the ma-
jor goal of these networks, to improve member firm competitiveness
(Lichtenstein, 1992).
Because this article reports a qualitative inductive study, its format dif-
fers somewhat from the formats generally used in reporting quantitative
research. Our description of the research methods includes an account of the
iterative data collection and analysis process appropriate for exploratory
research on new phenomena (Eisenhardt, 1989). The results of our explor-
atory research, which address our research questions, are presented in the

1 There are many definitions of networks (Nohria & Eccles, 1992). Our definition (1) cap-
tures the meaning of networks common in the small-to-medium-sized-enterprise practitioner
literature and among our respondents, (2) focuses on empirically observable activities, and (3)
distinguishes networks from related organizational forms such as trade associations. A national
newsletter promoting network development stated this: "Only the exceptional trade association
brings businesses face-to-face for purposes of developing trust or cutting business deals. Most
are collective voices to influence governments, not collective production systems to capture
markets" (Coffey, 1995: 6).

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 373

next section. We attempted to build theory from our findings by developing

testable propositions and an overall model. The final section discusses im-
plications and conclusions for research and practice.
Table 1 provides an overview of our three-phase research process. After
an initial groundwork phase, we used a two-stage cross-sectional design
following Eisenhardt's (1989) suggested sequence for exploratory, theory-
building research and Jick's (1979) "triangulation" research strategy, in
which we integrated qualitative and quantitative evidence. Consistent with
the goals for exploratory research, the study incorporated a replication rather
than a sampling logic (Yin, 1989). That is, instead of examining the fre-
quency of network structures and outcomes for statistical inference to a
network population, we viewed network and market firms as multiple cases
in which the unfolding evidence should either be similar or different across
firms. Thus, our goal was theory building rather than theory testing. Data,
collected during 1994 and 1995, were obtained through multiple sources,
including interviews, questionnaires, on-site tours of network firm opera-
tions, and archival documents (network grant reports, newsletters, articles).
Qualitative interviews conducted prior to survey development helped en-
sure construct and item applicability for respondents and provided insights
for interpreting survey results.
Groundwork Phase
During the groundwork phase (1993-94), we developed our research
questions, design, and samples (Table 1). During this time, one of the re-
searchers observed network meetings for two networks (one of which par-
ticipated in this study) and participated for a year on a task force charged
with small-to-medium-sized-enterprise network development. Following
Yin's (1989) suggestions for within- and across-case comparisons in quali-
tative research, we selected two same-industry networks with similar pur-
poses, formation periods, and membership sizes for study and then selected
same-industry market firms located in either of the two states the networks
operated in. The setting for this study, the secondary wood products indus-
try, is one of the "suffering industrial sectors" (Bosworth & Rosenfeld, 1992:
29) targeted by the U.S. economic development community for network
creation over the past decade. The industry, composed of firms that produce
finished and semifinished wood products like cabinets and furniture, has a
large proportion of small and medium-sized firms that could potentially gain
strategic advantage from network involvement.
Our sampling frame, A Catalogue of U.S. Manufacturing Networks
(Lichtenstein, 1992), documented the formation of 27 small-to-medium-sized-
enterprise networks between 1989 and 1991, primarily in the wood prod-
ucts, metal working, and apparel industries. We wanted to find 2 networks
for comparative study. Our criterion for inclusion in the sample was that a
network have ongoing activity and interaction among separately located
members. Through use of this criterion, we avoided networks that existed in
name only or for a single event, or whose member firms were located in one

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
374 Academy of Management Journal April

Cd
-4- 't 0 F--4
cd +-I 10,
lt 0
P-4
Cd
0

> t
.,.j
4-1
>
a) cn Cd
Cd
cn 4;; cd
cn
Cd cd cn 4-4
cd cn 0
0 u
v +., -,-I Cd
Cd > 0 co
cn cn
P..4 CO
Cd
;- 4-4 F--4
'7. .,..4 o cd 0
-4Cd
.,..I
cn a) co
0 17. 0
4-1 Cd v 0 +- Cd
a) cd a) a) to o
CI-8 c-n cp-d' 0
Cd
o cd 0
cd U C u .- 'O cd 0
.,.,
F..4 +-, P, cd - co
o o 0 cd
o o -zi u Cn o -00 cn
CY E-

Cd
9 cn 0
0 cn cn P-4 co 0
14 0 .,..I 0 co cn
0 cn
cd

cn cn
cn cd
to .-
1-4
4 ;-
0
>
cd
C5
4 4--, CD 4-4
0 P- +- > >
U 0 0
---I 0 ;.
.. a) cn
a) CO w
U cn
CX > cn +-I

CO CD
4--i
-pl co co
co
0 0 co co
C.
4
PL
C) co 0
P-4 0 P-4
0 41 0
0 Cd 0 P-4 cd 4-1 cd

4-4
0
co
0
co F--4 cd
F--4
0 0 a)
.-I 0 0 cn .-I
cn cn P, cd -r 0
P-4 0 Cd
a) -4 0 Cd
.,..4
7j
CO CO 0
F--4 F--4
Q
.,.I 0 0 ,
cd
+D ct o .,.I
Cd Cd 0 cn
cd a) P-.4
cn C)
u
a)
N a)
P-4 co
.,-, cj

7
.,..4 cd a) .- P4 cmd Cd co
0 0 Cd 0 -7:j +-
> 4-1 cn
cn -1-4 cn P4
cd
cn +J 0
cn cd
cn m 4-4 0 Cd 4-1 4-1
P-4 Cd cd
cd A W Ici --P-4
`4 .,..qP, U
114 , 4--4 1-4
a)
4-5 0
4-1
cn 0
cn P-4 P-, P-4 u :: I
0 o 0 C) C) -,
C) -'.,4 Cd cd C) Cd o
'.0 -.0
4-1 4-1 r,
Cd m cn P-4 cd cd

cn
Cd m cd
Ici Cd 4-1
cn r4 Cd 4-1
Cd 4-1 m
u C).-I 0
cn cd 0 Cd
.,..I
4-1 cd tg
4-, 0 cd
4-, 0
cn P-4
tq 4-1
4-1
Cd >
P-,
0 >
F--q M) u
Cd a) a) 4-, cn
Cd F-4
F. -,- F. C) O
"O
04 Cd P., 1-4 P., pi o
0 F-4
E CD

cn
4-1
cn

Cd
0 0
-.4 cn cn
Cd
4-1
0 Cd Cd
PZ4 m cn

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 375

a
Data
Results
The
Element
analysis Research

"results"
of
and
the
Selected
Identified
network
research
outcomes Research
exploratory,
comparative
firm-level
primary two-stage
groundwork structure Design
design
constructs:

phase
led 10
to states Selected Groundwork
Selected Network
network products
complete
the 10 interviews
interviews
network two Phase
reputational
membership Sample
and firms
directors
"focus" firmsmarket network industry
from
of for samples in networks
firms directories
of in Selection TABLE
stage
their two 1
and for nominated
questionnaires wood
one;
the
not data after
firms (e.g., Stage (continued)
use Identified Constant
Thematic
use Confirmed outcomes literature
"results" resources)
exchanges,
credibility, comparingusing each
interactive One:
of relationships
that
(e.g., major coding
them
as
stage access of
interorganizational
network enfolding
administrative comparison
interviewInterviews
market to structure)
one themes Qualitative
network and for
led firms and
organizational
interorganizational
to qualitative
approach Phase
do structures
the

data Stage
"focus" Network
of interview DevelopedFirm-level
network analyses
firm-level qualitative Two:
Descriptive
propositions and
stage and
structure:
structure
two. outcomespreliminaryoutcomes:
factoring
statistics,
plotting Questionnaires
andregarding Quantitative
from Network
questionnaire
Phase

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
376 Academy of Management Journal April

building. Preliminary interviews with network planners, administrators,

and member firms in 5 of the listed wood products networks produced 2 that
met our criterion. We labeled these 2 networks "alpha-net" and "beta-net."
Total manufacturing company membership listed in the directories of the 2
selected networks equaled 60 and 77 firms respectively (institutional mem-
bers like universities were excluded). One network operated in a southern
and 1 in a northern state. Table 2 provides descriptions of both networks.
After careful consideration, we decided that it made little sense to col-
lect data from all firms in each network, since many firms were members in
name only and had little involvement with network administration or other
members. Conversations with network administrators indicated that mem-
ber firms cited multiple reasons for their network inactivity, including "lack
of time," "lack of trust," "inconvenient meeting locations," and "difficulty
getting away from the business." Respondents from inactive firms would
have had great difficulty answering our questions and would have added
little value to our overall understanding of network activities and structure

