Vous êtes sur la page 1sur 559

Corporate Social

Capital and Liability

Corporate Social
Capital and Liability

Edited by
Roger Th.A.J. Leenders
School of Management and OrganiZOlion
University ojGroningen, The Netherlands

Shaul M. Gabbay
Davidson Faculty of Industrial Engineering and Management
Technicn - Israellflsli/utl! o!Tedllw{ogy

Springer Science+Business Media, LLC
" Electronic Services <http://www.wkap.ol>

Library of Congress Cataloging-in-Publication Data

Corporate social capital and liability I edited by Roger Th. AJ.

Leenders, Shaul M. Gabbay.
p. em.
Includes index.
ISBN 978-1-4613-7284-4 ISBN 978-1-4615-5027-3 (eBook)
DOI 10.1007/978-1-4615-5027-3

1. Business networks. 2. Social nelworks. 3. Industrial

organizatian. 1. Leenders, Roger Th. A. J. II. Gabbay, Shaul M.
HD69.S8 C67 1999
302.3'5 - - dc21

Originali)' published by Kluwer Academic Publishers 1999
Softcover reprint of the hardcover 1st edition 1999

AII rights reserved. No part of this publicatian may be reproduced, stored in a

retrieval system or transmitted in any form Of by any means, mechanical, photo-
copying, recording, or otherwise, without the prior written permission of the
publisher, Kluwer Academic Publishers, 101 Philip Drive, Assinippi Park, Norwell,
Massachusens 02061

Printed an acid-free paper.


CSC: The Structure of Advantage and Disadvantage
Shaul M. Gabbay & Roger Th.AJ. Leenders

Section I
CONCEPTUAL ISSUES - Theory, Models, and Measurement
Organizational Networks and Corporate Social Capital 17
David Knoke
2 Social Capital of Organization: Conceptualization. Level of Analysis.
and Performance Implications 43
Johannes M. Pennings & Kyungmook Lee
3 A Relational Resource Perspective on Social Capital 68
Luis Araujo & Geoff Easton
4 Social Capital by Design: Structures. Strategies. and Institutional Context 88
Wayne E. Baker & David Obstfeld
5 Corporate Social Capital and Liability: a Conditional Approach 106
to Three Consequenses of Corporate Social Structure
lIan Talmud
6 Dimensions of Corporate Social Capital: Toward Models and Measures 118
Shin-Kap Han & Ronald L. Breiger
7 Organizational Standing as Corporate Social Capital 134
Patrick Doreian
8 Customer Service Dyads: Diagnosing Emperical Buyer - Seller
Interactions along Gaming Profiles in a Dyadic Parametric Space 148
Dawn Iacobucci

Section II
STRUCTURE AT THE INDIVIDUAL LEVEL - Social Capital in Jobs and Careers
9 The Sidekick Effect: Mentoring Relationships and the Development of
Social Capital 161
Monica Higgins & Nitin Nohria
10 Social Capital in Internal Staffing Practices 180
Peter V. Marsden & Elizabeth H. Gorman
11 Getting a Job as a Manager 197
Henk Flap & Ed Boxman
12 The Changing Value of Social Capital in an Expanding Social System:
Lawyers in the Chicago Bar. 1975 and 1995 217
Rebecca L. Sandefur, Edward O. Laumann & John P. Heinz
vi - Corporate Social Capital and Liability

Section III
STRUCTURE AT THE INDIVIDUAL LEVEL - Social Capital and Management
13 Generalized Exchange and Economic Performance: Social Embeddedness of
Labor Contracts in a Corporate Law Partnership 237
Emmanuel Lazega
14 CEO Demographics and Acquisitions: Network Effects ofEducational
and Functional Background 266
Pamela R. Haunscbild, Andrew D. Henderson & Alison Davis-Blake
15 Public Service Organizations - Social Networks and Social Capital 284
Ewan Ferlie
16 The Dark Side ofSocial Capital 298
Martin Gargiulo & Mario Benassi
17 Social Capital, Social Liabilities, and Social Resources Management 323
Daniel J. Brass & Giuseppe Labianca

Section IV
STRUCTURE AT THE FIRM LEVEL - Social Capital in Collaboration and
18 The Triangle: Roles ofthe Go-Between 341
Bart Nooteboom
19 The Management of Social Capital in R&D Collaboration 356
Onno Omta & Wouter van Rossum
20 Technological Prestige and the Accumulation ofAlliance Capital 376
Toby E. Stuart
21 Networks and Knowledge Production: Collaboration and Patenting in
Biotechnology 390
Laurel Smitb-Doerr, Jason Owen-Smith, Kenneth W. Koput & Walter W. Powell
22 Supply Network Strategy and Social Capital 409
Christine Harland

Section V
STRUCTURE AT THE FIRM LEVEL - Social Capital and Financial Capital
23 Choosing Ties from the Inside ofa Prism: Egocentric Uncertainty and
Status in Venture Capital Markets 431
Joel M. Podolny & Fabrizio Castellucci
24 Corporate Social Capital and the Cost ofFinancial Capital:
An Embeddedness Approach 446
Brian Uzzi & James J. Gillespie
25 Venture Capital as an Economy of Time 460
John Freeman

CSC: An Agenda for the Future 483
Roger Tb.A.J. Leenders & Sbaul M. Gabbay

• Introduction
CSC: The Structure of Advantage
and Disadvantage

Shaul M. Gabbay
Roger Th.A.J. Leenders

Scholars of the firm have long concerned themselves with identifying the
differential characteristics that make some firms-and some of their members-
more successful than others. A recent approach to the study of success and failure in
the competitive marketplace is the theory of social capital. The theory of social
capital suggests that players gain access to various kinds of resources that accrue to
them by virtue of their engagement in various kinds of relationships. Social capital
theory is fundamentally concerned with the resources inherent within structures and
social exchange. Until now, social capital theory has mainly been applied to
individual actors-human beings. In this volume the central question is how is
social structure related to the attainment of goals of corporations and their
members (denoted below by the terms 'corporate players' or 'corporate actors')? We
suggest that Corporate Social Capital refers to the resources, inherent in the social
structure, that accrue to corporate actors. Social structure refers to a network of
actors who are in some way connected via a set of relationships.
In the current chapter, we will briefly discuss the concept of social capital in the
context of organizations and introduce the concept corporate social capital.
Subsequently, we will discuss the relationship and distinction between social
structure and social capital. In so doing, we will introduce the notion of corporate
social liability and focus our attention on the various levels of aggregation at which
social structure, social capital, and social liability reside. We will coin the acronym
CSC to denote the interplay of social structure and social capital/social liability , both
at the firm, intermediate, and individual levels. In the concluding chapter of this
volume, we will discuss the research and practical applications of further
developments. We will highlight critical questions that, in our view, should be
2 - Corporate Social Capital and Liability

resolved. In our final discussion, we will open new questions for future discussions.
At its initial stages, CSC is an emerging research agenda. This, of course, presents a
wide window of opportunity for future and further contributions.


The term 'social capital' first appeared in scientific literature around a century ago,'
but it was not until the mid-1980's that a theory of social capital was developed and
applied in the context of work and organizations. James Coleman was the first to
develop a comprehensive theory of social capital and his work inspired the diffusion
of the use of the theory in relation to the study of actors who are pursuing interest
driven goals. Coleman's seminal work-Foundations of Social Theory (1990)
sparked a host of studies applying social capital theory.2 The work of Burt (1992)
was explicit in its focus on corporate actors. In his study of managerial mobility in a
high technology firm, Burt was the first to introduce a quantitative measure for
social capital. Studying White et al.'s (1976) suggestion that the absence of ties may
provide an advantage, Burt (1992) showed that actors who are connected to
disconnected others (structural holes) advance faster in the corporation under study.
These studies explicitly used the term social capital. Many social network
studies, that did not explicitly employ the term 'social capital' have also proven
important in the development of social capital theory. For example, the work of
Granovetter (1973), The Strength of Weak Ties, greatly influenced the study of
social capital as did the work of Lin (1982; Lin et al. 1981) and the work of
Laumann (1973). The theory of social capital has gained a prominent place in a wide
range of scientific fields, spanning social, economic, and political research agendas.
For a discussion of the history and use of the theory of social capital we refer to
Portes and Sensenbrenner (1993), Gabbay (1995, 1997), Nahapiet and Ghoshal
(1998), and Woolcock (1998).


Although a large body of research has emerged on social capital, and a growing
group of researchers is now using social capital in their research, consensus on the
definition of social capital has yet to be established. Some authors equate social
capital with social structure, whereas others refer to the resources an actor can
mobilize through the social structure. In the current volume, we specifically refer to
the assets embedded within and available through networks of relationships. We
define social capital as the set of resources, tangible or virtual, that accrue to an
actor through the actor's social relationships, facilitating the attainment of goals.
Before moving on to the discussion to 'corporate' social capital, it is important to
discuss briefly three elements of our view of social capital. First, we view social
capital as goal-specific. A large number of social ties does not necessarily translate
itself into social capital. It only does so if these ties assist the actor in the attainment
of particular goals. An example of this is provided by the following event. In
February 1996, Ben Van Schaik, the CEO of Dutch aircraft builder Fokker, gave a
presentation to a potential alliance partner. 'We are the second largest aircraft builder
in the world,' he said, supporting his claim with a colorful bar graph-the bars
CSC: The Structure of Advantage and Disadvantage - 3

represented various aircraft manufacturers; their length represented the number of

the manufacturer's clients. With this graph, Van Schaik showed the audience that
Boeing/McDonnell Douglas had 846 clients, Fokker had 225, and the rest trailed
behind at a large distance.3 Unfortunately, the number of ties did not represent the
number of aircraft sold, nor did it represent the credit rating of the customers.
Fokker's clients were primarily small companies, leasing only one or two airplanes
(rather than buying them), and many of them 'forgot' to pay their bills. Fokker
declared bankruptcy only a few months later and was liquidated. Its 'much smaller'
competition is still alive and kicking. 4 Social network and social capital are different
things. A social network only conveys social capital if its social ties are beneficial
for the attainment of goals. In the case of Fokker, they clearly were not.
Second, an actor need not be conscious of the social capital he enjoys. The
social structure in which an actor is embedded may confer advantages to the actor,
without the actor even realizing it.
Third, the social structure that brings opportunities for the realization of
particular goals need not have been built in the pursuit of these goals-social capital
often is a by-product of other social activity.
Like other forms of capital, social capital is productive, making possible the
achievement of goals that would not be attainable in its absence. This goal
specificity has a number of implications.6 First of all, social structure may be
beneficial for the attainment of multiple goals, since the multiplex character of many
social relationships results in overlap of opportunity structures. Networks created for
one purpose may be employed for another-which often was not foreseen when the
actors initially engaged in a relationship with each other. In some situations, the
same social structure can be beneficial for the attainment of one goal, while
obstructing the attainment of others. 7 Social networks can have a positive effect in
providing network members with access to privileged resources, while lowering
transaction costs, but can, simultaneously, place high demands on these network
members, restricting their individual behavior and opportunities.
In this volume, we concern ourselves with corporations and their members. s
Corporate social capital, then, refers to:
The set of resources, tangible or virtual, that accrue to a corporate player
through the player's social relationships, facilitating the attainment of goals.
Although social structure and social capital are often equated in the social
capital literature, they are different entities. In this volume we therefore make an
explicit distinction between social structure per se and the outcomes of social
structure. When these outcomes are positive, helpful in attaining specific goals, we
say the social structure conveys social capital. But when social structure prohibits
and obstructs action, it produces social liability. Although the absence of social
structure precludes social capital from corning into existence, the two are distinct.
They are generated by related, but distinct, processes.


Just as bank accounts can run dry and can have a negative balance, social capital can
vanish. Social structure that provided social capital in the past may not necessarily
4 - Corporate Social Capital and Liability

do so in the present. The social structure required to sustain corporate social capital
may shift as transactions, activities, and conditions change and become more or less
complex. Relationships beneficial to the achievement of goals in the past may thwart
goal attainment in the future. In their study of how managers adapt to changing
environmental conditions, Gargiulo and Benassi (this volume) find that relational
structures that in the past had provided ample social capital for managers, later
increased the number of coordination failures for which these managers were
responsible. The network no longer provided resources, but had become a constraint,
impeding performance.
Brass (1984) and Blau and Alba (1982) found that relationships to the clique of
top executives in an organization had a strong positive relationship with power and
promotions. Top executives were likely to have more social capital in the form of
more (relevant) information to share with those who were connected to them. Since
men are more likely to maintain relationships to top executives than are women,9
women may be forced to forgo any preference for homophily in order to build
connections with the dominant coalition and share the social capital. Brass and
Labianca (this volume) therefore conclude that actor dissimilarity (in this case based
on gender) may affect interaction patterns and may consequently exclude some
people from sharing in the social capital. From the viewpoint of human resources
management, Brass and Labianca provide ample examples of the positive and
negative effects social relationships yield on social resource management outcomes
such as recruitment, socialization, turnover, job satisfaction, power, and conflict.
Leenders (1995) shows that the relations among social service workers may help
in preventing these workers from becoming burned out, but that feelings of burnout
are also contagious through these exact same relationships. Increases in burnout
experienced by social service workers also increased the level of burnout
experienced by their co-workers. Conversely, decreases assisted in decreasing
burnout among alters.
Gabbay (1995, 1997) shows that, for some actors, strong ties combined with
structural holes ('structural ports') were beneficial at the inception stage of their
business, but were detrimental for future expansion. Successful entrepreneurs
strategically changed their networks over time in order to maintain or build social
structures that would be rich sources of social capital.
Social structure translates into social liability in at least two different ways.
First, ongoing, strong social relationships may constrain the behavior of actors,
impeding their action and attainment of goals. IO For example, long-standing
relationships with customers may stifle the firm by monopolizing a disproportionate
share of its resources, inhibiting the firm from forming relationships with alternative
customers. Second, actors may be unfavorably affected in their opportunities by
negative ties in the social structure. In this volume, Brass and Labianca explore the
effects of this type of social capital. As an example of the social liability stemming
from negative ties, they argue that 'it is likely that an actor's negative ties within an
organization will prevent promotion, particularly if those negative relationships are
with influential others. Others may withhold critical information that worsens an
actor's performance or they may provide bad references in order to prevent a
promotion' (this volume: 324).
CSC: The Structure of Advantage and Disadvantage - 5

Organizations and their members continuously engage in social relationships.

Effective organization requires a constant balancing of the potentially opposing
forces inherent in social structures.

Relational structures can be recognized at different levels of analysis and
observation. Four levels of analysis normally characterize organizations. \I The
individual human being is the basic building block of organizations. The human
being is to an organization as a cell is to a biological system. One step higher, we
find group or departments, where individuals work together on group tasks. Next is
the organization itself: a collection of groups or departments. Above this level,
organizations themselves can be grouped into an interorganizational network. 12
Social capital and social structure are relevant at all of these different levels. Classic
methodological approaches require a researcher to choose one particular level of
analysis as the primary focus of a study. However, as we will discuss below, the
very nature of social capital runs through all these organizational levels. Social
structures at the individual level translate into social capital and social liability for
individual actors, but also translate into social capital effects at higher levels of
analysis. Similarly, the structure/capital connection also works its way down the
levels of analysis. A full study of social capital should thus incorporate structure and
capitallliability at mUltiple levels of analysis.