TABLE 2
Network Descriptive Characteristics
Characteristic Alpha-net Beta-net
Year formed 1989 1990
Structure Network administrative Network administrative
organization: Nonprofit organization: Nonprofit
organization composed of organization composed of
part-time executive director, full-time executive director,
one part-time staffer, three full-time staffers,
13-member board of directors 13-member board of directors
Network membership: 60 Network membership: 77
manufacturing companies, 9 manufacturing companies, 9
associate members (e.g., associate members (e.g.,
educational and economic educational and economic
development agencies) development agencies)
Objectives Joint marketing, production, Joint marketing, production,
and development for and development for
secondary wood products secondary wood products
firms in the state, focusing on firms, focusing on a
furniture manufacturers multicounty region of the
state
Member services Individual: Consulting, business Individual: Consulting, business
provided planning, technical planning, technical
assistance, market research assistance, market research
Group: Monthly membership Group: Quarterly board
and board meetings, monthly meetings and newsletters,
newsletter, training annual membership meetings,
workshops, demonstration trade shows, catalogs,
projects, trade shows, catalog, business and information
insurance program, business brokering, site tours,
and information brokering, technology transfer project
video library, showroom

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 377

since these firms were, for all practical purposes, not part of the focal net-
works. Thus, guided by our basic theoretical concerns, we selected a sample
of firms within each network. We used a reputational sampling technique
(Laumann, Marsden, & Prensky, 1983; Scott, 1991) that better allowed us to
address our research questions than either random or snowball sampling (cf.
Scott, 1991). Random sampling is based on the assumption that all potential
respondents in a population are "equally important" for addressing research
questions (Holsti, 1968: 654), yet our initial network selection experience
demonstrated that not all network members could equally address our re-
search objectives. Snowball sampling builds on the connections of specified
respondents, asking initial respondents to name network members with
which they interact, then asking named members for their connections; this
process creates an interconnected sample by definition. In contrast, asking a
key network informant to select a sample of firms based on their reputations
as active and participating in the network ensured that the sample firms
could address the project's outcome questions, while allowing interfirm con-
nections to emerge rather than be predefined.
The executive directors for the two network administrative organiza-
tions, who were actively involved in forming and maintaining their respec-
tive networks, were our key informants for sample selection. All selected
firms had to have been network members for at least one year. After we had
described the purpose of the study, the research questions, and our criteria
for sample selection to the directors, they selected 22 firms that were active
in alpha-net and 23 in beta-net based on detailed searches of their member-
ship lists. These reputational samples represented 37 percent of alpha-net's
and 30 percent of beta-net's total memberships (i.e., 22 of 60 firms and 23 of
77 firms, respectively) and were consistent with estimates that active mem-
bers often represent only 25 to 30 percent of total network membership
(Lichtenstein, 1992). We then randomly selected 10 network firms (5 from
each reputational sample) to participate in the in-depth interviews that
would comprise stage one. Finally, 10 market firms listed in manufacturing
directories from the two network states (5 from each state) were randomly
selected to participate in the first-stage interviews. The 10 market firms
represented product lines and sizes that were similar to the network firms'
in their respective states and were only selected if they were geographically
close enough to the appropriate network administrative organization (i.e.,
within 100 miles) that distance was not a reason they were not in the net-
work. Table 3 provides descriptive characteristics of all respondent firms.
Stage One: Qualitative Data Collection
During stage one, we collected and analyzed qualitative data from ex-
ploratory interviews of network and market firms to provide the explana-
tions necessary for understanding a new phenomenon and to identify major
themes for the second-stage questionnaire. After pretesting interview proto-
cols with network firms not included in our study, we conducted one- to
two-hour interviews (in person or by telephone) with the two network ad-

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
378 Academy of Management Journal April

TABLE 3
Respondent and Firm Characteristics
Characteristics Alpha-neta Beta-netb Market Firmsc
Respondent position 17 CEO-owners 20 CEO-owners 7 CEO-owners
2 top managers 3 top managers 3 top managers
Mean number of years in position 11.9 11.2 13.5
Mean number of hours per month
on network activitiesd 7 5 0.50
Mean firm size
Employees 14 26 24
Gross salese 550 3,300 2,000
Range of firm sizes
Employees 1-120 1-125 3-150
Gross salese 60-2,600 10-20,000 100-12,000
Mean firm age in years 13 15 23
Number primary product lines
represented 4 12 4
Mean number of years in networkf 3.74 (6) 2.57 (5)
Mean number of miles to
network administrative organization 76 42 54
a
Three nonrespondents not included; N = 19.
b
Includes complete sample; N = 23.
CN 10.
d
For market firms, mean hours on professional organization activities.
e In thousands of dollars.
'Numbers in parentheses show the maximum possible.

ministrators and with the CEO-owners or the top managers of the 10 previ-
ously chosen network firms. Our interview protocols provided a framework
for exploring network outcomes and structures while allowing us to pursue
other issues seen as important by respondents. Market firm protocols were
then developed and interviews conducted with their CEO-owners or top
managers to compare their market-based interorganizational relationships
with the network firms' relationships.
Consistent with a replication logic, data analysis, coding, and concep-
tual categorization occurred after each interview in stage one. Systematic
procedures for qualitative data collection and analysis included transcribing
interview notes shortly after each interview, creating field notes to describe
processes and impressions, and identifying thematic patterns in respondent
comments (Glaser & Strauss, 1967; Miles & Huberman, 1994; Yin, 1989).
After identifying patterns in the qualitative evidence, we followed Eisen-
hardt's (1989) suggestion and returned to "enfolding" literatures for trian-
gulation and conceptual clarification of the patterns. For instance, we con-
sulted both practitioner research (Rosenfeld, 1993) and the organizational
literature (Venkatraman & Ramanujam, 1986, 1987) to clarify emergent out-
come findings. As Table 1 indicates, exploratory interviews identified major
themes for comparisons; these included network structures (administrative
and interactive) and network outcomes, such as four types of interorganiza-

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 379

tional exchanges-business, friendship, information, and competencies. In-

terviews also confirmed network and market differences. Our findings are
presented in a later section.
Stage Two: Quantitative Data Collection
In stage two, we administered a questionnaire to the full samples of
firms in both networks to explore and validate data patterns uncovered in
stage one and to collect quantitative data more accessible through question-
naires than interviews. Like previous qualitative researchers (e.g., Eisen-
hardt & Bourgeois, 1988), we thought that results would be strengthened if
we provided quantitative measures that could be readily compared across
organizations and networks. We built second-stage questions around first-
stage interview results to ensure that network structure and outcome mea-
sures were appropriate for the networks being studied. After pretesting the
initial questionnaire in a third, unsampled wood industry network and in 1
firm from each of our selected networks (to ensure item applicability), we
contacted each network respondent by phone to obtain agreement to par-
ticipate.
Using our questionnaire, we collected data on firm characteristics as
well as sociometric, or network structure, data. These structural data were
obtained by asking respondents to indicate with which other sample firms
they had had business, friendship, information, and competency exchanges
over the past year. We identified these four types of exchange from the stage
one interviews. Interrater reliability was achieved by using only those ex-
changes that were confirmed by each partner in a given link (see the Ap-
pendix). Consistent with our use of reputational sampling, we did not ask
survey respondents to indicate their exchanges with all firms listed in their
network membership rosters (60 firms in alpha-net and 77 in beta-net). Our
survey pretesting determined that time-intensive sociometric questions for
such extensive membership lists would greatly decrease response rates and
accuracy while generating data on firms that had little or no network in-
volvement. For comparing outcome data, we asked respondents to indicate,
from a 26-item list generated from our exploratory interviews, which out-
comes had occurred as a result of their network participation. We received
completed surveys from 19 of the 22 firms in the alpha-net sample (86%) and
from all 23 firms in the beta-net sample.
Descriptive statistics were used to capture questionnaire data on organ-
izational and individual attributes such as major product line, firm size,
outcomes, and respondent position. We then applied network analysis tech-
niques using UCINET (Borgatti, Everett, & Freeman, 1992) and KrackPlot
(Krackhardt, Lundberg, & O'Rourke, 1993) software to analyze the sociomet-
ric questionnaire data (the Appendix describes our network analysis mea-
sures). Researchers generally caution against using network analysis on in-
complete network data (Knoke & Kuklinski, 1982). However, boundary
specification is seldom obvious (Laumann et al., 1983), and much of the
extant network research could be criticized for either over- or underinclu-

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
380 Academy of Management Journal April

sion. We considered our procedure for network inclusion appropriate and

excluded firms that were members in name only. Nonetheless, we acknowl-
edge cautions concerning missing data in network research and recognize
that our findings do not necessarily represent relationships and outcomes for
all firms listed in each complete network.
We adopted a contrast-comparison approach (Miles & Huberman, 1994)
for analyzing and generating meaning from the questionnaire data. That is,
we searched for data patterns that revealed juxtaposing or contrasting struc-
ture and outcome characteristics within cases (i.e., network-network) and
across cases (i.e., network-market). For instance, we compared descriptive
statistics on the major product lines of member firms, examining domain
similarity (Van de Ven & Ferry, 1980) across networks. In addition, we com-
pared common network structural measures, such as local centrality and
graph density, to uncover differences in the number of highly central firms
in alpha-net and beta-net. Finally, we used a qualitative factoring procedure
suggested by Miles and Huberman (1994) to analyze outcome data.