Relations Among Individuals

Intraorganizational networks have been the locus of exuberant research efforts. The
effects of network structure on, among many other things, the distribution of power,
job satisfaction, career opportunities, and productivity of individuals have been
central on the research agenda. Undoubtedly the most attention has been given to the
sources and consequences of the distribution of power in organizations, focusing on
such concepts as prestige, status, and control. Next on the agenda has been the
related topic of information flows in organizations. This type of research addresses
questions related to 'who has what information when and how'? Considering the fact
that we are only at the start of the information age, and are pushed along by rapid
developments in information and communication technology, this latter question is
likely to remain among the most popular objects of study for a long time.
The distribution of power and information is both an outcome and generator of
network structure. As an outcome, it portrays (part of) the social capital (and social
liability) that employees extract from their relationships. As a generator, social
capital affects intraorganizational structure. For instance, by the very nature of his
work, a top executive has many relational ties, spread throughout the organization,
and is at the intersection of numerous flows of information. This network position,
characterized by high centrality and the presence of structural holes, provides the
manager with ample opportunity for control, power, and influence: the manager
draws social capital from his network structure. In turn, his power gives him the
opportunity to surround himself with a network of loyal subordinates and use his
network ties to stay informed about events throughout the organization. By placing
trusted friends in strategic positions, a manager gains power by being well-informed,
6 - Corporate Social Capital and Liability

by having access to other (important) people in the organization, and by having

mUltiple people depend on him. Capitalizing on his social capital, the manager
effectively reengineers his position in the network. Ideally, this will then allow him
to harvest more social capital from his (new) network, or secure the capital drawn
from the old structure-social capital creates social capital.
Social structures among individual members of the firm relate to social capital
at various levels of analysis. First, intraorganizational networks yield social capital
for the individuals comprising these networks as individuals employ their direct and
indirect ties to fulfill individual goals. Occupying a central position in various
networks of relationships is often regarded as a source of power--power is social
capital to those who can utilize it for getting the job done. Having a tie with a
mentor can be both helpful and detrimental in securing jobs (see the chapter by
Higgins and Nohria in this volume). Social ties may help individuals to find (more
favorable) jobs both in the framework of internal market opportunities (Marsden and
Gorman, this volume) and in the context of external labor markets (Flap and
Boxman, this volume). In particular, having access to higher-status individuals is
often argued to be beneficial in the job search process.
Second, social structures at the level of individuals also impact social capital at
higher levels of analysis. For instance, Kratzer, Van Engelen, and Leenders (1998)
discuss how the structure of various types of relationships among members of R&D
teams affects their success. They contend that segmentation of the problem-solving
network among the team members is detrimental to the performance of the team, an
effect stronger for teams performing highly complex tasks than for teams
performing tasks of low complexity. R&D teams that agree on their basic product
development goals-but are characterized by a reasonable disagreement on how to
achieve these goals-tend to develop products with a much higher probability of
market success than teams whose members fully agree (or disagree) on these
issues.13 Based on the studies focusing on social contagion,14 it is possible to
conclude that the structure of social relationships between individuals affects their
level of consensus and, consequently, their group performance. The structure of
these relationships thus affects outcomes at the firm level.
Social structures at the individual level of observation can also have negative
effects at these higher levels of analysis. For instance, consider the so-called
grapevine-an informal, person-to-person communication network of employees
that is not officially sanctioned by the organization. Grapevines are sources of
rumors and gossip that spread quickly throughout an organization. Often,
management decisions circulate through grapevines days ahead of their official
announcement. 15 Because they feel threatened by it, managers often try to suppress
the grapevine, but find themselves confronted with a nearly impossible exercise.
Another example is related to the resilience of personal networks. Managers in
charge of (re)designing business processes often experience difficulties in breaking
through the power structures that exist between the firm's employees. 16 As Knoke
(this volume: 40) notes, 'power structures are highly stable and resistant to change.'
In addition to relationships with others within a particular firm, individuals also
maintain relationships with those outside of the firm. The relationship a lawyer
maintains with her clients often directly translates into status, financial turnover, and
CSC: The Structure of Advantage and Disadvantage - 7

profitability at the law firm level. In effect, the firm draws social capital from its
employee's contacts (who also draws social capital from the relationship herself).
Interlocking directorates connect firms-they lead to information sharing and policy
binding and can be a strong source of organizational power.17 In the chapters by
Knoke and by Pennings and Lee, the boundaries differentiating roles of individuals
are discussed extensively.

Relations Among Organizations

No organization really stands on its own. Almost every organization needs suppliers
and buyers, and deals with intermediaries. Ongoing exchange relations with other
organizations translate into more or less stable interorganizational networks.
Drawing from theories as diverse as contingency theory, resource dependence
theory, transaction cost theory, and population ecology, scholars are collecting data
and devising explanations of interorganizational relationships. Unfortunately,
however, these theories hardly devote any attention to how these interorganizational
relationships lead to a network of relationships. 18 They also ignore how these affect
opportunities and restrictions for the firms comprising these networks.
Recent trends influence the extent and intensity of the engagement in
interorganizational relationships. The unification of Europe has led to a surge of
mergers, alliances, cooperation, and licensing. More than ever before, European
companies are looking at the structures around them, actively investing in new and
reassessing old ties. Advantages are not necessarily in the creation of a large bundle
of ties, but in the creation and maintenance of relations that provide access to new
technologies, resources, and legitimacy.19 In other words, companies are actively
molding their networks so as to draw as much corporate social capital from them as
possible, with a long-term focus.
Europe is not the only stage where interorganizational relationships are crucial.
With strong trends toward globalization in virtually every sector, and with rapid
developments in information and communication technologies, it is possible and
necessary for many firms to enter into intercorporate relationships.
Just as individuals can tap social capital from their social networks in order to
facilitate some action or attainment of goals, so too can organizations extract
resources from their networks. Harland (this volume) shows how corporate social
capital is created through the buildup of strategic interorganizational supply
networks. Using the cases of Benetton and Toyota she illustrates how firms, that are
seeking to increase competitive advantage, actively pursue the role of a 'supply
network hub' to facilitate effective and efficient flows of materials and information.
Smith-Doerr et al. (this volume) examine the link between social structure and
intellectual output in the context of the biotechnology industry, in which
interorganizational collaboration is commonplace. Their interest is in the number of
patents organizations obtain as a result of their collaborative relationships. For
instance, they argue that firms that are more central in their collaboration network
will attain more patents in a timely manner, due to their deeper insights into the
knowledge-base of the field and their ability to combine and utilize a broad
configuration of partners' expertise in choosing promising lines of research and
framing patentable claims (this volume: 394). Not only does a favorable position in
8 - Corporate Social Capital and Liability

the network (as measured by centrality) lead to the generation of more patents, but
receiving more patents, in turn, also leads to an even more favorable network
position. In other words, social capital is both an outcome and a generator of social
Stuart (this volume) also studies the social capital that can be gained from
alliances. In concert with Smith-Doerr et al. (this volume), patents playa central role
in his study. Stuart's interest is in whether the status of an organization (as a
function of the organization's position in the patent citation network) has an effect
on the rate at which the organization acquires technologies from other firms.
Technologically prestigious firms are found to be able to access the technological
assets of other firms at a higher rate than are firms of lesser prestige. Interestingly,
the rate at which a firm acquires technology from its competitors is more strongly
positively related to the firm's technological prestige than is the rate at which the
firm supplies technology. In this sense, high-status firms gain a positive balance of
social capital from alliances with lower-status partners by taking more technology
than they give. Being at the other end of the relationship, the lower-status firm sees
some of the status of its partner reflected on its own technology, and draws a
different type of social capital from the same relationship.
The question of how status relates to social capital is also the focus of the
chapter by Freeman (this volume). Freeman shows that start-up businesses that
receive support from centrally connected venture capital firms (VCs) have an
increased probability of an initial public offering. In Freeman's study, the structural
position of one organization (the VC) produces social capital for another (the start-
up business). But status also brings a risk. As the analyses show, VC centrality also
increases the probability of a startup business being acquired. Here, the structural
position of the VC creates (potential) social liability for the entrepreneur. The VCs
themselves also experience risks. Freeman suggests that more central, high-status
VCs behave like option traders by making high-risk investments and walking away
from the losers. This could undermine the deal flow advantages that centrality
affords alter the central position of the venture capital firm.
Uzzi and Gillespie (this volume) are concerned with the question of how social
structure affects the probability of a small business securing capital. In partiCUlar,
they examine the relationship between social structure and lending practices. They
show that small businesses garner more loans at lower interest rates by increasing
the duration and multiplexity of their relationships with financial institutions. Their
findings clearly show that by strategically engineering social structure and by
creating a proper mix and intensity of ties to financial institutions, small businesses
can gain much social capital (in the sense of securing capital at decent prices).
So far, we have only concerned ourselves with the relationship between
interorganizational networks and social capital at the organization level. But these
structures also yield social capital at the individual level. For example, alliances or
consortia can provide employees with the opportunity to be trained or temporarily be
located at other organizations. Employees can also more easily change employers,
moving between firms within one alliance. With employability becoming
increasingly important in the workplace, these interorganizational relationships
CSC: The Structure of Advantage and Disadvantage - 9

Structure Structure
firm level individual level

Figure 1. esc framework: Interdependencies between social structure and social capital at
various levels of analysis

provide the individual employee with the social capital of learning new skills and
the opportunity to display his skills beyond the boundaries of the ·firm he is
(currently) affiliated with. Obviously, this also creates potential social liability for
the organization, as it may lose valuable employees. In this volume, Higgins and
Nohria study career opportunities of employees moving between companies.

Mixtures of Levels
Employees of every firm are involved in social relationships that extend beyond
their own firms. Consequently, firms themselves are part of wider social and
economic networks that can be expected to influence their relationships with
potential suppliers and customers.
Many relationships between organizations are mediated by individuals. Lawyers
bring their clients to their law firms. Formal relationships between firms often start
out as informal, personal relationships between representatives of these firms and
then become institutionalized. But not all interorganizational relationships are
mediated by individuals. As Pennings and Lee (this volume: 50) note 'as a legal
entity, the firm is capable of contracting, of acting as a partner in any market
relationship, including the setting up of joint ventures, the acquisition of another
firm, or the shedding of a business unit to other firms, etc .. '
An important aspect of the institutionalization of ties maintained by the firm's
members is that it shifts the agency of the social capital to the firm level. A
consultant, whose social contacts bring in large revenues for his firm, may try to
keep the relationship 'to himself.' By retaining the relationship in his personal
portfolio, he has a powerful means to strengthen his position in the firm. A manager
whose contacts lead to a formal, contractually regulated alliance transfers his claims
to the relationship (and the social capital inherent in it) to his organization.
10 - Corporate Social Capital and Liability

An organizational risk attached to the personal nature of relationships is that

these relationships may be lost when the employee leaves the firm. Consultants may
take their contacts with them to their new employer. 20 Organizations search for ways
to appropriate individual-level social capital, often by developing a reward system
that favors employees who bring in the most social capital. Salesmen often are
rewarded in proportion to their achievements in terms of sales and revenues;
successful consultants are offered the opportunity to become a partner in the firm.
Relational structure may have both positive and negative consequences for the
attainment of goals of corporate players at either level of analysis (see Figure 1).21
Strong inter-firm relationships (firm level structure) can provide the firm (firm level
social capital) with corporate social capital-they provide the firm with resources
while lowering risks and costs of opportunism. At the same time, these ties may
provide firm members (individual level social capital) with social capital by giving
them access to a number of other firms to which they can relocate to further their
individual careers. As a result, the firm is confronted with the corporate social
liability of seeing its valued employees leave (firm level social liability).
Relationships of individuals of firm members with others outside the firm
(individual level structure) can provide the firm with valuable clients (firm level
social capital). The ties among individuals in teams and departments (individual
level structure) affect their job satisfaction (indidividual level social capital) and
affects the productivity of the group (firm level social capital). As argued above, a
full study of social capital should deal with the association between structure and
capitalIliability at mUltiple levels of analysis.

To summarize, we have the following elements:
• Social structure and social capital are related but distinct entities;
• Social capital represents the resources that accrue to an actor through the actor's
social relationships, facilitating the attainment of goals. When social structure
hinders the attainment of goals, it yields social liability;
• 'Corporate social capital' refers to the social capital of corporate players: firms
and their members;
• Social structure and social capital can be distinguished at different levels of
analysis, at minimum at the levels of the firm and the individual;
• Social structure and social capital are not only associated within the same level
of analysis, but are outcomes and generators of each other at all of these levels;
• Social structure can provide social capital to one player, but social liability to
another. It can provide social capital for the attainment of one goal, but social
liability for another goal.

We have found it useful to use the acronym 'CSC' to denote the constellation of
{structure, capital, liability, level of analysis}. In this volume, we will employ this
acronym in the current chapter and in the concluding chapter.
CSC: The Structure of Advantage and Disadvantage - II


The study of social networks has been the focus of the work of organizational
scholars for over a decade. A growing body of studies from various areas and
disciplines has focused on the study of social networks in the context of
organizations. However, Salancik (1995) recently noted that the wealth of social
network studies and methodology has not yet produced an encompassing theory of
organization. He therefore suggests that 'a network theory of organization should
propose how structures of interactions enable coordinated interaction to achieve
collective and individual interests.'22 Theories of corporate social capital present a
(partial) answer to this call. CSC theory attempts to explain how social structure is
connected to organizational outcomes. It explicitly relates to (the interplay of)
structures at the individual level and the firm level, and to how these affect
attainment of relevant organizational goals. This invokes questions such as: What
kinds of social structures bring what kinds of outcomes? Which networks are good
for what actors and under what conditions? When are networks vessels for social
capital and when do they render social liability? The fundamental difference
between the study of social networks and organizations in general and the
overarching agenda of CSC is that, within the framework of CSC, social network
structures are the focus of attention in their explicit productive or destructive
association with goal attainment of corporate players. With this approach, CSC
shifts attention to more daring questions. We move from a descriptive set of findings
to a theoretically based orientation with practical implications. A CSC theory
provides a different framework for understanding organizations.

Organizational Paradigms
The study of corporate social capital suggests an analytical framework that cuts
across other theoretical frameworks that are eminent in the literature on
organizations--for example, Scou's (1992) paradigmatic typology of organizations
as rational, natural, and open systems. 23
The rational paradigm perceives organizations as 'collectives oriented to the
pursuit of relatively specific goals.' The basic assumption of this paradigm is that of
goal-oriented rationality. Rationality suggests a cost-benefit analysis--a balance
between the interest of actors, costs, and payoffs. The rationality-based premise of
our definition of corporate social capital is straightforward because of the explicit
focus on the functional (negative or positive) aspects of social structures in their
relation to the goals of corporate actors. Social structure in the framework of CSC
has costs and benefits and is directed towards the attainment of goals.
The natural paradigm perceives organizations as 'collectives whose participants
share a common interest in the survival of the system and who engage in collective
activities, informally structured.' It is largely in the framework of social networks
that collectives operate. These social networks can be measured, mapped, and
related to outcomes-at the firm and individual levels. The development, creation,
and maintenance of these social networks to fit corporate goals are what ultimately
transform these networks into sustained corporate social capital. esc is thus
explicitly related to the notion of organizations as natural systems. In the natural
paradigm, the survival of the system is a goal-the challenge in studying corporate
12 - Corporate Social Capital and Liability

social capital is to explain which social structures make the achievement of this goal
easier or more difficult. 24
The open system paradigm sees organizations as open systems 'embedded in-
dependent on continuing exchanges with and constituted by-the environment in
which they operate.' The open system perspective is a natural context for the theory
of corporate social capital, which emphasizes social structure in the relevant
environment of firms and their members. An important asset of corporations lies
outside and beyond their immediate boundaries. Firms are embedded in
interorganizational networks, affecting the goal attainment of corporate players. A
firm's environment is made of opportunities and potentially creative relationships
that may further the goals and achievements of corporate players. From a social
capital perspective, the boundaries of firms fade into their environment-
corporations are embedded in a networked environment. The relationships between
(members of) the firm and players outside the firm's formal boundaries affect the
efficiency and effectivity of the firm's internal organization and, at the same time,
can be modified to suit the firm's goals. The ramifications of the existence of
'external' ties on corporations are at the heart of CSc. 25
Placing the notion of corporate social capital in the paradigmatic discourse of
organization studies suggests that CSC contributes a new framework and an
additional dimension to the study of organizations and organizational processes.