NETWORK PARTICIPATION AND ORGANIZATIONAL OUTCOMES

Our first research question concerns what outcomes firms achieve from
network participation. The evidence revealed multiple outcomes, both eco-
nomic and noneconomic. Further, addressing our second research question,
the findings confirmed that these outcomes went beyond those available to
firms in traditional market relationships, supporting the value of small-to-
medium-sized-enterprise networks to involved member firms.
Outcomes in Small-to-Medium-Sized-Enterprise Networks
During interviews, we asked network respondents to describe what their
firms had achieved from network participation. Comments ranged from fi-
nancial performance outcomes ("We joined three years ago to expand our
sales of products. Last year our business expanded from a little over $2
million to well over $3 million") to outcomes that were clearly process
oriented ("I have learned about different ways of running the business that
I would have never thought of"). Although some short-term outcomes were
mentioned, most responses could be categorized as concerning an indirect or
long-range impact on performance, reflecting the long-term orientation of
network involvement discussed by Jarillo (1988), Powell (1990), Larson
(1992), and others. We identified four main categories of outcomes from the
network interviews: interorganizational exchanges, organizational credibil-
ity, access to resources, and financial performance. Table 4 presents quali-
tative findings illustrating these categories.
Interorganizational exchanges. The first outcome category, interorgani-
zational exchanges, refers to direct transactions or exchanges among net-
work firms, such as buying and selling, jointly producing and marketing a
product, and exchanging friendship and information with each other. For
instance, interview respondents indicated that although they had buyer-

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 381

Business
Friendship Information
Competencies
Interorganizational
Outcomes

exchanges
I I
The it
have met
each There's folks so new
major
shops found
a gallery. he
differentability I've
questions products
assemblewe
and other
[alpha-net]
I found
techniques
to as it. another
benefit metforout product
the makes had
shops.
to we Outcomes
share friendlier well
Thenthe
way couldn't as of
me. basically.
members. doing
about furniture
the through we similar Representative
could
tops
We ask all
We show
they're what both
I ideas
andboth maker
stronger kinds I
environment
know do. them so at Network
combine in anybody I of
[alpha-netl work Alpha-net
in market we the
make
theorganized
points each
andcannew theandtheon.
else.
in came
besthas workingaskcall...
other sell
We TABLE
the up Comments
meetings. 4
of
been andwith them any thebottoms
a of [alpha-net] withWe
a Participation:
anddesigned
several
different trust the

I
It's
we of to job
he's with now Interview
don't thingsAnotherdo There's share not
not a so sell
do furniture members
difference attitude.
this,
so set that similar
or to ... my
member [beta-net]
up I'mapproached certain Results
well. made are maker]
to it information.
not, and mine threatening
do you and instead
downfall
things amountthat for Representative
and.. of
and . myself members,
of I ask
feasible going me
he vice as [waste
for but to ... howto
. know Beta-net
can .. he
himversa cut call destroying
. differed he's wood]
to .. he'sso
help myopposed up it. to
to easygoing
hadhandles
megiveso
good
on that much throat" "do ... a other Comments
me at I [another
on so
type
things some kind friendship
want we of
I things how smaller

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
382 Academy of Management Journal April

Access
Financial to

Organizational
Outcomes
resources
performance
credibility

We The We
it's cost
small could Mostly
tripled have to shows
bottom ability
done business the
company notto
a our me.... university
shops
line, some promotion[alpha-net]
duefrom has do.talk
to a we forest
pretty
since led benefits
with
business, to through
goodwe small
members could thefrom Representative
and the established
job who radio notproducts
a
[alpha-net].
started do state,
for and
company
have Alpha-net
us. with Like catalog [alpha-net]
to
seriously TV
we a laboratory
which TABLE
and 4
grown relationship
come
spots
the ...
did.
individually.
with Comments
We their
improved
[alpha-net]. with from
trade which personally
the I
So our medium-size no as their
(continued)

We The
lot They
thisback
to little small been
joined ...
ourover involved
products. visibility,
amount
such
beginning.
in heresaying
[beta-net a of
$2 three By
new Last theis that
a company. PR
staff]
"here PR,
year
years .
million wood are .
. joining, that
contacts bank Representative
to ourago I
to the and high-visibility
that
well researched could
look [beta-net]
through
businessindustry."for Beta-net
over expandshows people be
us a was
$3 you. more
our .. partorganization
[beta-net]. can of
expanded getting
Comments
sales and
million, willingness from
of that ...
from to contact legitimate
duea came the
as it's
a be a
on

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 383

seller relationships with member firms and nonmember firms, they often
knew "what members can do . . . better than those outside [the network]."
Respondents from both networks also described joint business projects with
member firms. For instance, an alpha-net member stated this: "A couple of
us that met through [alpha-net] decided our products looked good together
so we had a joint show at the showroom. We sent mailings to both of our
mailing lists and, after some debate, shared our lists so we would know if we
had many overlaps. [The alpha-net director] advertised it for us as well ...
and we were pleased with the turnout." A beta-net respondent described
marketing his products through a beta-net catalog project: "[Beta-net] came
up with the money for a color catalog to market members' products. At first
we felt funny about being in a catalog with some competitors, since our
products were featured even on the same pages as competitors'! But we
stayed with the project and got some orders out of it." Respondents, par-
ticularly from alpha-net, also made it clear that the exchanges they described
did not occur prior to network participation and that their initial expecta-
tions were rather modest. As one alpha-net CEO commented, "There was a
certain level of fear that existed in me, that the network could be a negative
thing. But it was just a click of good ideas and sharing information, and
people saying 'How you do that?' or 'Where do I get that?' " Overall, al-
though exchanges did not occur equally among respondent firms, the quali-
tative evidence indicated clearly that network membership provided a group
of known firms with which members could and frequently did safely engage
in business, information, friendship, and competency exchanges.
Organizational credibility. The second category of outcomes uncovered
from our interviews, organizational credibility, refers to respondents' per-
ceptions that firm external legitimacy was enhanced through association
with the network. For instance, the network administrative organization, in
both cases a nonprofit organization, established relationships with local uni-
versities, large suppliers, and state agencies that smaller, individual firms
could not establish. In one example, the state agency representing the wood
products industry had no current database of manufacturing firms in the
state. Respondents described how alpha-net "established an industry pres-
ence" that generated state interest in updating the database and provided a
mechanism individual firms could use to show state decision makers "what
we have got." In a second example, beta-net's director arranged for a large
supplier to demonstrate new manufacturing technologies to network mem-
ber firms. Respondents indicated that previous attempts to obtain this in-
formation made by the small firms independently had been unsuccessful
because of "lack of company importance . . . these suppliers see us as un-
important players in the market. . . but when we contact them as a network
they see us differently." Thus, network respondents perceived that network
participation increased the visibility and credibility of member firms.
Access to resources. The third category of outcomes identified, access to
resources, refers to how network participation played an instrumental role in
respondents' accessing new markets, new product ideas, and other valued

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
384 Academy of Management Journal April

resources for their companies. For instance, respondents described how,

through network participation, they had been "introduced to a completely
new line of business," had "established contact with banks that lend to the
wood industry," and were "lined up with local and even national business
contacts."
Financial performance. Finally, respondents reported financial perfor-
mance outcomes in the interviews. Some respondents described economic
benefits that had occurred within a short time after joining the network: "As
a result of what's gone on with the connections I've made through [alpha-
net], and the company growth I've experienced, I've just bought the ten acres
in front of me so I can expand my operation." Other respondents described
a longer-term perspective on economic benefits: "I don't think that the main
emphasis of [beta-net] should be to bring someone's bottom line up imme-
diately. I think we should focus on ways to improve techniques, regardless
of how you are doing things now, and the bottom line will naturally come
up." In sum, the interview data provided ample evidence that network par-
ticipation benefited member firms across a range of financial and noneco-
nomic issues and that trust and long-term commitment pervaded network
thinking.
Outcome items. Drawing on the interview results, we then developed a
list of 26 potential firm outcomes resulting from network participation. This
list was included as part of the questionnaire completed by the full sample
of respondents in both networks. Table 5 presents the list of organizational
outcomes and the percentages of respondents in each network who reported
that a particular outcome had occurred. Overall, although outcomes were
not always achieved immediately, firms in both networks reported positive
results, supporting the interview findings. Outcomes were not equally strong
in both networks, however. On average, alpha-net respondents reported
60.77 percent of the outcomes as having occurred, and beta-net respondents
reported achieving only 48.38 percent of the same outcomes. This difference
is statistically significant at the .05 level on the basis of a t-test comparison
of means.
Obviously, our results would have been strengthened if we had had
access to objective data on outcomes. However, we can conclude with rea-
sonable confidence that over the broad range of outcomes respondents from
both networks reported as important, alpha-net members perceived network
participation as being more positive than did the members of beta-net. Beta-
net firms did report positive outcomes as a result of network membership,
particularly in certain categories. For instance, 83 percent reported they
were able to "generate new ideas for products or processes" as a result of
network affiliation. Nonetheless, across the full range of outcomes, beta-net
did not appear to benefit its members as much as alpha-net did.
Next, we analyzed these data using a qualitative factoring procedure
designed to identify "thematic commonalities" to reduce the list of 26 items
to a smaller subset of conceptually sorted outcome categories, or "factors"
(Miles & Huberman, 1994: 257). After identifying preliminary conceptual