It is common sense that relationships are important, both for individuals and for
corporations. It is also intuitively clear that not all relationships are (equally) useful.
Unfortunately, it is largely unknown which relationships, under what conditions, are
beneficial or obstructive to organizations or their members. It is often argued that
part of the explanation of which corporations (individuals) do better than others is
found in the relationships these corporations (individuals) maintain. But we are not
yet sure what exactly it is in those structures that benefit some and impede others.
The world is becoming increasingly complex and dynamic. Organizations must
incorporate even greater diversity to survive and thrive. More complexity compels
more organizations to (try to) increase their social capital. Organizations enter into
various forms of inter-corporate relationships. We are moving from core
competence to core capabilities, increasing the tendency toward hybrid forms of
organization. Changing labor contracts are now built on the notion of
employability--organizations offer less permanent contracts.
In a competitive and fast-paced world, the differences between the winners and
the losers will more strongly be determined by the way actors make use of their
opportunity structures. Social capital represents the pipeline to those opportunities.
Social capital theory has gained prominence in a wide range of academic fields,
particularly in sociology, economics, political science, and managementlbusiness
science and is appearing with increasing frequency in intellectual magazines and
professional reports.26 Woolcock (1998: 184) even claims that social capital is
'arguably the most influential concept to emerge from economic sociology in the last
decade.' He also points out that 'in social capital, historians, political scientists,
anthropologists, economists, sociologists, and policy makers-and the various
CSC: The Structure of Advantage and Disadvantage - 13

camps within each field-may once again begin to find a common language within
which to engage one another in open, constructive debate, a language that
disciplinary provincialisms have largely suppressed over the last one-hundred-and-
fifty years' (Woolcock 1998: 188).
Notwithstanding the substantial insights in the literature into the advantages
and-in a much smaller part of the literature-the predicament of social
embeddedness, researchers still lack a coherent theory for explaining how and when
social structure transforms itself into corporate social capital or into corporate social
liability. We were both disappointed and inspired by this. Trying to develop a more
coherent and encompassing framework for studying the interplay between social
structure and social capital in the context of corporate actors, we aimed at extending
the insights from various fields into a systematic exploration of the concept. We
therefore invited prominent scholars to contribute to this volume and suggested they
remain within their own field of expertise in writing their chapters. As a
consequence, some of the chapters have a slightly different take on corporate social
capital than we do in this chapter. Still, we believe that the chapters provide a
coherent overview of the field . Some of the chapters are of a theoretical nature,
some of them methodological, many of them empirical.
The first chapters in the book present conceptual issues dealing with CSC
theory, models, and measurement. These chapters delineate the pressing challenges
in these three domains. In the second and third section, CSC is discussed starting
from social structure at the level of the individual. The effect of social structure is
discussed as it facilitates or inhibits players in pursuit of their professional careers-
moving ahead in a competitive environment. The third section explicitly highlights
two basic characteristics of CSC: contingency and fragility-not every social
network conveys social capital and social networks carrying social capital at one
point in time can create social liabilities at another. Social structure at the firm level
is considered in the next two sections---highlighting ways In which
interorganizational networks bring social capital to the firm.
We had the pleasure of reading multiple versions of the chapters, moving from
rough outlines to the texts you find in this book. If, after reading several chapters,
the reader has an (increased) appreciation for the theory of corporate social capital, it
has all been worth it. We hope you will have as much pleasure reading the book as
we had putting it together. Enjoy!

We thank Laura Rittmaster and Judith de Kleuver for their help in producing this book. The assistance
and gentle pressure provided by Julie Kaczynski and Elizabeth Murry of Kluwer Academic Publishers
were instrumental for the project as well. We thank the School of Management and Organization (Cluster
of Business Development) at the University of Groningen and the Davidson Faculty of Industrial
Engineering and Management at the Technion for their support and resources.

l. The first appearance of social capital in the scientific discourse (see soc-net discussion on the
genesis of social capital) was in Marshal (1890) and Hicks (I942)-1hey, however, used the term to refer
to different types of physical capital. Hannifin (1920) employed the term in the context of community
studies. Again in the context of community studies, social capital was used by Jacobs (1965) and Hannerz
(1969). Loury (1977, 1987) used the term in the field of child psychology. Bourdieu (1972, 1980)
developed the term in reference to cultural capital. The first researchers who used social capital explicitly
in the study of organizations were Flap and De Graaf (l986}-in their investigation of job mobility.
14 - Corporate Social Capital and Liability

Coleman (1988, 1990) was the most influential in developing a general wide-scale theory of social capital
as it is mostly used by scholars today. The work of Bur! (1992) was important in its wide visibility among
students of organizations as well as for his explicit emphasis on actors ('players') who are described as
competing in the market-place. Putnam (1993ab) has been influential in his application of social capital to
macro development policy issues, some of which are used by macro world bank policy makers.
2. Even though most of Coleman's work did not deal with business organizations, it provided a lot of
inspiration to scholars of the firm who used his theories in their study of organizations.
3. Third largest was Airbus with 124 customers, fourth largest Bombardier with 107.
4. Koelewijn (1997).
5. Coleman (1990).
6. Also see Gabbay and Sato (1996)
7. During the Second World War, German soldiers were entitled to two liters of beer a week. The
Germans were not able to supply their troops that occupied The Netherlands, so they relied on Heineken
Company, a Dutch brewery, to distribute beer to the German troops. However, in order for Heineken to
distribute the yellow nectar to the Germans, the German army had to inform Heineken about the location
and movements of its troops. Heineken then provided the Dutch resistance with this information. The
business relationship with Heineken thus helped the Germans quench their thirst for beer, but also
negatively affected the German ability to protect strategically sensitive information.
8. With the term 'corporation' we refer to both for-profit and not-for-profit organizations (see, for
instance, the chapters by Ferlie and Doreian). Also, we both consider organizations that deal with tangible
combodities, and those that provide services (see, for instance, the chapters by Iacobucci, Lazega,
Sandefur et al.).
9. This finding is reported by Brass (1985a).
10. Gabbay (1997) calls this 'negative social capital.'
II. Daft (1995).
12. Gabbay and Stein (1999), studying infrastructural network projects and their effect on the Middle
East, develop the notion of 'regional social capital' inherent in country-to-country networks.
13. Van Engelen et al. (1999). Also see Amason and Schweiger (1994) and Dess and Priem (1995).
14. For an overview, see Leenders (1995b, 1999).
15. See Davis and Newstrom (1985), Hyatt (1989), Simmons (1985).
16. Exemplary is the resistance the Dean of a management school experienced when a new curriculum
was introduced that was completely different from the old one. In their fear of having to give up some of
their hobbyhorses and relinquish control over the new curriculum, a number of the professors mobilized
the social ties they had within the school. They were not fighting the introduction of a new curriculum per
se; they were merely fighting having to give up credit points offered for their personal favorite topics. The
formal organization set up for the (substantive) development of the outlines of the new curriculum only
provided these professors with limited influence. Still, at a meeting with the entire academic staff present,
one of them got up and said to the Dean 'why do you even think you can make these decisions? Let the
decisions be made where the real power is: with us!' He was (largely) right: the informal network still
maintained a lot of the power that, formally, belonged to the dean. Although the dean officially had the
authority, he would not be able to pass anything the informal network of professors wished to block.
17. See, Mintz and Schwartz (1985), Stokman et al. (1985), Useem (1979), and Pennings (1980).
18. Nohria, (1992a: 11).
19. Gabbay and Leenders (1999).
20. In some industries, contracts are occasionally signed that prohibit the leaving employee from
working with former clients for a set number of years.
21. The discussion of the interplay between social structure at the firm level and social structure at the
individual level is beyond the scope of this chapter.
22. Salancik (1995: 348).
23. We use here the most well known typology in the study of organizations. Other typologies can also
be connected to corporate social capital. For instance, see Allison's (1971) typology of organizations as
'rational' and 'political' actors, as discussed in Pennings and Lee (this volume).
24. Also see Walsh et al. (1999) on el~onic networks and Gabbay and Stein (1999) on country-to-
country networks.
25. Modem technology and the increasing use of computer mediated communication systems extend the
intuitive understanding of corporations far beyond these limited descriptions (Walsh et al. 1999).
26. See, for instance, World Bank (1997).

• Conceptual Issues
theory, models, and measurement
Organizational Networks
and Corporate Social Capital

David Knoke

Corporate social capital is defined as processes of forming and mobilizing social
actors' network connections within and between organizations to gain access to
other actors' resources. Following a brief overview of basic network concepts and
principles, I discuss alternative theoretical explanations for the origins, spread,
transformation, and erosion of social capital. Two sections next investigate how
network dynamics have reshaped corporate practices and changed the employment
contract between workers and their firms . In conclusion, researchers should conduct
more empirical investigations and construct better theories about the mechanisms
through which social capital networks change the fates organizations and their

Michael Eisner, the imperious chairman of the Walt Disney Company, hired his
long-time friend Michael Ovitz to fill the media giant's presidency in August, 1995.
That position had lain vacant for months following Frank Wells' death in a
helicopter crash. Ovitz began his Hollywood career humbly, as a tour guide at
Universal Studios followed by a stint in the William Morris Agency's mailroom. It
soared after he co-founded the Creative Artists Agency in 1975, which soon grew
into the entertainment industry's premier talent agency, representing a thousand film
personalities including Tom Hanks, Barbra Streisand, and Tom Cruise. Ovitz
became the Hollywood's most-powerful and most-feared figure, personally
brokering such mega-deals as the Matsushita-MCA merger and SONY's acquisition
of Columbia Pictures. In 1995, he lured CBS news executive Harold Stringer to
18 - Corporate Social Capital and Liability

head Tele-TV, a new video-program joint venture of three regional telephone

companies. Tele-TV was relying on the super-agent's connections to procure
production deals with the Hollywood studios. But Ovitz's surprise defection to
Disney, which was allied with a rival group of telecommunication companies, left
CAA scrambling to hold onto its business and entertainment clients.
Industry observers assumed that Disney hired Ovitz to strengthen its ties to top
Hollywood talent, to manage its recent $19 billion takeover of Capital Cities!ABC,
and eventually to succeed Eisner as chair: 'It provides the company with Ovitz's
wide network of contacts and skills,' said one stock analyst. 'Frank Wells was always
a sounding board for Eisner. Ovitz will playa similar role' (Wells 1995). Yet after
just fourteen acrimonious months, unable to agree with his boss on how best to run
the company, Ovitz left the Magic Kingdom 'by mutual agreement.' Denying rumors
of a personal feud, Eisner said, We have been doing business together while being
friends for many years, and I know that both our professional and personal
relationships will continue' (Van de Mark 1996). To salve his wounds, Ovitz took a
reported $90 million severance package.
Although exceptionally dramatic in its particulars, the Disney episode
exemplifies the use of social capital to achieve individual and collective goals,
especially the importance of intra- and interorganizational connections for employee
careers and corporate performances. Social actors-persons, groups, organizations,
nations--continually interact in pursuit of their interests, and through those
processes they generate complex webs of social relations that advance or thwart
their goals. Three key issues, regarding the significance of social relations as social
capital, permeate this chapter: 1) How do actors in organizations create, reproduce,
and change their network ties to other participants? 2) How do actors' positions in
multiplex network structures facilitate or constrain their acquisition of social capital
and affect personal and organizational performances? And 3) how do these network-
formation and performance-outcome processes interact and mutually shape one
another over time? That is, selection and influence may operate reciprocally:
connections may originate because well-performing organizations prefer ties to
other outstanding actors, while superior performances may result from the social
capital those network ties generate. Answering those crucial questions draws
research attention to temporal sequences of network formation and consequences for
corporate social capital at both individual and organizational levels of analysis.
The next section defines social capital as the processes by which social actors
create and mobilize their network connections within and between organizations to
gain access to other social actors' resources. Then I briefly develop some basic
network concepts and principles. I next review competing accounts of the origins,
proliferation, modification, and destruction of social capital. The following two
sections investigate how network processes are reshaping corporate practices and
changing the employment contract. The conclusion appeals for more research and
theory construction about how social capital networks alter the fates of organizations
and their participants.
Organizational Networks and Corporate Social Capital- 19


Following suggestions by Bourdieu (1980, 1986) and Loury (1977), James S.
Coleman defined social capital as social-structural relationships that an individual
could mobilize in actions 'making possible the achievement of certain ends that
would not be attainable in its absence' (Coleman 1990: 302; see also Coleman 1986,
1988). In contrast to other forms of capital (physical, financial, human, and cultural
capital), social capital resides neither in a person's skills and knowledge nor in such
physical instruments as tools and machines. It is embedded in social connections and
is 'created when the relations among persons change in ways that facilitate action'
(Coleman 1990: 304). Indeed, for many purposes, a network tie is itself a form of
capital. People and organizations actively shape their social relationships to obtain
better opportunities and benefits. By forging large volumes of connections to
numerous, diverse, and well-endowed contacts, a social actor gains potential access
to the assets controlled by those contacts:
Social capital is at once the resources contacts hold and the structure of contacts in a
network. The first term describes whom you reach. The second term describes how you
reach. (Burt 1992: 12)
An actor's social capital encompass many types of relations, including trust and
confidence, obligations and cooperation, and information. These 'moral resources'
ultimately depend on shared social norms that sustain and strengthen the cooperative
bonds among actors in a social system. The dynamics of social capital may parallel
other forms of capital growth, with initial investments in riskier relations
subsequently paying higher returns to the investors. For example, rural communities
and small towns are typically pervaded by extensive webs of mutual social and
economic assistance among the residents. Over time, these neighborly exchanges
generate high reservoirs of trustworthiness that outstanding obligations to
reciprocate will likely be repaid in future times of need. Persons in such
interdependent social systems benefit from greater amounts of social capital 'credit'
on which to draw, as demonstrated by heroic yet futile sandbagging efforts to
prevent the Red River from inundating North Dakota's and Minnesota's riverine
towns during the spring of 1997. Similar endeavors occur in The Netherlands,
involving periodic evacuations of thousands of residents from lowland areas.
Several social network analysts propose that people mobilize their chains of
direct and indirect social connections, particularly their 'weak ties' to higher-status
persons in stratified social systems, to accomplish personal goals such as finding
jobs and achieving upward social mobility (Granovetter 1973; Lin, Ensel, and
Vaughn 1981; Boxman, De Graaf, and Flap 1991; Lin 1995a; Bian 1997; see the
Higgins and Nohria chapter in this volume). Job-seekers obtain timely information
about openings and necessary qualifications primarily through informal
communication channels rather than from formal employment institutions (Rees
1966). An applicant's chances to secure a job or a promotion may depend less on
human capital, in the form of school credentials and examination scores signaling
achieved skills, and more on obtaining favorable assessments from key advocates
and gatekeepers who can vouch for the applicant's reliability. For example,
20 - Corporate Social Capital and Liability