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 385

TABLE 5
Frequency of Reports of Firm Outcomes Resulting from
Network Participation
Outcome
Outcome Type a Alpha-net Beta-net
Improved my firm's financial performance (e.g.,
sales, profitability, ROI) a 42% 52%
Improved my firm's operational performance
(e.g., product quality, new product
introductions) a 43 56
Generated new ideas for products or processes a 79 83
Improved my management skills f 48 48
Changed the way I organized work or people f 42 30
Added personnel to my business a 32 39
Accessed new equipment or production
processes a 63 43
Accessed new techniques or skills a 74 48
Accessed new contacts or suppliers a 89 74
Accessed new markets for my business a 74 61
Jointly marketed products or services with
network member firms f 52 48
Jointly designed or developed new products
with network member firms f 42 26
Jointly bid on projects with network member
firms f 37 39
Jointly produced a product or service with
network member firms f 42 26
Shared specialty services or technologies with
network member firms f 79 52
Discussed common problems with network
member firms f 95 70
Increased my interactions with competitors
within network membership f 79 43
Viewed competitors as potential resources for
my business f 84 43
Became a customer of network member firms f 42 52
Became a supplier to network member firms f 58 56
Referred jobs to network member firms f 89 61
Subcontracted jobs with network member firms f 26 30
Clarified my understanding of my firm's
competitive capabilities f 74 39
Gained "trade secrets" from competitors in the
network membershipb - 37 13
"Bought into" or accepted concept of
cooperation among firms (even competitors)
through a network f 74 61
Increased my firm's credibility through
association with the network a 84 65
Mean for all outcomesc 60.77* 48.38
Mean for transactional outcomesc 64.44 57.89
Mean for transformational outcomes' 60.19** 45.25

a
An "a" denotes a transactional outcome, such as enhanced resource acquisition or per-
formance gains. An "f" denotes a transformational outcome, or a change in the ways the net-
work firms' CEO-owners or top managers think and act. There are 9 transactional items and 16
transformational items.
b Item dropped from conceptual factor analysis.

The total number of items was 26; transactional = 9, transformational = 16.

* p < .05
** p < .02
This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
386 Academy of Management Journal April

factors, we searched the organizations literature for conceptual clarification

of these categories and finally returned to our outcome data to verify the fit
between the data and our emerging factors. We continued this process until
we determined that our enfolding literature (Eisenhardt, 1989) did indeed
improve our understanding of the outcome data and provided support for
our conceptual categorization.
Transactional and transformational outcomes. The qualitative factor-
ing and enfolding literature procedures revealed a strong conceptual fit with
Beije and Groenewegen's (1992) categorization of transactional and transfor-
mational network activities and outcomes. Transactional outcomes are en-
hanced resource acquisition or gains in performance. A beta-net respon-
dent's statement that network participation "opened my eyes to other mar-
kets" is an example. We identified 9 such outcome items; Table 5 identifies
these items. Transformational outcomes are changes in the ways the man-
agers of network firms think, act, or both. For instance, an alpha-net respon-
dent described how network participation changed his approach to working
with competitors: "Without [alpha-net], I would say 'I don't want to divulge
any of my secrets . . . these guys are my competition,' but there's a lot of
things that these other guys know a lot more about than I do . . . so that has
made [joining the network] worth it." A beta-net respondent described "a
joint bid I won with another member that led me in a whole new direction.
That [joint bidding] is not something I would've done without [beta-net]."
We identified 16 transformational outcomes. One item (the third-to-the-last
item in Table 5) was dropped from the qualitative factor analysis because it
was listed so infrequently by network members (only 13 percent of beta-net
firms) and because it involved using the network in ways that we viewed as
being potentially damaging to network viability.
Results from this second stage of analysis of outcome data revealed some
interesting findings. It was clear that firms in both networks did see benefits
from their participation in the network, but alpha-net firms found their
involvement to be advantageous regarding both transactional (64.44%) and
transformational (60.19%) outcomes. In contrast, beta-net firms appeared to
achieve more transactional (57.89%) than transformational outcomes
(45.25%). Beta-net firms thus appeared to view their network as an extension
of traditional market relationships and were less inclined to seek full ben-
efits of network membership, which would have required more direct inter-
action and sharing with other network members. Comparing results across
networks, the reported occurrence of transformational outcomes was signifi-
cantly higher among alpha-net firms than among beta-net firms (p < .02), and
there was no significant difference between the two networks for transac-
tional outcomes.
To summarize, our findings on outcomes indicate that member firms
appear to achieve substantial benefits from network involvement. Powell
(1990), Jarillo (1988), Saxenian (1994), and others have argued strongly that
networks are worthwhile despite their costs. Our work supports this view
but also goes further, indicating that both transactional and transformational

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 387

outcomes of network involvement occur. Although firms in both small-to-

medium-sized-enterprise networks studied reported favorable outcomes in
the interviews, the survey data indicated that, at least for transformational
outcomes, alpha-net firms benefited more. Our findings on outcomes are
stated as a formal proposition. Below, we explore in depth the details of
what network characteristics might be related to differences in outcomes, in
the section addressing the research question on network structure.
Proposition 1. Firms in small-to-medium-sized-enterprise
strategic networks are likely to achieve a broad range of
favorable transactional and transformational outcomes as
a result of their network involvement. Transactional out-
comes are most likely to be consistent across SME net-
works, whereas transformational outcomes are likely to
vary across networks.

Network-Market Firm Comparison

The second question that guided our research was whether outcomes for
firms in SME networks differ from outcomes for firms that are not network
members, or market firms. In particular, we were interested in discovering if
nonnetwork firms pursued ongoing interorganizational relationships within
traditional market arrangements that enabled them to achieve the same types
of outcomes as network firms. Table 6 presents findings from our interviews
with the sample of 10 comparable market firms. The evidence indicates that
the exchanges, organizational credibility, and access to resources available
to firms through the networks were not pursued through interorganizational
relationships by market firms. The market respondents indicated little or no
interest in developing ongoing interorganizational exchanges with competi-
tors, viewed organizational credibility as an issue for other firms, and rarely
if ever used competitors or other similar firms in their industry to access
organizational resources. We could not accurately determine whether finan-
cial performance was stronger for network or nonnetwork firms, although
the interviews made it clear that market firms achieved financial perfor-
mance through independent and competitive activities rather than through
cooperation with other firms.
Market respondents who interacted with competitors often did so "by
chance" or by "running into someone" at a supplier or trade show. Thus,
both network and market firms utilized contacts such as personal and busi-
ness relationships, but only network firms interacted regularly with com-
petitors and other, similar firms in their industry. Overall, our comparison of
network and market firms indicated that network membership provided an
external structure beyond market mechanisms through which network firms
generated new interorganizational exchanges, expanded organizational le-
gitimacy, and accessed organizational resources. This finding clearly sup-
ports the view held by Powell (1990), Larson (1992), and Provan (1993), that
networks are a unique organizational form and not simply an interorganiza-

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
388 Academy of Management Journal April

TABLE 6
Attitudes toward Cooperation among Competitors: Market Evidence
Cooperation Representative Market Company
Outcomes Responses Summary
Interorganizational exchanges
Referrals I am glad to refer, except in my line of
business. I never do it then.
Friendship I am friendly with all my customers ...
for one thing, I like to pick their
brains and get ideas on new sources
... but I am not friendly with
competitors.
Resources I use competitors as resources . . . when I
am in a crunch.
No, I never use competitors as
resources ... for one thing, you can't
make any money with the markup Few interorgani-
you have to add to their product. zation exchanges.