Figure 1. Schematic of a multi-level social capital network

university hiring and promotion procedures typically require letters of

recommendation from trusted assessors willing to testify about a candidate's
worthiness. Thus, people's chances for improved career outcomes are often
enhanced by their ability to tap into (,borrow') the scarce social resources held by
actors beyond their immediate social circles. Resources accessed through network
relations become an individual's social capital, regardless of whether those ties were
explicit investments made by the person in expectation of a future payoff, or were
overtly mobilized during instrumental actions (Lin 1995b). The wider the range and
the more diverse the contacts directly and indirectly available to a job-seeker, the
greater that person's chances for ultimate success. Thriving in a highly competitive
labor market favors candidates possessing heterogeneous networks over applicants
whose opportunities are restricted by their redundant, impacted networks (Burt
1992: 195; 1997).
The idea that social networks constitute an actor's social capital investments can
be readily extended to social groups and corporate entities. Just as an individual can
mobilize her personal contacts' social resources for purposive action, so can a
formal organization activate various resource networks to achieve its goals. Figure 1
illustrates a hypothetical social system involving complementary exchange networks
at macro- and micro-levels of analysis (for simplicity, I omit intermediate groups
such as teams and departments). The circles represent the formal boundaries of two
organizations: Bankright, a financial corporation, and Squarebilt, a construction
company. This example highlights two key employees: Jane, a Bankright
commercial loan officer, and Dick, Squarebilt's chief controller. Pairs of directed
lines (arrows) represent social exchange relations linking these workers and their
enterprises. Within each organization, the employees supply resources essential to
their firms' operations, specifically financial expertise and experience in loan-
making and acquisition. As returns on these human capital investments, their firms
reward Jane and Dick not only with salaries, fringe benefits, and job-training
opportunities, but also with increased social capital via their relationships with other
company employees, for example, authority, social status, and collegiality.
The two pairs of horizontal lines depict resource exchanges that span
organizational boundaries at both levels. As agents for their organizational
principals, Dick and Jane directly negotiate business loans, but apart from these
roles they may also maintain personal and professional ties, such as common school,
Organizational Networks and Corporate Social Capital - 21

church, and voluntary association affiliations. At the institutional level, the two
companies are linked primarily through their long-standing banker-client
relationship. Such financial transactions are often reinforced by solidary connections
between their chief executive officers, who belong to the same social clubs and
political parties, and whose families may even intermarry. These instrumental
networks payoff by turning one actor's social resources into the social capital
investments of other individuals and corporations. Squarebilt gains access to
Bankright's resources indirectly whenever Dick mobilizes his ties to Jane, who
possesses authority to make construction loans. Alternatively, were Jane to reject his
loan application, Dick could mobilize his connections to Squarebilt's CEO to tap
into the firm-level authority of Bankright's CEO and bring pressure to bear from
above on Jane to reverse her decision.
Actors often nurture and manipulate their social relations as deliberate strategies
for coping with uncertainties arising from dependence on external environments for
many critical resources needed to 'get the job done.' Explaining how people and
organizations actually behave requires considering factors beyond the purely arm's
length economic transactions occurring in spot-market exchanges between
anonymous buyers and sellers, where efficiency criteria are allegedly the paramount
determinants. Many economic relations are embedded within larger social, political,
and legal contexts (Granovetter 1985), which constrain participants' choices and
actions according to normative and political criteria transcending pure cost-benefit
calculations. These contexts can be conceptualized and empirically modeled as
multiple networks, connecting diverse social actors with varying interests and
resources, activated for individual and collective purposes. Hence, researchers
ignore network structures at the peril of providing incomplete insights into
organizational structures and processes.
Mixed competitive and cooperative modalities suffuse many types of network
interactions. At times actors mobilize their social capital to gain personal advantages
over their adversaries, while in other circumstances they jointly coordinate actions
for collective benefit. For example, corporate employees engage in self-serving
career strategies, seeking out mentors or networking with superiors to advance up
the promotion ladder. In contrast, new management practices and workplace designs
stress teamwork and collaborative responsibility for production, encouraging
workers to pool their skills and social capital to improve group performance.
Similarly, both modalities operate at the level of organizational strategy where plans
to achieve global corporate goals are implemented. Firms operating in the same
industry generally compete for customer loyalties and form exclusionary supplier
relations, yet they may also collaborate in strategic alliances and joint ventures with
expectations of mutual gains. An important task for network theory is to explain
under which conditions zero- and positive-sum interactions are more likely to occur.


This section briefly reviews some fundamental issues facing network researchers. I
explicate these concepts and principles in a nontechnical manner, which should
suffice for understanding the chapters in this volume. Readers seeking deeper
22 - Corporate Social Capital and Liability

knowledge should consult such didactic texts as Knoke and Kuklinski (1982) or
Wasserman and Faust (1994).

Identifying Actors and System Boundaries

Networks are social constructions arising from exchanges and joint activities among
participants in a social system, defined as a 'plurality of actors interacting on the
basis of a shared symbol system' (Parsons 1951 : 19). The actors in a network may be
designated at varying levels of analysis: individuals (children in a kindergarten
class); small groups (work teams on an automobile assembly line); formal
organizations (corporations in business association); coalitions (lobbying alliances);
even nations (members ofthe World Trade Organization).
Identifying a social system's boundaries, and hence its size, requires specifying
which potential members are relevant to the system's functioning. Investigators
using a nominalist strategy typically achieve conceptual closure by including all
actors possessing one or more key characteristics (Laumann, Marsden, and Prensky
1983). Nominal designations often restrict network membership to incumbents
occupying formal positions, for example, directors of Fortune 500 companies or
middle managers at Apple Computer. The alternative realist approach to boundary
specification assumes that system participants themselves can best identify who
belongs. Uncovering the network members' subjective meanings requires a
researcher to designate 'a substantively defined criterion of mutual relevance or
common orientation among a set of consequential actors' (Knoke and Laumann
1982: 256). Typically, potential network members carry out a reputational ranking
of other actors according to their importance to the system's performance. Actors
enjoying high reputations, indicating that their peers believe they must be taken into
account, are included but actors with low or no reputations are dropped because of
their marginal importance.

Total and Ego-Centric Networks

For a small social system, researchers may be able to obtain data on connections
among all system participants, comprising a total network. 'Small' in this context
means anywhere from a dozen to several hundred actors. To construct a total
network among G actors requires each actor to report on existence of ties to all other
system members, typically by checking a previously compiled name list (see
questionnaires for U.S. energy and health domains in Laumann and Knoke 1987:
401 -500).
Some social systems are either too large or too weakly connected to collect total
networks. In such instances, the only feasible alternative representative is to draw a
sample from the target population and to elicit the direct network ties of each
sampled actor (an ego-centric network). For example, a high-tech firm studied by
Burt (1992) employed 3303 managers who had few direct or indirect connections to
one another. He sampled 547 managers and ask them to describe their social and
work-related contacts. An ego-net procedure identifies a unique set of network
alters, typically through a two-step name-generator protocol. First, an informant ego
lists the most important people with whom specific types of interaction occurred.
Then, ego describes each alter's key attributes, such as age and gender, the nature
Organizational Networks and Corporate Social Capital - 23

relations with each alter (e.g., closeness and frequency of contacts), and perceived
direct ties among the alters. Importantly, because the alters are not subsequently
interviewed, ego's self-reported information remains unconfirmed. The ego-net
approach is the only plausible network methodology for general population surveys
(Marsden 1987) and large samples of diverse organizations (Kalleberg, Knoke, and
Marsden 1995).

Network Relational Contents

The next task is deciding what substantive relationship to measure. Network' s
relational contents refer to some relatively homogeneous tie among actors.
Relational contents seem to fall into two general categories: I) A transaction
involves exchanges where one actor yields rights of control over some physical
commodity or intangible value to another actor, possibly in expectation of eventual
reciprocation; 2) Joint actions require actors to co-participate in an event located in
specific time and space, without relinquishing control over resources. A minimal
joint action is mere co-presence, for example, software firms attending the annual
Las Vegas COMDEX trade fair. More intense joint activity involves coordinated
efforts to achieve common goals, for example, lobbying the Congress on legislative
proposals (Knoke et al. 1996).
Deciding which types of network content to operationalize should be guided by
a project's theoretical objectives: what substantive relations are useful for
understanding important actor and system behaviors? In practice, multiplex ties link
social actors into complex webs, meaning that no such creature as 'the network'
exists for any system. Typically, more than one relational content may be studied in
a given project.
The following typology roughly classifies relations by increasingly formality:

Minimal awareness, with actors reporting whether they 'know about' or 'have ever
heard of others in the system. Because recognition is typically unreciprocated,
mutuality in choices is rare, thus differentiating the 'stars' who enjoy high visibility
from their anonymous fans in sports, politics, and science networks. Among
organizations, corporate reputations for quality products and services similarly
differentiate the well-regarded from the invisible players (Fombrun 1996).
Common presence at the same events, or membership in the same collectivities,
disregarding direct interaction. Mass public assemblies, such as political rallies and
athletic contests, anchor one end of a continuum, while co-participation in restricted-
access enterprises, such as private schools and social clubs, implies the existence of
potentially cohesive social classes (Domhoff 1975).
Information Exchange
Routine and regular communication of data, whether about scientific-technical
matters (a supplier's current catalog of available products and prices) or socio-
political affairs (claimant organizations' positions on legislative bills). Such
information may be widely broadcast through press releases and email-server
24 - Corporate Social Capital and Liability

networks, or more narrowly targted on specific recipients, for example, a marketing

unit providing the sales department with consumer survey results.
Confidential information intended to help a recipient gain advantage over
competitors. An advisor transmits her superior knowledge and experience to an
advisee, sometimes with no expectations beyond gratitude and deference.
Organizational superiors mentor their favored underlings (Keele 1986; Noe 1988b),
expert consultants familiarize their organizational clients with R&D opportunities,
and venture capitalists scrutinize start-up investment prospects. (See the chapter by
Freeman in this volume.)
Confidence in the dependability of another actor's promises, reducing the chances of
opportunistic behavior by one's partners. Trust relations run a risk that such reliance
will prove ill-founded, but their efficiency in reducing transaction costs is an
important precursor for building long-term exchanges between people and
organizations. (See the chapter by Nooteboom in this volume.)
Expressions of sympathy, empathy, or commiseration in times of need, including
instrumental actions that demonstrate solidarity with an afflicted actor. At the
individual level, friendship and kinship bonds comprise the most obvious varieties
of support. Support relations among organizations involve public legitimation, for
example, testimonials regarding one company's willingness to rally behind another
enterprise confronting political or legal difficulties.
Financial Aid
Transfers of money, credit, or physical facilities, other than arm's-length market
purchases of goods or services. Some financial ties seem altruistic, as in
corporations' philanthropic donations to charities (Galaskiewicz 1985), while other
exchanges attach implicit quid pro quo strings, such as corporate contributions to
parties and candidates via political action committees (Mizruchi 1992). Other
financial exchanges entail explicit ownership and control connections, such as the
debt and equity transactions linking corporate and banking members of Japan's
famous keiretsu (Gerlach 1992a).
Legitimate power to expect that commands will be obeyed, backed up by sanctions
for failures to comply. Employment relations in corporate and public bureaucracies
typify intraorganizational authority networks, while a regulatory agency's power to
constrain an industry embodies interorganizational ties. The interlocking directorate,
connecting corporations and banks (Mintz and Schwartz 1985; Stokman, Ziegler,
and Scott 1985) and nonprofit organizations (Useem 1979), may reflect hybrid
authority patterns where agents of outside organizations help set policies binding on
participants inside another organization.
Collaborative arrangements involving two or more organizations that combine
resources to pursue common or complementary objectives, often involving
applications of uncertain technologies or entry into risky markets. These cooperative
strategies vary in the substantive content of interorganizational cooperation and
Organizational Networks and Corporate Social Capital - 25

types of governance structures designed to monitor and control the partners'

behaviors. (See the chapter by Stuart in this volume.)
The nine types of ties classified above are illustrative, but not exhaustive, of the
diverse relations that network analysts may find useful for their theoretical interests.

Magnitude of Ties
Network informants should also indicate the magnitude or value of their relations.
At a minimum, only a dichotomous coding-presence (1) or absence (O)-might be
recorded, for example, using name list checkoff. The unchecked alters yield
important data, since a network's structure depends as much on its gaps as on its
direct connections. More detailed magnitude codings assign scalar values reflecting
each tie's relative strength. For persons, tie strength usually refers to subjective
intensity of commitment, for example, a friendship study requesting egos to indicate
which people are their 'casual,' 'close,' or 'best friends' (Leenders 1996). For
organizations, relational magnitudes may involve objective data, for example, the
dollar amounts of loans from commercial banks to manufacturing firms, or the
numbers officers sitting on other companies' boards of directors.
Another important consideration is the time span observed: too short and
important but infrequent relations may be overlooked; too long and dormant ties
might mistakenly be treated as current. Unfortunately, the temporal dimension
hasn't been well-integrated into research procedures. Most network projects yield
static snapshots of a long-established network, without revealing their origins,
evolution, and ultimate fates. For example, we know little about whether informal
ties between employees of different companies subsequently generate organizational
alliances, or whether the opposite causal process occurs. Despite evident theoretical
payoffs from understanding network dynamics, data collection has not kept pace
with recent methods for investigating network changes over time (Wasserman and
Iacobucci 1988; Frank 1991; Zeggelink 1994; Snijders 1996; Leenders 1996).

Network Forms
Basic network forms describe the patterns connecting system actors regardless of
their specific relational contents. Figure 2 displays a hypothetical chooser-by-chosen
binary adjacency matrix and its associated graph. Think of the {ABCD} subset as a
production department located in one building, while the {WXYZ} subset is a
geographically dispersed salesforce. Actors A and W are these units' respective
heads. The relational content is regular communications about work. Each matrix
row represents a potential sender and each column a receiver of dichotomous social
ties. A 'I' entry indicates that the row actor communicated with the column actor,
while '0' means that no communication occurred. Graphs depict actors as labeled
points and their relations as arrows pointing from sending actors to receiving targets.
Because every communication tie is reciprocal, all ten arrows are doubled-headed.
The network volume is the total number of ties and its density is the proportion
of observed ties to the number of possible connections not counting self-ties (i - g
for a g-actor system). The example has a volume of 20 ties and a density of (20/56)
= .357. Actor connectedness counts the number of non-zero entries in a matrix row
26 - Corporate Social Capital and Liability



/' /'
c A ... • W I(

'" D

A 0
B 1


C 1 1 0 1 0 0 0 0
D 1 1 1 0 0 0 0 0
W 1 0 0 0 0 1 1 1
X 0 0 0 0 1 0 0 0
y 0 0 0 0 1 0 0 0
Z 0 0 0 0 1 0 0 0

Figure 2. Graph and matrix representations of a hypothetical eight-actor network

(out-degrees) and column (in-degrees). Thus, department heads A and W are the
best-connected actors, with out- and in-degrees = 4. Next come B, C, and D with
three ties each, trailed by the dispersed salespeople with just one connection to their
boss. Successively mUltiplying a matrix by itself reveals the minimal path length
required to connect pairs of actors. Visually, a path can be traced across directed
arrows between pairs, with the length being the number of steps needed to connect
that dyad. For Z to pass a message to C requires a path of length = 3: (ZW) + (WA)
+ (AC). The four production members are connected by one-step paths (direct ties),
but the salespeople require 2-paths to reach one another.
A and W enjoy unique and powerful roles in the system, an insight confirmed
by measures of actor centrality (see Freeman 1977, 1979). Basically, a central actor
participates in a large volume of social relations, with refined centrality concepts
differentiating among the type or 'quality' of connections. The simplest centrality
measures is an actor's in-degrees, measuring the sheer volume of ego-centric
contacts received from alters. Closeness centrality captures the extent to which ego
maintains connections to many alters who themselves have many non-overlapping
ties, thus enabling ego to reach many others by relatively short paths. A and W each
have the highest closeness scores (70), while B, C, and D enjoy somewhat higher
closeness (50 each) than the less-connected X, Y, and Z (43.8 each). Betweenness
centrality reflects an actor's ability to mediate many connections between
subgroups, thereby potentially leveraging greater impact on system activities.
Because the salespeople are less connected than are the production employees, W
has a higher betweenness score than does A (15 to 12), while the other actors'
betweenness centrality scores are O.
Organizational Networks and Corporate Social Capital - 27

The final network forms cluster actors into positions, which simplifies the
system's social role structure. Two basic approaches involve cohesion and
equivalence criteria. A clique is a network sub-set in which all dyads are maximally
connected (all reciprocal direct ties occur, yielding a density of 1.00). By this
rigorous definition, only the {ABCD} cluster comprises a genuine clique. Two
actors are structurally equivalent to the extent that they display identical or very
similar patterns of ties to all other alters, regardless of their ties to one another
(Sailer 1978). For example, firms in an industry that buy from the same sources and
sell to the same customers are fundamentally interchangeable competitors from the
market's perspective. The four equivalent blocks are {A}, {W}, {BCD}, and
{XYZ}. Automorphic equivalence identifies actors i and j are automorphically
equivalent if, after removing the 'names' of the actors from the nodes, nodes i and j
are impossible to distinguish. In the example, the sets of automorphic ally equivalent
actors are {AW}, {BCD}, and {XYZ}. In contrast to structural equivalence, which
puts A and W into separate positions because they supervise different individuals,
they are automorphically equivalent because they are connected to corresponding
others-their work-unit subordinates.

By the 1980s, converging environmental pressures began restructuring forever the
ways organizations would relate to their competitors, employees, customers, and the
larger society. While every analyst offers a favor~te list of key factors driving
organizational change, the six master trends cited by the Hay Group (Flannery,
Hofrichter and Platten 1996) seem particularly concise and comprehensive: rapidly
expanding technologies; growing global competition; increased demand for
individual and organizational competencies and capabilities; higher customer
expectations; ever-decreasing cycle times; and changing skilled personnel
requirements. A seventh trend, at least in the U.S., is increased investor pressure on
companies to improve their short- and long-term financial performances (Useem
1996). As corporations grew increasingly exposed to international competition, they
sought new ways to remain viable by slashing costs and prices, improving
production performance, and responding rapidly to technological innovations and
fickle consumer preferences. With consumer demand simultaneously globalized and
fragmented, niche markets for specialty goods and services supplanted cumbersome
mass-production systems run by 'Fordist' principles. Firms perceived performance
gains from unbundling their internal hierarchical structures and deinstitutionalizing
the product-unrelated conglomerate form (Davis, Diekmann, and Tinsley 1994).
These incessant pressures to achieve corporate flexibility and specialization drove
organizations to restructure their internal employment systems along much more
participatory lines. They also compelled companies to reach outside their traditional
boundaries to form long-term collaborative relationships enabling them to stay afloat
in an increasingly cutthroat world economy.
One consequence of the strategic search for new competitive advantages was the
proliferation of many new interorganizational forms . Figure 3 presents an alliance
typology, modified primarily after Yoshino and Rangan (1995 : 8). At one extreme
are pure market relations, whose transactions require no enduring collaboration by
28 - Corporate Social Capital and Liability

exchanging parties. At the other extreme are hierarchical arrangements in which one
firm assumes full authoritative control over the other, absorbing the participants into
a unitary enterprise. Between these extremes fall various 'hybrid' arrangements that
are neither clearly markets nor hierarchies but typically blend elements from both
types (Jensen and Meckling 1976; Williamson 1975; Powell 1987; Heydebrand
1989). An appropriate label might be the N-form or 'networked' organization, to
emphasize that relations are central to these mixed structures. For Yoshino and
Rangan (1995: 5), a strategic alliance's critical characteristics are partner firms that:
1) remain independent after the alliance is formed; 2) share benefits and managerial
control over the performance of assigned tasks; and 3) make continuing
contributions in one or more strategic areas, such as technology or products.
Based on this definition, they classified licensing and franchising as traditional
market contracts because one company grants another the right to use patented
technology or production processes in return for royalty payments. However, Figure
3 reassigns franchising under non-equity partnerships, since many distribution

Hierarchical relations Subsidiaries


Equity partnerships Minority equity investments

Equity swaps
Joint ventures
Business groups

Non-equity partnerships Franchising

Small firm networks
Joint R&D, production development, manufacturing,
marketing, distribution, service & other functions

Multi-participant alliances Data banks and information clearinghouses

Standards-setting consortia
Government-sponsored R&D consortia
Trade associations
Cooperatives for purchasing, marketing
Action sets for lobbying campaigns

Market relations Spot exchanges

Arm's length buy-sell
Short-term subcontracting
Figure 3. Types of interorganizational alliances
Modified after Yoshino and Rangan (1995)
Organizational Networks and Corporate Social Capital - 29

franchisers such as McDonald's Corp. exercise centralized coordination to safeguard

their corporate interests while leaving ownership and operations to the local
entrepreneur (Reve 1990: 148; also Osborn and Baughn 1990).
While some alliances require only a bilateral (dyadic) relations, other forms
involve mUltiple participants structured into complex 'alliance networks' (Gomes-
Casseres 1996: 52). Resource and authority commitments also vary considerably
among the hybrid types, particularly the extent of equity exchanges among partners.
Space limitations prevent detailed discussions of each interorganizational form, but
brief distinctions among the most important types may be helpful. An action set is a
short-lived organizational coalition whose members coordinate their efforts to
influence public policy decisions, for example, the passage of legislative acts
affecting the coalition members' collective interests (Knoke et al. 1996: 21). Cartels,
or pools, are unstable alliances formed to constrain competition by cooperatively
controlling production and/or prices in specific industries (Fligstein 1990: 39). The
'trusts' flourishing in late 19th-century America - in railways, heating oil, steel,
aluminum, sugar, salt-were outlawed but cartels periodically arise elsewhere, for
example, OPEC.
Small-firm networks (SFNs), are a late 20th century innovation involving large
numbers of very small firms (often with fewer than 10 employees) interacting on a
long-term basis, 'sharing information, equipment, personnel, and orders, even as
they compete with one another' (Perrow 1992: 455). They are complex clusters of
raw materials suppliers, producers, financial service, marketing, and distribution
firms. SFNs arise primarily in clothing, toys, publishing, motion pictures,
construction, light machinery, and electronics industries rather than in heavy
manufacturing or extractive industries. For example, Bennetton, an Italian apparel
producer, owns very few facilities but parcels out almost all production tasks to
hundreds of firms employing thousands of workers (Clegg 1990: 120-125; Kanter,
Stein, and Jick 1992: 228). The regional economies of Emilia Romagna in northern
Italy (Brusco 1982; Lazerson 1988) and Baden-Wiirttemberg in southwestern
Germany (Herrigel 1996) are the most famous SFN exemplars, as is perhaps Silicon
Valley in northern California (Saxenian 1994). SFNs should not be confused with
the more prevalent small supplier networks dominated by a large industrial firm
such occur in Germany's Ruhr Valley (e.g., Grabher 1993) and the American and
Japanese automobile industries (Womack, Jones and Roos 1990).
Four main types of equity partnerships occur. In a minority equity investment,
one firm buys an interest in another by direct investment, perhaps only a 3-5% stake,
'although the active involvement of the management of the partner-company is
retained and the assessment of expertise of the company can be made without
complete integration' (Hagedoorn 1993a: 132). Examples abound in high-tech fields
where large corporations use minority shareholding to access start-up firms'
technologies. An equity swap involves mutual direct investments of partners. A joint
venture occurs when 'two or more legally distinct firms (the parents) pool a portion
of their resources within a jointly owned legal organization' (Inkpen 1995: 1) that
serves a limited purpose for its parents. For example, the well-known NUMMI
automobile plant in Fremont, California, enabled General Motors to learn about
Japanese management techniques while Toyota gained a U.S. foothold (Adler 1993).
30 - Corporate Social Capital and Liability

Joint ventures may involve 50:50 ownership between two parents (Lewis 1990: 173-
192) or unequal equity shares among mUltiple partners.
Finally, a business group is a coherent collection of firms bound together at an
'intermediate' level between short-term strategic alliances and the unitary
corporation (Granovetter 1994: 454). East Asian partnerships among manufacturers,
suppliers, and financial institutions-such as the Japanese keiretsu (Lincoln, Gerlach
and Takahashi 1992; Gerlach 1992a) and Korean chaebol (Steers, Shin and Ungson
1989)-exemplify business groups spanning multiple fields that are integrated
through complex debt, interlocking directorates, and equity ownership patterns.
Although some observers argue that Chrysler Corp.'s comeback allegedly resulted
from an American weak-tie version of keiretsu, its cooperative relations with parts
suppliers involve neither the equity investments nor the management connections
that Toyota and Nissan have with their suppliers (Dyer 1996b). (See the chapter by
Pennings and Lee in this volume.)
The following subsections examine explanations of three aspects of
interorganizational alliances: why they occur, how they develop, and their
consequences. In seeking to understand network development over time, we should
ask whether any single theory can account for such diverse alliance phenomena or
whether distinct theories are required?

Theories of Alliance Formation

Two prominent theoretical explanations of why organizations engage in nonmarket
relations are transaction cost analysis and resource dependence theories, respectively
emphasizing economic and socio-political factors. Transaction costs determine
where to draw an economically efficient boundary between an organization
(hierarchy) and its environment (market), in other words, whether to make or buy a
particular function. Oliver Williamson (1975) argued that the most important factor
driving organizational efforts to economize is asset specificity, the extent to which
investments are specialized to particular recurrent transactions between buyers and
sellers. The greater the specificity, the more likely are the parties to 'make special
efforts to design exchanges with good continuity properties' (Williamson 1981:
555), thus effectively locking them both into prolonged bilateral exchanges. For
example, a corporation requiring only intermittent legal advice is likely to retain an
outside legal firm, while a company with persistent legal problems may create its
own in-house legal department. Thus, interorganizational ties arise from specialized
investments that would lose their value if transferred to another exchange partner.
Otherwise, market exchanges will be more cost-efficient. A second core assumption
of transaction cost analysis is that at least some actors are 'given to opportunism'
(Williamson 1981: 553), that is, dishonesty and dissembling about preferences and
information. Employing a different terminology, Williamson argues that
interorganizational relationships that convey social capital, are at risk of being
turned into social liability by opportunistic actions of one of the parties. The
necessity to monitor partners' performance and safeguard against duplicity can
increase interorganizational transactions costs. For example, Seagram's Universal
Studios successfully sued Viacom Inc., claiming the latter's launching of a
Organizational Networks and Corporate Social Capital - 31

competing cable TV network violated their agreement on joint operation of the USA
Network (Shapiro 1997).
Resource dependence explanations of alliance formation emphasize inherent
tensions between organizational resource procurement needs and the desire to
preserve freedom of strategic decision making. Intercorporate relations arise from
interdependencies and constraints among organizations: situations where one
organization controls the critical resources or capabilities-such as money,
information, production and distribution skills, access to foreign markets-needed
by another organization. Alliances tend to occur more often among interdependent
than between independent firms , that is, where complementarity rather than
similarity prevails. However, organizational efforts to manage problematic external
interdependencies 'are inevitably never completely successful and produce new
patterns of dependence and interdependence' (Pfeffer 1987: 27). Dependence
theorists argue that network ties arise from managers' efforts to control the most
troublesome environmental contingencies through complete or partial absorption
(e.g., mergers or joint ventures).
In their drive to acquire critical resources from network partners, organizations
risk losing control of their own destinies (social liability). Resource dependence
generates interorganizational power differentials that constrain firms' opportunities,
since organizations tend to comply with demands from the more powerful actors in
their environment. 'Organizations seek to form that type of interorganizational
exchange relationship which involves the least cost to the organization in loss of
autonomy and power' (Cook 1977: 74). Given an opportunity set of potential
alternative providers, a company will optimally choose a partner that can best satisfy
its resource needs while imposing minimal constraints on its discretionary actions.
For example, confronted with many suppliers capable of providing equivalent-
quality inputs, a large manufacturing firm is likely to purchase from the smallest
supplier, thereby gaining power to impose terms and conditions. Similarly, a small
supplier would prefer to spread its business across many customers, thereby
avoiding the loss-of-control stemming from dependence on a single partner.
Few analysts have explicitly tested hypotheses about alliance formation drawn
explicitly from the transaction cost or resource dependence perspectives. Pfeffer and
Nowak (1976) found that resource interdependencies (high exchange of sales and
purchases) among companies in technologically intensive industries significantly
increased joint venturing at the industry-level of analysis. Zaheer and Venkatraman
(1995) tested hypotheses about interorganizational strategies drawn from transaction
cost economics and social exchange perspectives, using data from a mail survey of
329 independent insurance agencies. Their two dependent variables were 'vertical
quasi-integration' (the percent of total premiums handled by an agency's 'focal
carrier,' the company with which an agency conducted most of its business) and
Joint action' (a multi-item scale measuring planning and forecasting activities with
the focal carrier). Although transaction-specific assets predicted quasi-integration,
neither uncertainty nor reciprocal investments were statistically significant. Instead,
quasi-integration and joint action were both positively related to mutual trust
between agency and carrier, a relationship opposite to the transaction cost
hypotheses but consistent with social exchange theory.
32 - Corporate Social Capital and Liability

Resource dependence principles seemed more helpful than transaction cost

concepts for understanding cooperative networks between new biotechnology firms
(NBFs) and established corporations in the 1980s (Barley, Freeman, and Hybels
1992; Kogut, Shan, and Walker 1992; Powell and Brantley 1992; see the chapters by
Smith-Doerr et al. and Stuart in this volume). Complementary resource needs drove
strategic alliances, primarily involving exchanges of financial support for technical
expertise. The small, innovative R&D laboratories typically lacked funds, public
legitimacy, and in-house capability to market their products and maneuver through
the regulatory maze. Hence they allied with diversified, resource-rich
pharmaceutical, chemical, and agricultural companies able to provide sustaining
resources. In turn, these established firms welcomed collaborative agreements as
means to acquire tacit knowledge and to learn new technological skills from their
NBF partners (social capital). As relationships accumulated and stabilized over time,
the network positions occupied by individual organizations constrained their access
to information regarding potential alliance partners (social liability). 'It is the
structure of the network, rather than attributes of the firm, that plays an increasingly
important role in the choice to cooperate' (Kogut, Shan, and Walker 1992: 364).
Two studies of changing networks patterns in other fields underscore the
importance of past ties on future actions. Leenders (1995b) reanalyzed dyadic data
from the social service networks of two Pennsylvania counties between 1988-90.
Informants named the organizations with whom their agencies maintained relations,
such as coordinating client treatments or sharing funds and personnel, including ties
mandated by the state. In both counties, estimated dyad-transition models revealed
that 'reciprocity both increases actors' inclination of creating and maintaining ties
and decreases the inclination of withdrawing ties' (Leenders 1995b: 193). Although
he did not use the term corporate social capital, the evolution of these
interorganizational networks clearly fits such an interpretation.
In a study of dyadic international corporate alliances, Gulati (1995a) found
evidence consistent with both resource dependence (which he called 'strategic inter-
dependence') and social structural explanations. Using a 1980-89 panel of 166
corporations operating in three worldwide sectors (U.S., Japanese, and European
new materials, industrial automation, and automotive products firms), he conducted
event-history analyses on a variety of dyadic alliances ranging from arms-length
licensing agreements to 'closely intertwined equity joint ventures' (1995a: 634).
Strategically interdependent firms (i.e., those companies operating in comple-
mentary market niches) formed alliances more often than did firms possessing
similar resources and capabilities. Previously allied firms were more likely to
engage in subsequent partnerships, suggesting that 'over time, each firm acquires
more information and builds greater confidence in the partnering firm' (l995a: 644).
Beyond a certain point, additional alliances reduced the likelihood of future ties,
perhaps prompted by fears of losing autonomy by becoming overly dependent on a
single partner. Indirect connections within the social network of prior alliances also
shaped the alliance formation process: previously unconnected firms were more
likely to ally if both were tied to a common third-party, but their chances of
partnering diminished with greater path distances. Gulati concluded that 'the social
network of indirect ties is an effective referral mechanism for bringing firms
Organizational Networks and Corporate Social Capital - 33

together and that dense co-location in an alliance network enhances mutual

confidence as firms become aware of the possible negative reputational
consequences of their own or others' opportunistic behavior' (1995a: 644). His
results reflected a logic of clique-like cohesion rather than status-competition among
structurally equivalent organizations.