Expanded credibility I think joining [an association to improve

firm credibility] is more for the Low use of cooper-
companies that are high-volume and ation to expand
new-product-oriented. credibility;
I think using a network for increasing practice seen as
credibility would be mainly for the relevant to other
craft industry. firms.
Access to resources
New technologies Mostly from salesmen. They work the
whole state, so they get to know
what others are doing, and pass it on
... we get ideas that way.
New suppliers will find us at trade
shows.
New product ideas Some product ideas come from customers
... some from architect designs ...
we work off drawings and improve
on them, we build in our techniques.
Competitors' We learn about the capabilities of the
capabilities firms through the media . . . stories
and articles about what so and so is
doing.
Importance of external I probably know 90 percent of what I
links for accessing need to know, and am connected to
resources everyone my business needs ... .I
don't feel the need for [association
membership] anymore.
We are very busy . . . so the problem is
when do you find the time to join or Low use of cooper-
participate? If I saw a reason to join I ation to access
would, but I don't. resources.

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 389

tional contracting mechanism midway on the continuum between market

and hierarchy (Thorelli, 1986; Williamson, 1985). Our finding can be stated
as a formal proposition:
Proposition 2. Firms in small-to-medium-sized-enterprise
strategic networks are more likely than market (nonnet-
work member) firms to obtain outcomes related to increas-
ing interorganizational exchanges of friendship, informa-
tion, business, and competencies and increasing organ-
izational credibility and access to resources.

NETWORK STRUCTURE AND ORGANIZATIONAL OUTCOMES

The third research question focused on small-to-medium-sized-
enterprise network structure and the relationship between structure and the
outcomes of firms' network participation. Broadly stated, our findings indi-
cated that despite similarities in the structure of their administrations, the
roles of the two network administrative organizations and the interactive
structures of member firms differed substantially. Our evidence suggests that
these differences in structure could be attributed to differences in within-
network domain similarity. We were then able to draw on these findings on
domain similarity and subsequent network structure, as well as on our quali-
tative data, to help explain why the alpha-net firms reported significantly
more outcomes, particularly transformational outcomes, than the beta-net
firms.
Network Structure and Domain Similarity
One of the clear findings from our qualitative research was that small-
to-medium-sized-enterprise network firms made a sharp distinction be-
tween the network administrative organization, or the network-as-organiza-
tion, and the structure of interactions among member firms, or the network-
as-interaction. Although the network administrative organization and its
leadership were critical for the formation of the network and were active in
carrying out key network activities, its role was seen as somewhat separate
from that of network members and was viewed differently in the two net-
works. Table 7 presents interview data reflecting the opinions and views
held by the CEO-owners and top managers in each network on the network-
as-organization/network-as-interaction distinction.
Survey responses on links among network firms and with the network
administrative organization also confirmed the distinction between the net-
work-as-organization and the network-as-interaction. Our UCINET analysis
of confirmed responses on links (see the Appendix) indicated that both
networks' administrative structures were totally centralized around the net-
work central offices but that interactions among firms differed substantially,
both across firms and across networks. Total centrality meant that all the
firms surveyed had interacted with their central network offices during the
year of data collection. In network analysis terms, the administrative office

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
390 Academy of Management Journal April

TABLE 7
Differentiating the Network-as-Organization from the
Network-as-Interaction: Qualitative Evidence

Representative Alpha-net Comments Representative Beta-net Comments

I was thinking of [alpha-net] the I think the central office of [beta-net] is the
organization versus the operating of the most important ... then you kind of
members. The intermingling of the branch out from there . . . to get to know
members is different from [alpha-net] the the other members.
organization. The network is the
membership, and [alpha-net] is the spirit
behind the whole project.
Maybe what has happened in the genesis of I think what happens is that as the
[alpha-net] is we've made a distinction associations are made amongst members of
between [alpha-net] as a separate core [beta-net], you start to get the bypass
instead of being just one of the members. thing, which is really what you want ...
It has almost a separate direction ... to when firms start dealing with each other
promote the business, and not necessarily directly one to one, and the original
promote the [individual] membership ... organization starts to become a database
if you promote the business, you are manager or clearinghouse for information
promoting the membership. ... like [beta-net] is doing.
I saw alpha-net as an opportunity to It has perhaps worked better since the group
network, and meet other firms. The [member meetings] disbanded. When
meetings have been a lifeline .. . and the [beta-net] set up an office .. . we could go
[alpha-net] organization itself is a valuable in and work with the staff individually, as
asset as well, such as a hub on a wheel, opposed to working with the group ... in
with the spokes coming into it. meetings.

was the only node in both networks to have a 1.0 local normalized degree
centrality. In contrast, alpha-net and beta-net firms also interacted among
themselves, but much less than they did with the network central offices.
The range for all sample firms' local normalized degree centrality was 0 to
.67; thus, there was considerable variance in connectedness, and no member
firm in either network was nearly as connected as the network administra-
tive organization.
Although both networks had fully centralized administrative structures,
we found major differences in the roles of those structures in the two net-
works. Alpha-net's administrative office had run regular monthly meetings
for board members and the general membership since the network's incep-
tion. Alpha-net staffers also distributed newsletters, meeting agendas, and
other communications monthly and facilitated frequent group activities for
network members. In contrast, although beta-net decision makers initially
ran monthly meetings, within the first year they decided in favor of quarterly
board meetings and newsletters and annual membership meetings only.
They also decided to use a mechanism for coordinating member activities,
the netting circuit, in which network staff used facsimile and telephone to
contact individual network firms, rather than all members, regarding oppor-
tunities. Thus, the administrative structures for alpha-net and beta-net had

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 391

differing degrees of administrative coordination. Alpha-net's administrative

office used a high degree of administrative coordination through frequent
member firm interactions facilitated by its staff. Administrators at beta-net
used a low degree of coordination, through individualized, as-needed com-
munications and activities.
How can these large differences in the network administrative structures
of two small-to-medium-sized-enterprise networks in the same industry be
explained? What was apparent to us was that the two networks were com-
posed of different mixes of firms based on product line. What Van de Ven
and Ferry (1980) referred to as differences in domain similarity characterized
the two networks. The 19 sampled firms of alpha-net included only 4 pri-
mary product lines, representing all but 1 of the product lines sold by the
network's members: home furnishings and furniture, millwork, cabinetry,
and lumber remanufacturing; only the arts and crafts product category was
absent. Variance in product lines was even narrower, with 11 (58 percent) of
the 19 sample firms representing a single product line: home furnishings and
furniture. Given its homogeneity, the goal of alpha-net was clearly to aid
furniture manufacturing and sales activities. Alpha-net's head administrator
noted, "We targeted a specific product category, narrowly defined as furni-
ture. The dynamic of what we started then pushed it [network membership]
in that direction." Alpha-net members thus had substantial domain similar-
ity, with significant overlap in product lines, resource needs, competencies,
and potential customers.
In contrast, beta-net's respondents were far more heterogeneous; what
they sold included all 12 major product lines of their networks' members.
This included the primary business lines in alpha-net plus additional lines
such as high-end sports, marine, and garden products, and musical instru-
ment components. Despite being in the same general industry, beta-net firms
could be categorized as being more complementary than competitive, having
only partially overlapping resource needs, competencies, and potential cus-
tomers. As one network executive described the membership development
activities of beta-net, "[The network director] traveled around the region
contacting and recontacting all types of small and mid-sized wood compa-
nies listed in several directories . . . but he also visited many companies not
in the directories but just because he saw them during his travels." Beta-net
members thus had far less domain similarity than the firms of alpha-net.
For domain-similar networks comprised of potential competitor firms,
like those in alpha-net, high coordination appears appropriate, because
these firms must work together regularly to overcome their instincts to com-
pete rather than to cooperate. These firms require an external mechanism to
support and facilitate ongoing exchanges, especially at the early stages of
network development. In contrast, for complementary firms, like those of
beta-net, high coordination through a central administrative structure is
probably less critical and can even be viewed as something of an interfer-
ence. For instance, a beta-net respondent commented that "things ran better"
once monthly meetings were discontinued and she began interacting di-