Trust as Corporate Social Capital

The formation of successful strategic alliances between corporations hinges on
creating and sustaining relationships among the partners based on mutual trust. At
the individual level, we consider a person trustworthy if 'the probability that he will
perform an action that is beneficial or at least not detrimental to us is high enough
for us to consider engaging in some form of cooperation with him' (Gambetta 1988c:
217). At the interorganizational level, trust provides a foundation for one firm to
achieve some degree of social control over another's behavior under conditions of
high uncertainty. From a transaction cost perspective, the social capital of trust
expectations may provide an efficient mutual deterrent to both partners' temptation
to opportunism or malfeasance, thereby reducing alliance costs relative to more
formal control mechanisms such as written contracts (Gulati 1995a: 88-91). Hence,
interfirm trust relations fall conceptually somewhere between the polar logics of
hierarchical authority and market price relations (Bradach and Eccles 1989: 104;
Sako 1991).
The business-risk view of trust stresses confidence in the predictability of one's
expectations hedged by formal contractual means such as insurance (Luhmann
1979). Ring and Van de Ven (1994) emphasized an alternative psychological
conceptualization of trust as confidence in another's goodwill, of faith in the
partner's moral integrity. In their approach, trust constitutes a fundamental type of
organizational social capital, a strong-tie relationship between an ego firm and the
alters comprising its organizational field . Organization attributes and network
relations interact over time. As a company builds a reputation among its peers for
fair dealing and impeccable reliability in keeping its promises, that reputation itself
becomes a prized asset useful for sustaining its current alliances and forming future
ones. Reputed trustworthiness signals to potential partners that an organization is
unlikely to act opportunisticly because 'such behavior would destroy his or her
reputation, thus making the total outcome of the opportunistic behavior undesirable'
(Jarillo 1988: 37).
The social psychological explanation of trust is rooted in basic social exchange
principles, including conformity to such norms as reciprocity, commitment,
forbearance, cooperation, and obligations to repay debts (Stinchcombe 1986;
Bradach and Eccles 1989: 105; Lewis and Weigert 1985). Because typical interfirm
transactions are widely separated across time, trust reinforces these ties by invoking
such principles as that exchange values should balance over the long run, and that
each partners' payoffs should be roughly proportional to their contributions to any
joint enterprise. As trust relations became historically institutionalized in modem
industrial societies (Zucker 1986), initial arms-length market transactions grew
increasingly suffused with many normative connotations, generating and upholding
34 - Corporate Social Capital and Liability




Figure 4. Trust is an intervening factor in the alliance formation process

the moral communities within which trustworthiness conveyed great importance in

members' decisions whether to continue or break off relations.
What are the macro-level sources of trust among organization? Figure 4
proposes that interorganizational alliances emerge over time with trust occupying a
pivotal role between antecedent conditions and consequent alliance formations. Note
the feedback loop in which trust shapes the form of alliance, while events occurring
during the alliance may subsequently transform the interorganizational trust
relations, either reinforcing or weakening each partner's belief the other's
trustworthiness. Thus, trust and alliance relations mutually change one another as
interactions accumulate over time.
As suggested in Figure 4, communication networks structure an organization's
ability to screen and evaluate initial information about potential alliance partners.
These exchanges involve factual data about alters' interests and competencies, but
also provide indirect evidence about other organizations' trustworthiness via path
connections to knowledgeable peers in an organizational field. The more central an
organization's position within a field's communication network, the greater its
visibility and hence more informants are available to testify regarding its reliability
and integrity. Organizations located in peripheral positions have fewer opportunities
to become familiar with potential alliance partners and for their own trustworthiness
reputations to become vetted by the field.
A second set of antecedent factors fostering or thwarting trustworthiness are
macro-structural conditions. Imbalances in the resources controlled by each
organization (such as their financial size or market shares) may impede trust creation
because of unequal partners' inability to satisfy their reciprocity obligations. Pairs of
organizations that share similar or complementary characteristics are more likely to
develop strong trust relations. Tacit understandings and taken-for-granted
assumptions may be rudely violated when partners have little in common. For
example, many cross-border alliances, undertaken between foreign partners to gain
access to local markets, are fraught with pitfalls stemming from incompatible
national cultures (Lewis 1990: 253-278; Lorange and Roos 1992: 177-204; Bleeke
and Ernst 1993: 12-13; Gilroy 1993). Even domestic alliances can suffer from
clashing corporate cultures. A major instance was the office-network software
Organizational Networks and Corporate Social Capital - 35

producer Novell Inc.'s disastrous effort to integrate its subsidiary WordPerfect

Corp.'s 'close-knit and insular' staff with the parent organization's profit-driven style
(Clark 1996). After two years of plummeting market share and stock prices, Novell
sold WordPerfect to Corel Corp. at one-tenth its original $1.4 billion acquisition
The feedback loop between trust and alliance depicted in Figure 4 implies a
temporal dynamic to changing governance forms through accumulating
interorganizational experiences (Smith et al. 1995). Many alliances begin with
formal linkages that expose the partners only to small risks. Because the
organizations as yet have few bases for trusting one another, equity-based contracts
(,hostage-taking') predominate as legal protection against potential opportunism. But
after partners gain confidence in one another through repeated testing, then 'informal
psychological contracts increasingly compensate or substitute for formal contractual
safeguards as reliance on trust among parties increases over time' (Ring and Van de
Ven 1994: 105). This substitution process is succinctly summarized in Gulati's
(1995b) affirmative answer to the question 'does familiarity breed trust?' Because
strong-tie trust relations can counteract firms' fears of the partner's betrayal of
confidence, governing alliances through legal documents yields to relations
governed by interorganizational trust. Reduced transaction and monitoring costs
make informal social control the preferred cost-effective alternative to both market
pricing and hierarchical authority. Consistent with these expectations, Gulati's
(1995b) analysis of multi-sector alliances found strong evidence that formal equity-
sharing agreements decreased with the existence and frequency of prior ties to a
partner. Domestic alliances less often involved equity mechanisms than did
international alliances, supporting claims that trust relations are more difficult to
sustain cross-culturally.
Another crucial developmental issue concerns the relative agency of
organizations versus individuals qua persona in creating interorganizational trust. In
general terms, network analysis needs to resolve its quandaries about the role of
human agency in social action, 'the capacity of socially embedded actors to
appropriate, reproduce, and, potentially, to innovate upon receive cultural categories
and conditions of action in accordance with their personal and collective ideals,
interests, and commitments' (Emirbayer and Goodwin 1994: 1442). Applied to the
present context, a central question is whether trust relations operate at the
organizational level, or whether trust encapsulates purely interpersonal phenomena?
As noted above, some theorists emphasize that trust originates in the social
psychology of interpersonal interactions, and thus often evokes strong emotional
overtones of sharing and caring for the welfare of one's partner (McAllister 1995).
As the employees who occupy key boundary-spanning roles try to cope with their
organizations' environmental uncertainties, they socially construct strong bonds of
mutual confidence and trust with their counterparts in other organizations that may
affect interorganizational behavior. For example, a study of company decisions to
switch auditing firms found that the individual attachments of such boundary-
spanners as the company's chief executive, financial, and accounting officers
attenuated the pressures arising from changing resource needs (Seabright et al.
1992). If only people can manifest beliefs and emotional attachments, then trust may
36 - Corporate Social Capital and Liability

reside wholly within the individual fiduciaries who establish and nurture trust
relations on behalf of the organizations they represent. The potential for
intermingling the reputational social capital of people and organizations spawns
some knotty dilemmas for intraorganizational control: exactly who legally and
morally owns the trust relations in which both employers and employees have
invested? This question is not a trivial concern for firms, as reflected by such
practices as 'noncompete' clauses restricting local television news personalities from
working for rival stations after severing their employment ties, and in law-suits
against lawyers and talent agents who defect to rival firms, taking along their client
lists (Tevlin 1997). In the most extreme instances of trust violation, agents may
pilfer major corporate secrets for their new employers, as in Jose Ignacio Lopez's
alleged transfer of General Motors procurement data to Volkswagen.

Alliances Outcomes
The belief that interorganizational networks offer corporate social capital in the form
of performance benefits superior to both markets and hierarchies is widespread
among social scientists and corporate managers. Networks are allegedly 'lighter on
their feet' than hierarchies (Powell 1990: 303). They enable organizations and their
agents to respond rapidly to emerging contingencies, particularly gaining timely
access to swiftly changing technological knowledge and data essential for survival
and prosperity. Yet the evidentiary basis for such claims remains remarkably slim.
Researchers have proposed numerous criteria for judging alliance 'success,' ranging
from mere organizational survival to economic performance levels above industry
norms. One difficulty in assessing performance outcomes is that most alliances
explicitly seek only limited purposes and are intentionally short lived, so duration
alone may be an inappropriate yardstick. When an alliance terminates in one
partner's acquisition ofthe other, as in the majority of cases (Bleeke and Ernst 1993:
18), does that outcome constitute a failure of the alliance? A success for one
organization but a failure for the other?
Embeddedness in interorganizational alliances seems to contribute to
participants' survival chances compared to organizations engaging only in arm's-
length market transactions. Uzzi (1996a) used both ethnographic and quantitative
methods to study the impact on firm failure of the mUltiple network ties among 23
New York better dress apparel ftrms. 'Social capital embedded ness' indicated
whether a contractor had a network tie to a business group, typically formed around
CEOs who were kin or colleagues from previous jobs. Other measures involved the
proportion of work exchanged between organizations and the degree to which the
ego firm maintained arm's-length or embedded ties with partners. Uzzi's logit
analyses showed that 'firms that connect to their networks have greater chances of
survival than do firms that connect to their networks via arm's-length ties' (1996a:
694). But optimal networks were a mix of both types of relations:
A crucial implication is that embedded networks offer a competitive form of organizing
but possess their own pitfalls because an actor's adaptive capacity is determined by a
web ofties, some of which lie beyond his or her direct influence. Thus a firm's structural
location, although not fully constraining, can significantly blind it to the important effects
of the larger network structure, namely its contacts' contacts. (Vzzi 1996a: 694)
Organizational Networks and Corporate Social Capital - 37

Organizations enter alliances with many motives and strategic objectives, including:
speed of entry into new product or geographic markets; faster cycle times in
developing or commercializing new products; improved product or service quality;
gaining technical skills, tacit knowledge and competencies; sharing costs; spreading
risks and uncertainties; monitoring environmental changes. Bleeke and Ernst (1993)
relied on unpublished reports and interviews with insiders of 150 top companies in
the U.S., Europe and Japan to determine that, in 49 cross-border alliances, 51 % were
successful for both partners while 33% were mutual failures. Alliances were 'more
effective for edging into related business or new geographic markets' (1993: 18)
while acquisitions worked better for core businesses and existing areas. Other
conditions leading to success included alliances between equally strong partners,
evenly split financial ownership of the joint venture, and autonomy and flexibility
for the joint venture to grow beyond the parents' initial expectations and objectives.
Empirical evidence regarding the financial outcomes of strategic alliances is
scarce, with network studies of investment banking and the stock exchange a notable
exception (Eccles and Crane 1988; Baker 1990; Podolny 1993). For example Chung
(1996), analyzing cooperative exchanges among 98 top investment banks involved
in new stock issues in the 1980s, found that the best long-term performers (measured
by amounts underwritten) were involved in a strategy of exchange initiation, which
also led to subsequently higher popularity and expanded participation in stock deals.
However, few researchers have studied whether joint venture partners recover their
capital investments, or whether such collaborations yield a higher returns than
available from alternative resource allocations. Theorists tend to emphasize only the
social capital emerging from networks, while ignoring potential social liability
inherent in interorganizational relations, specifically that social embeddedness may
exert a drag on market efficiency. For example, Sako (1991: 239) speculated that a
major disadvantage of obligational contractual relations is '[r]igidity in changing
order levels and trading partners [and] potential lack of market stimulus.' Similarly,
the impact of trust on alliance success remains uninvestigated. Trust presumably
fosters goal attainment by facilitating the favorable resolution of conflicts inevitably
cropping up during joint operations. Given its subjective basis, high mutual trust is
likely to correlate with feelings of satisfaction about the partner's performance and
contributions. Researchers might inquire whether collaborators feel their venture is
worthwhile and whether they would repeat the alliance for other purposes or to
recommend their partner to other firms seeking to form strategic ventures. On the
negative side, trust and other obligational norms may attach organizations too
strongly to their partners, carrying relations beyond rationally efficient limits by
resisting swift dissolution of inefficient or inequitable situations. Clearly many
opportunities await for imaginative research on the outcomes of interorganizational
alliances. .