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
392 Academy of Management Journal April

rectly with the [beta-net] staff" for business planning and "worked one-on-
one" with other beta-net members.
Some network administrative organization involvement is clearly
needed if a network is to act as a network. In the case of beta-net, the
administrative office and a key supplier were focal points for network ac-
tivities. The focal role of this key supplier had its roots in the period prior to
network formation. Observations and interview data indicated a history of
interaction among some beta-net firms and this supplier. For instance, beta-
net's director used existing state directories to begin an early network "mem-
bership drive," then asked company contacts "who they knew" who might
be interested in the network. In contrast, alpha-net's director had had no
current state directory to guide initial contacts, so early membership activi-
ties involved "getting out on the road" to "find out. . . who was doing what,
and where." Further, our market findings regarding the lack of interorgan-
izational relationships among our market sample underscores the small like-
lihood that alpha-net's domain-similar firms would have had interactions
prior to network initiation. Consequently, the beta-net firms' history of in-
teraction, which the alpha-net firms lacked, would have further reduced the
need for strong central control after beta-net was formalized. This was not
clear to the administrators at first, but resistance by beta-net firms caused
them to scale back their efforts at network-wide integration and coordina-
tion.
Differences in network structure were also in evidence at the level of
firm-to-firm interaction. We refer to this level as the network interactive
structure, which excludes links to the network administrative organization.
Figures 1 and 2 illustrate this interactive structure at the time of survey
administration for each network; structures are based on confirmed links. As
noted, we developed the network diagrams using UCINET software with a
plotting subroutine called KrackPlot. The figures also show firms' overall
exchange density values and local normalized degree centrality values for
the network interactive structures. Results indicate that the structures of the
two networks differed both in density and local centrality. Density among
alpha-net sample firms was twice that among beta-net firms (.25 versus .12),
and alpha-net included 7 firms with high local centrality-5 small furniture
manufacturers, a cabinetmaker, and a lumber remanufacturer had centrality
values ranging from .33 to .67. In contrast, beta-net included only 1 highly
central firm (centrality = .50), a relatively large raw wood broker/supplier,
and the next highest centrality score was .23.
In sum, at the level of member firm interaction, alpha-net was highly
integrated, with a large number of firms (7 of 19) highly involved in the flow
of activity within the network. In addition, alpha-net's administrative office
was very active in network coordination. Beta-net showed evidence of only
modest levels of firm interaction, and much of this interaction was through
a single supplier. Although beta-net's network administrative organization
was linked to all sampled firms, its role regarding the full network was less
active than that of its counterpart at alpha-net.

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 393

Overall Network
AC a18 a
Code 6:
Firm lal3:MI:500:17
Company
HF:
exchange
Label
na:O]
Key
density la9:HF:220:221
+
value Code HF: Total
= Product

.25.

Confirmed
Sales a2:HF:250:4
+
125:
Annual ja1O:HF:207
($000s)
Alpha-net

al
al2:MI1206:1
/ Links,
i
(.00) 56:
DegreeLocal
Centrality al
Normalized a18:HF:125:56 :Hna:67 FIGURE
1
excluding

1a11:MI:2,60:
:LR:990:44

W
a17:CA:115:33
MI LRHF CA
Administrative
Producta16:MI:130:61ja:F

Home
Lumber Code a3:HF:400:44a7H754
Millwork Cabinetry
(4) Key
(3)

furnishings Organizationa
(number
and lal5:HF:na:lll
of
remanufacturer
la8:CA:300:6alHFn17
(1)
firms)
furniture
a5
(11)
50

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
 394 Academy of Management Journal April

Overall Network
Code b2:
Firm lbl4:LO:na:O|
Company
bb4:LR:210:9
exchange ic Label

Key
density
Code WB: Total
Product lbl8:LR:20,0O:
value

.1 bD22:GA:1,180:23
2.
Sales
Confirmed
Annual3,600:
($O000s)

Beta-net

(.00) 50:
DegreeLocal Links,
Centrality
~~~~~+
Normalized
b2

exclud'ing
G
3:AC:1o:23 FIGURE
LRLOHFGFA AC 2
Product
Arts
It
Network
Home &
Lumber
Logging Garden Code
jb6:CA:4
Cabinetry
(1) Key
(3) crafts :1
b16:WC:13,000:5|
(2)
furnishings b11:MC:na:231
furnishings
and (number b2:WB:3,600:5
(1)
remanufacturer of lbl3:MP:2,00:91 Administratio
(6)
firms)
furniture
(1)

WB
WC SEMPMI MC
Organizationa

Wood
Wood
Sports
Marine
~~~~~~~3:CA:40:
Musical bA4
_b2l:MJ:60:91
Millwork
chips 131:b5:LR:3,800:0
broker (2) b8:MC:3,50:
(2)(1)
products lblO:LR:25:011OR40
equipment
(1) components
(1)
(2

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 395

In a network with relatively low domain similarity among members, like

beta-net, firms have the opportunity to meet their resource needs by repli-
cating traditional, nonnetwork relationships and by linking to such member
organizations as suppliers of critical resources. In contrast, networks of do-
main-similar firms appear to develop decentralized network structures, even
though such a structure means interacting with former or potential competi-
tors. In part, such interactions arise because firms have no real alternatives
within the network; in addition, they are more likely to view each other as
equals and to hope to gain strategic advantage through cooperation. Thus,
interaction among domain-similar firms tends to be spread out among many
firms rather than be concentrated in a single firm. This may mean, of course,
that the development of interactions among domain-similar firms in a net-
work will proceed rather slowly, as we found.
Works by Harrigan (1988) and by Oliver (1990) offer evidence that inter-
organizational asymmetries provide a stabilizing influence on interorganiza-
tional relationships, since asymmetric firms are likely to need each other for
critical resources, and they benefit by interactions with dissimilar organiza-
tions. These stabilizing asymmetries would not inherently exist among domain-
similar firms, which are more likely to be competitors. We argue that domain-
similar firms would enter networks with decision-making and resource sym-
metries that do not provide inherent network stability or ready-made links
among firms. Networks of "peer" organizations would thus need to develop
multiple stabilizing connections. For alpha-net, this requirement meant mul-
tiple firms having a position of centrality and a strong central coordinating
body. Our findings contrast with Walker's (1988) argument that centrality is less
relevant for understanding domain-similar interorganizational relationship
structures than for understanding domain-dissimilar ones. We conclude from
our results that domain similarity has an important impact on network struc-
ture, which can be stated as a formal proposition:
Proposition 3. Small-to-medium-sized-enterprise network
structure can be explained by the domain similarity of the
firms that compose a network. Specifically, a domain-
similar network is likely to have a strong central admin-
istrative structure and an interactive structure of high
density within which multiple firms are central. A net-
work having relatively low domain similarity is likely to
have a moderately weak central administrative structure
and a low-density interactive structure with few central
firms.

Network Structure and Outcomes

How do the characteristics of a network's structure help explain the

outcomes attained by member firms from network participation? Our results
(Tables 4 and 5) indicate that firms in both networks reported a wide range
of positive outcomes stemming from their involvement. Thus, it is not un-

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
396 Academy of Management Journal April

reasonable to conclude that small-to-medium-sized-enterprise network suc-

cess can be attained through a variety of different network structures. How-
ever, alpha-net firms did report more outcomes than firms in beta-net, and
more transformational outcomes in particular, which we believe can be ex-
plained by differences in network structure.
Research on the tie between network structure and outcomes has mostly
focused on the position of individual organizations within a network, pri-
marily local centrality, and on how such centrality might affect outcomes
(Barley et al., 1992; Walker, 1988). Provan and Milward (1995) offered an
alternative to this approach, looking more broadly at overall network struc-
ture and network-level outcomes. Our research aims at something between
these approaches. Like Provan and Milward, we looked at overall network
structure rather than at the positions of individual firms within that struc-
ture. Unlike these researchers, however, we were interested in understand-
ing how structure was related to the outcomes experienced by the firms that
comprised a network, rather than in outcomes, like client well-being, that
may be externally determined.
Provan and Milward (1995) provided evidence that networks achieve
higher effectiveness through centralized integration than through decentral-
ized integration. All firms sampled in both our networks were fully con-
nected to their respective network administrative organizations and, consis-
tent with Provan and Milward's conclusion, both alpha-net and beta-net
were effective. In addition, alpha-net, with the higher number of outcomes,
had an administrative office that was far more heavily involved in efforts to
coordinate and integrate the network than the beta-net office. Where our
results differ from theirs is that alpha-net was not only centrally integrated,
but was also integrated at the firm level, with high density scores and an
interactive structure that was far more extensive than that of beta-net. In
contrast, the network that Provan and Milward found to be most interactive
at the firm or agency level was least effective.
Alpha-net was successful because it had both strong administrative
links through its administration (the network-as-organization) and a strong,
largely decentralized interactive structure (the network-as-interaction).
These dual structures were critical for maintaining coordination and inte-
gration among the largely domain-similar entrepreneurial firms of alpha-net,
which were far more used to competing than cooperating. Provan and Mil-
ward's organizations were domain-dissimilar health and human service
agencies that were likely to cooperate to some extent anyway, even in the
absence of strong central coordination. If the firms in alpha-net were to
achieve any real benefit from network involvement, the links holding the
network together had to be strong. Our interviews led us to conclude that the
socialization process experienced by former competitors linked in a single,
strong network forces them to break old ways of thinking and to work hard
to benefit from network participation.
The effectiveness of beta-net was enhanced by interactions between the
network administrative organization and individual member firms, which

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 397

were appropriate in view of the heterogeneity of the firms in the network.

Beta-net firms simply did not have as much in common as alpha-net firms
did, and frequent group meetings to share common concerns and problems
were not viewed as critical for success. What also helped to pull the network
together, however, was the strong central position of a key supplier, an
advantage that beta-net firms would not have had through traditional mar-
ket-based buyer-supplier relationships. Despite this advantage, beta-net was
clearly less effective than alpha-net in achieving transformational outcomes
for member firms. Because firms had relatively low domain similarity, they
had less in common and were less likely than alpha-net firms to be changed
in significant ways. In general, the costs of lack of cooperation were less in
beta-net, but the benefits of cooperation were also fewer.
Alpha-net's interactive structure of multiple central, domain-similar or-
ganizations is consistent with descriptions of Italian industrial districts (In-
zerilli, 1990) as having, rather than single central firms, fluid relationships
and coordinating roles that flow back and forth among firms depending upon
opportunities and activities. Beta-net's structure of one highly central firm
among domain-dissimilar ones is consistent with Lorenzoni and Ornati's
(1988) case descriptions of the lead firm in a constellation of complementary
firms. Although both network models can produce positive outcomes for
members, each model is appropriate for a particular mix of firms, and the
types of outcomes attained will not necessarily be the same.
Based on our findings and insights on the relationship between small-
to-medium-sized-enterprise network structure and outcomes, we propose
that, in general, the level and range of outcomes attained by firms in net-
works can be explained by aspects of network structure. More specifically,
Proposition 4a. For domain-similar firms, numerous
transactional and transformational outcomes will occur
when a small-to-medium-sized-enterprise network's ad-
ministration is central and active in network-wide coor-
dination and when network interactive structure is dense
and decentralized.
Proposition 4b. For firms having low domain similarity,
numerous transactional (but not transformational) out-
comes will occur when a small-to-medium-sized-enterprise
network's administration is central and active in interac-
tions with individual member firms and when interfirm
exchange density is low but centralized.

CONCLUSIONS

Our overall conclusions are summarized in the simplified model pre-

sented in Figure 3. We have found that involvement in an SME manufac-
turing network can be advantageous for firms and have both transactional
and transformational outcomes. The model shows that the extent to which

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
398 Academy of Management Journal April

these outcomes will occur depends on the domain similarity of the firms in
a network and on how the network is structured. Our results indicated that
two aspects of network structure, administrative and interactive structure,
intervene to facilitate the process of network involvement in ways that can
lead to favorable outcomes for member firms. The type of network structure
that is most appropriate for achieving good results depends on the domain
similarity of the firms in a network.
Our model clearly links network structure and outcomes. Nonetheless,
we do not argue that structure alone, even when consistent with the domain
similarity of participants, is sufficient to produce good outcomes for member
firms. Both a network's administrative office and the firms themselves must
be committed to norms of cooperation and must work toward accomplishing
network, as well as firm, goals. We found high levels of commitment and
strong leadership in the administration of both alpha-net and beta-net, and
the firms we selected for study were obviously ones that were interested in
network involvement. Our study offers preliminary evidence that the struc-
ture of SME networks is important for network effectiveness, but primarily
as a way of facilitating interactions among firms (and network administra-
tors) that are committed to the network concept. Finally, our comparison of
network and market firms in the same industry demonstrates that interorgani-
zational relationships within purposefully constructed small-to-medium-
sized-enterprise networks are unique and that network involvement can
yield positive results that are not likely for nonnetwork firms.
We recognize certain limitations in our research design. First, our net-
work sampling illustrates the "boundary specification problem in network
research" (Laumann et al., 1983: 18) and brings to mind case study research-

FIGURE 3
Proposed Model of SME Network
Structure and Outcomes

Outcomes from
Domain Similarity t Network Participation
among Network Firms _ *Transactional
* Transformational

Network Structure
* Administrative
(network-as-organization)
* Interactive
(network-as-interaction)

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 399

ers' concerns over case representativeness (Miles & Huberman, 1994; Yin,
1989). Our network selection experience provided compelling reasons for
reputational sampling, which incorporated systematic and documented pro-
cedures to minimize nonrepresentative or biased results. Nonetheless, re-
sults are based on an involved subset of the firms in each network and could
thus lead to conclusions that would not apply to firms listed as being part of
the network but having little involvement.
A second limitation of the study is its inclusion of only two networks.
Obviously, we cannot claim that our results are generalizable to other net-
works, perhaps even within an SME network context. Nonetheless, we be-
lieve strongly that the relative paucity of comparative empirical research on
network outcomes and the near absence of such work in small-to-medium-
sized-enterprise networks justified our research design. Since the SME net-
work phenomenon is relatively new in the United States, there are no in-
dustries with large numbers of networks. By examining two networks in the
same general industry, we were able to make comparisons that might not
otherwise have been possible. In addition, careful selection and in-depth
comparison of two networks allowed us to build relationships with network
and firm managers and collect data that would not have been possible using
a less intensive approach. Our aim was to conduct a study that would draw
on both qualitative and quantitative data, not to test hypotheses, but to
develop insights that could form the basis of a general, though preliminary,
theory of small-to-medium-sized-enterprise networks.
Implications for Network Practice
Network planners and organizers should find it helpful to know that
firms can achieve multiple outcomes, both economic (transactional) and
noneconomic (transformational) from network participation. However, these
outcomes do not appear to be attained equally across all network forms, and
the timing of network success may also vary. For instance, the administrative
offices of networks of domain-dissimilar firms can take advantage of existing
links, familiar value-chain relationships, and the potential for complemen-
tary exchanges to build external network credibility rapidly. This is not
possible in networks of domain-similar firms, which are likely to need more
time and nurturing by the administrative office to achieve the trust needed
for network success. Despite advantages of timing, results from this study
indicate that relationships among domain-similar firms provided a broader
range of outcomes, including potentially longer-lasting organizational
change outcomes, than the relationships among domain-dissimilar firms.
Network planners and organizers attempting to work with domain-similar
firms must be willing to maintain a longer-term perspective and be willing
to work closely with network members.
A second implication for practice is that our work supports the key role
of a network administrative structure. Although network descriptions by
Jarillo (1988) and Thorelli (1986) depict networks as having separate admin-
istrative organizations, as they did in our study, this model is not universal,

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
400 Academy of Management Journal April

particularly in Europe. We suggest that in the traditional arm's-length U.S.

business environment, network administrative organizations are critical;
they can help overcome the lack of a tradition and infrastructure supporting
network relationships among firms. Such structures, absent in the United
States, do exist in Europe. Consistent with Saxenian's (1990) conclusion, our
study indicates how important a network administrative organization can be
even though its role in network integration may differ with the mix of par-
ticipating firms. Only when an administration is imposed on network firms
by outsiders, like government entities, is network effectiveness likely to be
undermined.
Finally, qualitative data from interviews with representatives of market
firms indicated that these firms only minimally used interorganizational
relationships. The market respondents provided multiple reasons for their
lack of participation in interfirm relationships, such as lack of time, no
perceived need, and fear of losing proprietary information. U.S. network
organizers face the challenge of overcoming these perceptions if they intend
to develop SME strategic networks on a larger scale.
In sum, this research supports the idea that strategic manufacturing
networks among small-to-medium-sized firms are a viable option to com-
petitive, market-based relationships. Because SME networks are new in the
United States, researchers have a unique opportunity both to study the phe-
nomenon as it is evolving and to have an impact on the forms it takes. This
study provides a framework for further research on this emerging topic.

REFERENCES
Baird, I., Lyles, M., & Orris, J. 1992. Cooperative strategies for small businesses: A comparison
of aligned and nonaligned firms. Conference Proceedings of the Midwest Division of the
Academy of Management: 220-226.
Baird, I., Lyles, M., & Reger, R. 1993. Evaluation of cooperative alliances: Integration and
future directions. Paper presented at the annual meeting of the Academy of Management,
Atlanta.
Barley, S., Freeman, J., & Hybels, R. 1992. Strategic alliances in commercial biotechnology. In
N. Nohria & R. Eccles (Eds.), Networks and organizations: Structure, form, and action:
311-347. Boston: Harvard Business School Press.
Beije, P., & Groenewegen, J. 1992. A network analysis of markets. Journal of Economic Issues,
26: 87-114.
Boje, D., & Whetten, D. 1981. Effects of organizational strategies and contextual constraints on
centrality and attributions of influence in interorganizational networks. Administrative
Science Quarterly, 26: 378-395.
Borgatti, S., Everett, M., & Freeman, L. 1992. UCINET IV version 1.0. Columbia, SC: Analytic
Technologies.
Bosworth, B. 1995. Interfirm cooperation: The points of intervention. Firm Connections, 3(1):
2-5.
Bosworth, B., & Rosenfeld, S. 1992. Significant others: Exploring the potential of manufac-
turing networks. Chapel Hill, NC: Regional Technology Strategies.
Brusco, S., & Righi, E. 1989. Local government, industrial policy and social consensus: The case
of Modena (Italy). Economy and Society, 18: 405-424.