The macro-change forces noted above that reshaped interorganizational relations
also wreaked enormous transformations inside factories, offices, and clinics. During
a prolonged and painful decade of downsizing, reengineering, and restructuring
exertions, more daring or desperate corporations implemented flexible new designs.
38 - Corporate Social Capital and Liability

Rigid bureaucratic hierarchies yielded to experiments in cross-functional teams that

devolved increasing volumes of information, technical skills, and managerial
responsibilities down to the front-line worker level (Katzenbach and Smith 1993).
Employees were prodded to contribute to restructuring decisions by such schemes as
job enrichment, quality circles, job rotation, gain-sharing, and stock ownership
plans. The social control of organizational performance became increasingly
internalized through corporate cultures based on Deming and Juran's total quality
management principles, 'a set of powerful interventions wrapped in a highly
attractive package' (Hackman and Wageman 1995: 339). TQM emphasized the
never-ending collaboration between management and workers for continuous
learning and quality improvements, assessment of customer requirements, scientific
monitoring of task performance, and process-management to enhance team
effectiveness. These high-performance innovations were all intended to lower
supervisory costs and increase employees' work-life morale, thereby raising
corporate productivity, quality, and profitability (Levine 1990; Lawler 1992).
During the last half century, the implicit employment contract binding workers
and firms changed from a virtual guarantee of long-term job security to one
emphasizing employability (Cappelli et al. 1997). In the insightful words of Intel
Corp.'s vice president for human relations, 'You own your own employability. You
are responsible' (O'Reilly 1994: 47). By flattening managerial hierarchies and out-
sourcing formerly internalized staff functions, firms shortened or eliminated many
traditional internal labor markets that had provided career ladders for regular
promotions to ever-higher levels of responsibility, prestige, and pay. Instead, jobs
evolved from fixed positions into flexible bundles of tasks that were subjected to
periodic restructuring to grapple with organizational contingencies in tumultuous
world economic markets. Jobs metamorphosed into project-based appointments
through which multiply-skilled employees rotated in short-term assignments on their
way to newer projects inside the firm or with other employers. Temporary and
subcontracted workers became the fastest growing segments of the U.S. labor force
by the 1990s (Belous 1989; Parker 1994). The proliferation of computerized
communication (Internet and intranets) and production-control systems (CADI
CAM), coupled with escalating customer demands for made-to-order goods and
services, drove the relentless quest for continual upgrading of employees' technical
and interpersonal skills. Firms deployed a multi-track approach, searching for new
workers with requisite competencies, training current employees in-house, and
forging ties to external vendors of job-training services such as junior colleges
(Kalleberg, Knoke, and Marsden 1995).
From these gales of creative destruction emerged a new corporate form-the
network organization, whose external alliances were discussed in the preceding
section (Miles and Snow 1995). Its distinguishing internal features are multiplex
exchange ties among the firm's loosely coupled divisions, departments, work
groups, and the individual managers and employees. It breaks down hierarchical and
functional barriers, replacing them with task-specific units connected through
communication, advice, and trust networks (Krackhardt and Hanson 1993). The
Organizational Networks and Corporate Social Capital - 39

network organization
creates autonomous units, but it increases the volume, speed, and frequency of both
vertical and horizontal communication within the organization to promote collaboration .
... The result is an organization with superior performance characteristics for the 1990s.
Network management is, in the end, management by empowerment. (Limerick and
Cunnington 1993: 61)
Intraorganizational networks operate according to a logic of economically efficient
asset allocation. Rather than transferring goods and services by centrally
administered prices, the quasi-autonomous units are subjected to internal market
discipline when buying and selling resources, thus assuring they will continually
seek to improve their performance (Snow, Miles and Coleman 1992: 11). But,
corporate networks also function politically and socially in ways that defy strict
economic utility maximization principles. In particular, network relations offer
employees a prime source of social capital for developing rewarding careers under
the new employment contract terms which stipulate greater personal responsibility.

Networks and Career Capital

Employees have always used networking activity as important strategies for getting
ahead in their companies. A worker's personal networks comprise crucial social
capital investments that are as essential for career development as her or his human
capital assets of knowledge, skills, and experience. Employees survive and thrive by
learning how to construct and manipulate ego-centric networks that provide
advantages in the competitive scramble for jobs, project assignments, promotions,
and rewards. Networking abilities assume an even greater significance for
employees of the new forms of network organizations, where formal positions are
ill-defined and perpetually changing.
In a study of senior managers of a computer firm, Burt (1997) examined how
social capital affected rapid promotion. He measured social capital as constraints on
personal networks, that is, concentrated on fewer contacts. Persons whose networks
span more 'structural holes' are well-positioned to broker the flow of information
and to coordinate and control interactions between unconnected people on opposite
sides of the hole. Not only were managers with less-constrained networks promoted
relatively early, but the effect varied with the number of competitiors. The
correlation between network social capital and promotion was stronger for those
managers with few peers compared to those in positions where many people did the
same work. The social capital payoff was higher for people in unique corporate roles
'because such managers do not have the guiding frame of reference provided by
numerous competitors, nor the legitimacy provided by numerous people doing the
same kind of work' (Burt 1997: 356). Thus, their entrepreneurial networks offered
access to more rewarding opportunities.
Research on gender differences in network dynamics sought to explain how
personal networks are converted into corporate advantages. In a New England
advertising and public relations firm, Herminia Ibarra (1992) found differential
patterns of homophily (tendency to form same-sex ties) among the 80 male and
female employees. Men tended to concentrated their ties across multiple networks
(communication, advice-seeking, support, friendship, and influence) primarily on
40 - Corporate Social Capital and Liability

other men. Women employees differentiated according to network contents,

obtaining social support and friendship from their female co-workers and
instrumental access through ties to higher-status men. That is, expressive and
instrumental ties coincided for men, but were inversely correlated for women.
Consequently, men seemed to receive higher returns than women on their social
capital investments, in the form of greater network centrality.
Similar gender-differentiated network propensities occurred among 63 managers
of four large corporations (Ibarra 1993a), with men relying more on weak-tie
homophilous networks and women forging more strong expressive ties to other
women. The relationship between managers' ego-net strategies and their potential
for promotion, as judged by supervisors and human resources staff, were also
conditional by sex. High-potential women and low-potential men placed greater
relevance on expressive networks, such as trust and reciprocity, while high-potential
men and low-potential women stressed instrumental ties. Ibarra concluded that
women's preferred network strategies placed them at a disadvantage relative to their
male peers: The 'entrepreneurial' network pattern characteristic of successful male
managers is less effective for females who require stronger network ties to achieve
the same level of legitimacy and access to resources' (Ibarra 1993a: 27).

Networks as Power Resources

Social power is a structural property of the relationships among actors in a social
system, rather than inherent in individuals' formal roles or personalities (Pfeffer
1981: 3; Knoke 1990: 1). Hence, power and political action in organizations are
rooted in the multiple intraorganizational networks connecting the participants. Even
formal authority that assigns legitimate rights to control corporate human resources,
as exhibited in an organization chart displaying supervisor-subordinate positions,
should be viewed as just one of several networks conveying political implications,
including such informal ties as communication (Pfeffer 1992: 111-125), advice,
support, trust, friendship, and horizontal workflow (Brass 1984). Knoke (1990)
argued that the primary analytic power relations of every social system are reducible
to two basic exchange networks that follow differing logics under which actors
affect one another's behavior. Influence occurs when 'one actor intentionally
transmits information to another that alters the latter's action' (1990: 3), while
domination involves controlling another actor's behavior 'by offering or withholding
some benefit or harm' (1990: 4). For example, an employee may induce a co-
worker's collaboration by persuading the co-worker that cooperation is in the mutual
interests of both firm and employee (influence), or by offering resources essential
for the co-worker's project (domination). Over time, these informal political
relations become institutionalized as the company's intraorganizational power
structures, with the employees occupying the dominant and influential positions
affecting their less-powerful colleagues' perceptions, cognitions, beliefs, and
behaviors. Intra-organizational power structures are highly stable and resistant to
change as the persons in power seek to perpetuate their advantages. Structural
change occurs mainly as the result of major external shocks, for example, corporate
takeovers or technological innovations that drastically rearrange existing political
relationships (Burkhardt and Brass 1990: 105).
Organizational Networks and Corporate Social Capital - 41

A core theoretical proposition, derived from resource dependence principles, is

that actors who occupy the more central positions in intraorganizational networks
can exercise greater political power. While centrality may generate power for an
employee, the reverse causal process may also operate over time: people seek to
establish connections to the most powerful organizational players, in expectation of
enhancing their own power through these contacts. Incumbents in central locations
enjoy a variety of advantages over peripheral positions: through their proximity to
others in communication exchanges they can acquire more timely and useful
information; can better control the flow of resource exchanges and mobilize support
for initiatives; can mediate and broker deals between interested but unconnected
parties; and, through boundary-spanning ties to external organizational actors, they
can direct the organization's strategic objectives (e.g., Kanter and Myers 1991). In
short, 'network centrality increases an actor's knowledge of a system's power
distribution, or the accuracy of his or her assessment of the political landscape . ...
Those who understand how a system really works can get things done or exercise
power within that system' (Ibarra 1993b: 494).
As noted above, network methodologists developed alternative measures of
network centrality, including in-degree, closeness, and betweenness scores (Brass
1992). Rather than treating all relations as making equal contributions to each
person's centrality, a prominence index assigns higher centrality scores to
employees who are also highly central within the organization (Knoke and Burt
1983; Bonacich 1987). Thus, network centrality as prominence extends the principle
that 'it's not what you know, but whom you know,' to emphasize that your own
power depends importantly on 'whether the whom that you know has power.'
A few empirical investigations have uncovered compelling evidence that the
attribution of power covaries positively with employee centralities in
intraorganizational networks (see overviews by Krackhardt and Brass 1994; Brass
1995b). In a study of communication, friendship, and workflow networks among
140 nonsupervisory employees of a newspaper, Brass and Burkhardt (1992) reported
numerous statistically significant correlations between three types of centrality
scores and reputations for power (as attributed by supervisors and by peers). The in-
degree measure proved to be a stronger predictor than closeness or betweenness,
suggesting that a large volume of direct contacts may be necessary for coalition
formation and also provide the best mechanism for 'learning the network' (1992:
211). Krackhardt (1990) investigated the effects of betweenness centrality and
cognitive perceptions of both friendship (trust) and advice-giving networks on the
power reputations of 36 employees of a small entrepenurial firm. Controlling for
formal position in the company, persons who were more central in the friendship
network and who had more accurate cognitions of the advice network were rated as
more powerful by others. Neither advice centrality nor friendship accuracy had
statistically significant bearings on reputed power.
Finally, network centrality appears to affect some work-related perceptions and
activities. Ibarra's (1993b) analysis of which employees adopted problem-solving
innovations a New England advertising firm showed that centrality (prominence in
five types of network relations--communication, advice, support, influence and
friendship) 'was the most significant predictor of administrative innovation roles ...
42 - Corporate Social Capital and Liability

and mediated the effects of various of various non network variables on innovation
involvement' (1993b: 492). But centrality was not a statistically significant factor in
the adoption of technical innovations. Burkhardt and Brass's (1990) longitudinal
analysis of computer adoptions in a federal agency also found similar patterns, with
early adopters' power and centrality increasing more than later adopters. In further
analyses of the advertising firm data, Ibarra and Andrews (1993) showed that advice
network centrality and friendship network proximity to varying degrees each
affected perceptions of such organizational conditions as risk-taking, acceptance,
information access, interdepartmental conflict and autonomy.

The preceding review of research and theory construction about organizational
networks and corporate social capital suggests that we are collectively investigating
several critical issues. Researchers are probing the social forces that lead to the
formation of intra- and interorganizational ties, their persistence, and their
severance. We have fragmentary understanding of how global network structures
simultaneously facilitate ('social capital') and constrain ('social liability') the
opportunities available to people and organizations in pursuit of their interests. And
we now better appreciate corporate social capital as both a generator and an outcome
of strategic actions embedded in complex social structures. Still missing is a
comprehensive framework to coordinate and accelerate the efforts of numerous
scholars toward a more coherent and cumulative research program that could
integrate the diverse facets of these elusive phenomena. After decades of network
analysis developments, we have abundant conceptual and methodological tools with
which to forge such a synthesis.
Two generic tasks should be intensified in tandem. First, researchers should
track social capital across multiple levels within and between organizations. At the
intraorganizational level of analysis, research designs could examine the
concatenation of multiplex relations among employees, work groups, departments,
and divisions into complex yet reproducible assemblages that maintain the identity
and integrity of the corporation as a bounded social actor. At the interorganizational
level, investigators must examine the detailed mechanisms through which social
network investments yield individual and collective benefits to alliance members.
Until we gain a clearer picture of how relations between firms shape economic and
political outcomes, our perceptions of the emergent N-form organizations will
remain fuzzy. The second major task for corporate social capital researchers should
be to collect and analyze longitudinal data about changing network structures and
processes. Current knowledge is cramped by the cross-sectional nature of most
research designs. Many methodological advances promise boundless opportunities
to expand the temporal dimension of social capital dynamics. We need to learn how
seemingly minor changes in specific connections, involving a handful of critical ties,
can cascade rapidly through a network, radically transforming its shape and
functions. And we need to integrate unique events into the actor-relation dualism,
thereby increasing our capacity to capture the historical forces changing corporate
social capital.
Social Capital of Organization:
Conceptualization, Level of Analysis,

and Performance Implications

Johannes M. Pennings
Kyungmook Lee

In this chapter we explore the benefits of social capital and the harmful effects of
social liabilities. Following Allison (1971), two models of the organizations are
juxtaposed: those of the Rational and Political Actors. The issues of social capital
require different perspectives when its implications for performance are addressed.
The mediation through individuals takes a prominent place in the Political Actor,
and moves to the background in the Rational Actor. The issue of aggregation from
the member to the organization is primarily an issue when we view the organization
as a Political Actor in which the members' social capital aggregates to that of their
organizations. Two illustrative cases that fit the two models are then presented, the
industrial business groups in Japan and Korea on the one hand, and the popUlation of
professional services firms in the Netherlands on the other. In the case of business
groups we point to both the benefits of social capital and the drawbacks of social
liability. When we shift to the study of professional services firms, we demonstrate
that social capital as a distinct organizational resource diminishes the likelihood of
dissolution. The implications for social capital and social liability are exposed and

Organizations are presumed to have boundaries. They are endowed with various
kinds of assets on which they make ownership claims, and which are protected with
isolating mechanisms such as patents and contracts. They are liable for their
products and services. Also, they have members whose inclusion in the organization
is usually beyond dispute. In fact, the firm as a collection of individuals is often
44 - Corporate Social Capital and Liability

bracketed when considering the competitive game it is playing with other firms. Yet,
organizational boundaries are precarious and permeable. Organizations have
exchange relationships with suppliers and clients, collude with competitors, and
forge all kinds of alliances because they cover only part of the value added in their
value chain. In their positioning across the chain they face such decisions to 'make
or buy' components and supplies, whether to share or even outsource R&D efforts,
or to operate on a stand-alone basis. Their coherence and integrity might decline and
bundles of resources often unravel into discrete parts, but these resources might also
become combined-for example in divestments and acquisitions, respectively.
Organizations are embedded in a web of relational ties. In the present chapter,
the term social capital captures important aspects of this relational web. Social
capital of organizations constitutes a distinctly collective property that might be
mediated by individuals, yet is uniquely organizational. Social capital complements
financial and human capital as assets that are more or less valuable, scarce and
imperfectly tradable (Barney 1991). Social capital is even more unique and difficult
to appropriate than these other types of assets as it hinges on the continued
involvement of two or more parties. Firms, as repositories of unique resources
require complementary assets in order to compete successfully. Social capital is
crucial in bundling intangible assets and provides the absorptive capacity to merge
proprietary knowledge with that of others. Organizations need to coordinate their
interdependencies in the value chain and negotiate a position in their industry. By
forging external networks, the organization maintains optimal boundary conditions
and remains in tune with external trends and events. At the same time, its boundary
structures preserve an organizational modicum of identity and protection against
erosion of its assets.
The social capital benefits seem beyond doubt; less intuitive might be the cost of
social liability. Social embedded ness endangers a firm's appropriability regime, and
might also envelop the firm too tightly into a web of ties that stifles its ability to
change or impedes its innovative capability. While network relationship is often
viewed as conferring various benefits, we should therefore also examine its
undesirable consequences.
As numerous chapters in this book indicate, social capital refers to resources
inherent in sustained long term relationships and associations. The concept
originates in sociology, with two writers standing out: Bourdieu (1980, 1994) and
Coleman (1990). In this chapter we extend their representation of social capital by
treating it as a unique organizational resource. We will further reflect on the nature
of organizations, and ask how such human aggregates or their social organization
are capable of possessing social capital. As with human capital, we need to dwell on
the tension between individual and organizational levels of analysis. While it is
tempting to 'anthropomorphosize' the firm as a human aggregate and impute an
ability to mold its surrounding network, we need to ask how such semblance comes
about, who the agent is, and what collective motives are operating. After having
dwelled on these issues, we explore the implications for organizations of having
accumulated social capital. We do so by contrasting two contrasting settings, i.e.,
business groups and professional service firms, as these stylized forms might
respectively illustrate the firm as rational and political actor, and by implication, the
Social Capital of Organization - 45

sort of aggregation issues that color the reason we depict their social capital. Below,
we belabor these two metaphors to highlight aspects of corporate social capital. We
conclude by spelling out implications and future research opportunities.