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 401

Brush, C., & Chaganti, R. 1993. Competitive and cooperative strategies in new and small
ventures: A preliminary exploration. Paper presented at the Babson Entrepreneurship
Research Conference, Houston.
Coffey, M. 1995. Trade associations in the '90s: Knowledge centers and network brokers? Firm
Connections, 2(2): 1-7.
Cook, K. 1977. Exchange and power in networks of interorganizational relations. Sociological
Quarterly, 18(1): 62-82.
Cook, K., & Yamagishi, T. 1992. Power in exchange networks: A power-dependence formula-
tion. Social Networks, 14: 245-265.
Dess, G., & Robinson, R. 1984. Measuring organizational performance in the absence of objective
measures: The case of the privately-held firm and conglomerate business unit. Strategic
Management Journal, 5: 265-273.
Eisenhardt, K. 1989. Building theories from case study research. Academy of Management
Review, 14: 532-550.
Eisenhardt, K., & Bourgeois, L. 1988. Politics of strategic decision making in high-velocity
environments: Toward a midrange theory. Academy of Management Journal, 31: 737-
770.
Galaskiewicz, J., & Marsden, P. 1978. Interorganizational resources networks: Formal patterns of
overlap. Social Science Research, 7(2): 89-107.
Glaser, B., & Strauss, A. 1967. The discovery of grounded theory: Strategies of qualitative
research. London: Wiedenfeld & Nicholson.
Harrigan, K. 1988. Strategic alliances and partner asymmetries. In F. J. Contractor & P. Lorange
(Eds.), Cooperative strategies in international business: 205-226. Lexington, MA: Lex-
ington Books.
Hartwig, F., & Dearing, B. 1979. Exploratory data analysis. Beverly Hills, CA: Sage.
Holsti, 0. 1968. Content analysis. In G. Lindzey & E. Aronson (Eds.), The handbook of social
psychology: 596-692. Cambridge, MA: Addison-Wesley.
Inzerilli, G. 1990. The Italian alternative: Flexible organization and social management. Inter-
national Studies of Management & Organization, 20(4): 6-21.
Jarillo, J. 1988. On strategic networks. Strategic Management Journal, 9: 31-41.
Jick, T. 1979. Mixing qualitative and quantitative methods: Triangulation in action. Adminis-
trative Science Quarterly, 24: 602-611.
Knoke, D., & Kuklinski, J. 1982. Network analysis. Newbury Park, CA: Sage.
Krackhardt, D., Lundberg, M., & O'Rourke, L. 1993. KrackPlot: A picture's worth a thousand
words. Connections, 16(1&2): 37-47.
Larson, A. 1992. Network dyads in entrepreneurial settings: A study of the governance of
exchange processes. Administrative Science Quarterly, 37: 76-104.
Laumann, E., Marsden, P., & Prensky, D. 1983. The boundary specification problem in network
analysis. In R. Burt & M. Minor (Eds.), Applied network analysis: A methodological
introduction: 18-34. Beverly Hills, CA: Sage.
Lichtenstein, G. 1992. A catalogue of U.S. manufacturing networks. Gaithersburg, MD: U.S.
Department of Commerce, National Institute of Standards of Technology.
Lipnack, J., & Stamps, J. 1993. TeamNetfactor: Bringing the power of boundary crossing into
the heart of your business. Essex Junction, VT: Oliver Wright.
Lorenzoni, G., & Ornati, 0. 1988. Constellations of firms and new ventures. Journal of Business
Venturing, 3: 41-57.

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
402 Academy of Management Journal April

Marsden, P. 1990. Network data and measurement. In W. R. Scott & J. Blake (Eds.), Annual
review of sociology, vol. 16: 435-463. Palo Alto, CA: Annual Reviews.
Miles, M., & Huberman, A. 1994. An expanded sourcebook: Qualitative data analysis. Thou-
sand Oaks, CA: Sage.
Nielsen, N. 1993. Letter from Denmark: The network program concludes. Firm Connections,
1(1): 6.
Nohria, N., & Eccles, R. 1992. Networks and organizations: Structure, form, and action.
Boston: Harvard Business School Press.
Oliver, C. 1990. Determinants of interorganizational relationships: Integration and future direc-
tions. Academy of Management Review, 15: 241-265.
Perrow, C. 1992. Small firm networks. In N. Nohria & R. Eccles (Eds.), Networks and organ-
izations: Structure, form, and action: 445-470. Boston: Harvard Business School Press.
Pfeffer, J., & Salancik, G. 1978. The external control of organizations: A resource dependence
perspective. New York: Harper & Row.
Powell, W. 1990. Neither market nor hierarchy: Network forms of organization. In B. M. Staw &
L. L. Cummings (Eds.), Research in organizational behavior, vol. 12: 295-336. Green-
wich, CT: JAI Press.
Provan, K. 1993. Embeddedness, interdependence, and opportunism in organizational supplier-
buyer networks. Journal of Management, 19: 841-856.
Provan, K., & Milward, B. 1995. A preliminary theory of interorganizational network effective-
ness: A comparative study of four community mental health systems. Administrative Sci-
ence Quarterly, 40: 1-33.
Rosenfeld, S. 1993. Oregon's key industries program: Assessing the impact of manufacturing
networks. Chapel Hill, NC: Regional Technology Strategies.
Saxenian, A. 1990. Regional networks and the resurgence of Silicon Valley. California Man-
agement Review, 33(1): 89-112.
Saxenian, A. 1994. Regional advantage: Culture and competition in Silicon Valley and Route
128. Cambridge, MA: Harvard University Press.
Scott, J. 1991. Social network analysis. London: Sage.
Selz, M. 1992. Networks help small companies think and act big. Wall Street Journal, Novem-
ber 12: B2.
Shortell, S., & Zajac, E. 1988. Internal corporate joint ventures: Development processes and
performance outcomes. Strategic Management Journal, 9: 527-542.
Smith, K., Carroll, S., & Ashford, S. 1995. Intra- and interorganizational cooperation: Toward a
research agenda. Academy of Management Journal, 38: 7-23.
Stearns, T., Hoffman, A., & Heide, J. 1987. Performance of commercial television stations as an
outcome of interorganizational linkages and environmental conditions. Academy of Man-
agement Journal, 30: 71-90.
Thompson, J. 1967. Organizations in action. New York: McGraw-Hill Book Company.
Thorelli, H. 1986. Networks: Between markets and hierarchies. Strategic Management Jour-
nal, 7: 37-51.
Van de Ven, A., & Ferry, D. 1980. Measuring and assessing organizations. New York: Wiley.
Venkatraman, N., & Rarnanujam, V. 1986. Measurement of business performance in strategy
research: A comparison of approaches. Academy of Management Journal, 11: 801-814.
Venkatraman, N., & Ramanujam, V. 1987. Measurement of business economic performance: An
examination of method convergence. Journal of Management, 13: 109-122.

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions
1997 Human and Provan 403

Walker, G. 1988. Network analysis for cooperative interfirm relationships. In F. J. Contractor &
P. Lorange (Eds.), Cooperative strategies in international business: 227-240. Lexington,
MA: Lexington Books.
Wasserman, S., & Galaskiewicz, J. 1994. Advances in social network analysis. Thousand Oaks,
CA: Sage.
Williamson, 0. 1985. The economic institutions of capitalism. New York: Free Press.
Yin, R. 1989. Case study research: Design and methods. Newbury Park, CA: Sage.

APPENDIX
Network Analysis Measures
We calculated common network analysis measures, including local degree centrality and
graph density. Local or normalized degree centrality identifies the number of direct, adjacent
connections for each network member and is represented as a proportion between 0 and 1
(Scott, 1991). Density, or the "number of lines in a graph expressed as a proportion of the
maximum possible number of lines," also varies from 0 to 1 (Scott, 1991: 74).
Local centrality and density values were based on the number of links confirmed for each
sample firm and network administrative organization. We identified a confirmed link when a
firm reported engaging in the past year in one of the four exchange types (business, friendship,
information, and competency) with another firm and the other firm also reported the exchange.
If firm A reported an exchange with firm B but firm B reported no exchange in the past year with
firm A, no link was counted. Reciprocation rates calculated for all links were .90 (alpha-net) and
.92 (beta-net), representing high interrater reliability for linkage measures.

Sherrie E. Human is an assistant professor of management at Texas Christian Univer-

sity. She received her Ph.D. degree from the University of Kentucky. Her research
interests include interorganizational relationships such as strategic networks and fran-
chises, as well as new venture processes and the entrepreneurs involved in them.
Keith G. Provan is a professor in the College of Business and Public Administration at
the University of Arizona. He holds joint appointments in the School of Public Admin-
istration and Policy and the Department of Management and Policy. His Ph.D. degree is
from the State University of New York at Buffalo. Professor Provan's research interests
focus primarily on the effectiveness, structure, and governance of interorganizational
relationships and networks and on the analysis of health care systems.

This content downloaded from 139.184.14.159 on Thu, 01 Oct 2015 12:44:07 UTC
All use subject to JSTOR Terms and Conditions