It has not been customary to view organizations as embedded in a network of
relationships, although person based networks have been used to describe a firm' s
external linkages (e.g., Levine 1972). Much of the pertinent literature has focused on
individuals (e.g., Burt 1997; Coleman 1988; Granovetter 1985; Uzzi 1997a), their
place in some larger network, and the impact it has on their behavior and attitudes.
Many views stand in sharp contrast with an 'over-socialized' view of man.
Economists tend to couch transactions in personal, self-interest seeking terms. As
parties in a market, people engage in 'arm's-length' relationships and their
interaction is solely conditioned by the need for exchange. Contrary to a utilitarian
tradition, norm theory in modern sociology assumes that people are overwhelmingly
sensitive to the expectations of others (Wrong 1961). Sociologists often stress the
structural context within which parties meet, and such a context might give rise to a
small number of conditions in which actors develop personal bonds, based on trust
and mutuality. Uzzi (1997a, this volume) calls such links 'embedded ties.') Within
such bounds, utility maximization is often suspended for the sake of preserving
reciprocal, even altruistic relationships. The next issue involves the extension from
the individual as a party onto himself versus the individual as an 'office holder.' Size
also matters; for example a market with single proprietorships entails rather different
inter-firm networking than the US banking world in which firms are tied together,
for example, through interlocking directorates.
Entrepreneurs, new ventures, and small firms differ markedly from large
corporations in terms of the links they maintain. The links that bind them might vary
from those that are heavily endowed with trust to those that fit the arm's length
relationships. The large corporation is prone to have arm's length relationships with
external actors, but as we will see, they often invest in boundary spanning systems in
which personally mediated links are discernible. Small firms are more likely to
develop bonds of trust and mutual adjustment with external actors such as suppliers
and clients, although some conditions give rise to arm's length relationships.
We need to position these distinctions against the 'model' of the firm, which is
often implicit (Allison 1971; Simon 1957; Thompson 1982). Organizations have
often been viewed as 'rational actors' (Allison 1971) or have otherwise been treated
as unitary economic agents. As a singUlar agent, the firm might be embedded in a
multiplex web of inter-firm relationships as manifest in contracts, joint ventures,
stock cross-holdings, etc. As units with clear legal boundaries and other 'isolating
mechanisms' firms complement each other in the value chain. The ties that bind
them can be viewed as social capital for coordinating inter-firm activities. If we,
however, view organizations as human aggregates, as Allison (1971), for example,
stipulates in his organization as 'political actor,' we might attribute to that
organizational social capital by virtue of the aggregate social capital of its members.
The presumption of firms endowed with social capital appears non-problematic but
46 - Corporate Social Capital and Liability

the implications are rather different in the two scenarios thus depicted. In this
chapter we visit the issue of firm as rational versus political actor in greater detail.
In this chapter, for the sake of the argument, we juxtapose the rational actor
caricature with its political actor counterpart and examine the social capital as an
integral part of these models. 2 In the case of the firm as rational actor, we treat
individuals as a component in what often appears to be a multi-layered network;
partly mediated by individuals and partly by other linking vehicles. In the case of the
firm as political actor, the link will often be personal and fit the characterization of
simple tie, based on trust and tacitness. 3
We want to stretch the concept of social capital such that it might become an
extension of the individual as an office-holder in an organization and, consequently,
become an accessory for his firm's functioning. For example, an early study by
Pettigrew (1974) on the 'politics of organizational decision making' narrates the
position of an information technology specialist as a boundary spanner between his
firm and external vendors. As office-holder his significance derives from the quality
of internal and external embeddedness. We might then ask whether the office
holder's network connections can be combined with that of others into an index of
organizational social capital. Furthermore, inter-firm links might also be discernible
beyond the IT specialist, for example, by the long term outsourcing of data storage
and retrieval services, or the presence of a hot line with the IT consultants. Such a
link is not 'simplex,' but what might be called 'multiplex.' The Pettigrew example
illustrates the transition from the firm as a human aggregate to the firm as a
coherent, singular entity where the issue of aggregation becomes bracketed, or
remains altogether outside the purview of the observer.

It is problematic to move from the individual to the organizational level of analysis
when analyzing inter-firm networks. The issue of aggregation from the member to
the organization is primarily an issue when we view the organization as a Political
Actor in which the members' social capital aggregates to that of their organization.
Nevertheless, people associated with the organization as Rational Actor carry out
deeds on behalf of their firm, and while the model is agnostic about their integrity,
we could focus on their role as distinct linking mechanisms as well.
At face value, the individual-collective distinction seems merely conceptual, not
'real.' The issue oscillates between two frames: do individuals as agents or office-
holders connect organizations and other human aggregates? Or do organizations and
other human aggregates connect individuals? In this chapter, we are mostly
concerned with the first type of framing. Nonetheless, we recognize that many inter-
firm links condition the intermediation of individuals. In abstracting away from
individuals as mediators of inter-firm links we shift from the view of the firm as a
'political' actor to that of a 'rational' one (Allison 1971). The level of analysis
becomes moot and little need exists for acknowledging cognitive, cultural, or
strategic differentiation-whether in the organizational core or at its boundaries.
To the extent that aggregation surfaces as a salient feature, we should abandon
the neoclassical notion of the firm as a unitary actor with a well defined preference
ordering and whose strategy betrays a clear and unambiguous preference ordering.
Social Capital of Organization - 47

Its membership has a singular identity. The challenge for firms is to consolidate
divergent identities into a coherent one such that they might even approximate the
firm as a unitary integrated actor. The members are assigned to interlocked sets of
roles and they develop informal sets of hierarchical and horizontal relationships with
other people inside and outside the organizations. A large chunk of organizational
social capital exists by virtue of the individuals whose relationships span
organizational boundaries.
Some organizational participants are more contributory in their social capital
than others, depending on their involvement in the focal firm and its transacting
partners. Indeed, not all members are equivalent in their ability to leverage their
social capital for the firm. Members vary not only in their contribution to external
ties but also in their participation in the organization (e.g., Cohen, March, and Olsen
1972). When aggregating the social capital of members to arrive at a stock index of
firms, there is also the issue of redundancy. A network link is redundant if the
marginal increase in benefits from acquiring or maintaining that link equals zero.
Redundant ties have been well documented at the individual level, e.g.,
Granovetter's (1995) 'weak' versus 'strong' tie and Burt's (1992) presence or absence
of 'structural holes.'
The aggregation of the networks of organizational participants is prone to have
redundant contacts. The number of members maintaining contact with
representatives of other organizations might produce 'stronger' ties that are
particularly beneficial for the transfer of sophisticated knowledge. For the
transmission of information or what might be called 'explicit knowledge,' such
strong ties are hardly efficient (compare Hansen 1997). Furthermore, not all social
capital of members aggregates to the social capital of the organization. The social
contacts of certain organizational members may have little or no instrumental value
for their organization.4 Only overlapping membership in groups and organizations,
that are operationally or strategically relevant, matter when aggregating individual
social capital to that of the organization; the most common example is interlocking
directorates (Pennings 1980; Stokman, Ziegler, and Scott 1985).

Boundary Spanner or Multiple-Group Membership

The concept of overlapping membership as a way to represent an individual's social
capital should also be invoked to revisit the issue of a firm's boundaries. If members
vary in their inclusion in the focal organization, their external contacts should vary
in value as well. Even if organizational members have valuable external ties, they
become a valuable component of the firm's social capital only if the members enjoy
access to certain peers-for example, those with power, information, and other
resources. If inclusion is highly partial, their individual social capital becomes
marginalized for the firm as well.
For simplicity's sake, organizational members might be stratified into a core
group, a regular or associate group, and temporary or marginal workers. The core
group consists of essential employees who are long-term employees and owners.
Their fate is usually tied to that of the organization. The regular or associate group
consists of rank-and-file employees who have been involved in the organization for
'some' time and face good prospects to join the core group. Many members who
48 - Corporate Social Capital and Liability

participate in that tournament will 'plateau,' become sidetracked or might even be

terminated, however. The temporary or marginal category include temporarily hired
workers and employees of sub-contractors, i.e., workers who fill the jobs not
requiring firm-specific skills and who have little chance of moving into another
category of members. 5
It follows that the social capital associated with the core group is more
important for the organization than that of the regular group. The reason is two-fold.
First, members in the core group are more likely to use their social contacts on
behalf of the organization. Consistent with the garbage can model (Coh~!1, March
and Olsen 1972), these members have the highest 'net energy load,' as their fate is
closely tied to that of the organization. Second, they are likely to maintain more
valuable social contacts for the organization. They are more central to the access
structure, and enjoy higher positions with more power and authority. Many of the
firm-relevant social contacts are based on the job and title of individual members. A
CEO becomes a board member of a peer organization, supplier or some other
organization; a partner in a consulting firm befriends senior executives in the firm he
works for, etc. Compared to the employees in regular or temporary groups,
members in the core group tend to have social ties with people who occupy higher,
more visible and more prestigious positions in their organizations. In other words,
people who have social contacts with members in the core group of a focal
organization tend to have more valuable resources at their disposal for the focal
organization than do the people who have primarily social ties with members in its
more peripheral ones. Core members also stay longer with their organization such
that their organization stands to benefit more from their social capital. Overall, we
need to focus on the nature of the employment relationship to weigh an individual's
ability to link his firm with other ones.
Figure 1 provides a graphical display of organization stratification in terms of
magnitude of personal inclusion.
There are also other ways to compartmentalize the firm as a community of
people who are endowed with human capital, and who are differentiated by skill,
function, types of markets, products, or technologies. Firms have either a functional

Marginal: free agents, sub contractors,

contingent workers, strategic alliance
guest employees

Associate: temporary employees,

'in-transit' workers

Core: long term employees, partners,

owners, managers, residual claimants

Figure 1. Stratification of firms based on partial inclusion of their members

Social Capital of Organization - 49

or divisional (and in many cases some hybrid) structure whose boundaries define
identities. In fact, although finns proclaim to be a hierarchy that economizes on
transaction costs (Williamson 1975), they in fact comprise numerous sub-cultures,
with their own identity and parochial interests. While hierarchy and lateral linkages
integrate disparate units, they often face major hurdles in consolidating their skills or
knowledge, or more generally in bundling their contributions to the common good
(Brown and Duguid 1997; Kogut and Zander 1996). A firm's internal networks such
as heavy duty project managers (Clark and Fujimoto 1991), overlapping teams, and
interdepartmental career paths become vehicles for knowledge migration, but such
networks are often comparatively deficient because specialization impedes
knowledge transfer, especially knowledge that is difficult to package. Ironically,
communities of knowledge within the firm have often easier access to like-
communities in other firms than they do with the sister departments within their own
firm. The implication is that such external networks are often more efficacious in
bridging the firm with external actors than do networks that embrace the total
organization. By way of example, we might consider a firm's participation in an
'invisible college' less problematic than its participation in a trade association
(Powell 1990; Lazega, this volume).

Multiplex versus Personal Forms of Organizational Boundaries

At the level of inter-organizational relationships, we could make an even stronger
argument about the individually anchored social capital of organizations. When the
vendor of a software firm leaves, he might appropriate the connections with clients
that he has built up during his tenure. One might thus argue that the social capital of
organizations is tied up in the individuals they employ.
Yet, as with all intangible assets, social capital can also be treated as an
intangible asset that is not exclusively buried in personal networks. Social capital is
often 'depersonalized' or is couched in mUltiplex forms. Interorganizational links
established through individuals might begin to lead a life of their own. Or such links
become embellished by other glue such as contracts, traditions, and institutional
arrangements. The members who are then a complement to a system will in fact also
be governed by the norms and beliefs that are endemic to local social arrangements.
When links become multiplex, they cease to be dependent on individuals who act as
brokers. By way of examples, patent citations signal proximity of knowledge among
organizations and can be examined as a conduit for inter-firm knowledge transfer.
Cartels amount to a clique with shared norms where the members are firms rather
than people. A set of firms might be tied through mutual share holdings. Affiliation
among organizations, such as keiretzu in Japan, chaebol in Korea, or business
groups in Sweden illustrate bundles of inter-firm connections that cannot be reduced
to middleman-members.
Strategic alliances such as joint ventures, R&D partnerships, and minority
investments embody nodes in webs of inter-firm networks in the telecommunication,
micro-electronic and biotechnology industries (e.g., Ajuha 1998, Hagedoorn and
Schakenraad 1994; Omta and van Rossum, this volume). Severing some of these
linkages might be impossible. For instance, Microsoft has extensive lock-in
agreements with PC makers and their suppliers and PC manufacturers in fact have
50 - Corporate Social Capital and Liability

contracted for the pre-arranged installation of Microsoft's operating system in what

used to be called 'IBM-compatible' personal computers. Biotechnology firms'
entrenchment can be inferred from patent citation networks in which their
intellectual property is more or less linked with that of other firms ; the tightness of
their links is derived from the proximity as measured by relative citation frequencies
(Stuart, Hoang and Hybels 1997, see also the chapters in this volume by Stuart and
Smith-Doerr et al.).
All of this requires us to dissect the ingredients of inter-firm networks into at
least three categories: 6

Any sort of association between two or more firms, including equity cross-holdings,
patent-ties, licensing agreements, R&D partnerships, equity joint venture
agreements, gatekeepers, or interlocking directorates.
Human mediated links, such as interlocking director or guest engineer. Ties can be
'neutral,' reflexive (Pennings 1980) or even universalistic versus parochial and
Human mediated ties that are particularistic, as for example the guest engineer who
has an OEM employment status but resides on the premises of a supplier.

In short, corporate social capital bifurcates into personalized and depersonalized

forms, with relationships often augmented with ties and links; while in other
instances, the link might persist without the benefit of a relationship. This distinction
often corresponds to a simplex versus a multiplex web of network connections.
Multiplex 'links' appear to be more congruent with the rational actor metaphor of
Allison, while 'relationships' feature prominently in treatments of organizations as
political actors. Table 1 furnishes some examples. First, the organization itself can
have a link with other organizations that is instrumental for its functioning.
Affiliation among organizations, such as keiretzu in Japan or chaebol in Korea is a
social link of the organization itself rather than of organizational members. As a
legal entity, the firm is capable of contracting, of acting as a partner in any market
relationship, induding the setting up of joint ventures, the acquisition of another
firm, or the shedding of a business unit to other firms, etc. Indeed, independent of
their members, the organization often maintains social capital through the repetitive
exchanges with other organizations. The pattern of exchanges has stabilized, even if
the individual members who participate in the process have been changed (Chung
1996). Investment banks perpetuate their collective efforts when they syndicate
public offerings (Chung, Singh, and Lee 1995). Semiconductor firms joined
SEMATEe when they sought to acquire greater economies of scale.
Whether one assumes a personal or impersonal link (or a hybrid form
comprising both links and relationships) between organizations, links constitute the
ingredients of arrangements that govern the firm-environment interface. In some
cases the arrangements can be viewed in their own right, but their efficacy in
Social Capital of Organization - 51


Figure 2. Boundary transaction system comprising four individuals among two organizations

managing external dependencies depends critically on the quality of the relationship

with internal and external decision makers. Adams (1976) was one of the first
writers to review such arrangements. He refers to so called 'boundary transaction

Boundary Transaction System

Social capital fits with the notion of more or less permeable boundaries of
organizations that become spanned by a 'boundary transaction system' (Adams
1976). Figure 2 provides a graphic representation. As Table 1 indicated, such
systems diverge into pairs of individual dyads such as the interlocking director or
guest engineer whose role in maintaining the firm's network depends critically on a
balanced overlap between the inside and the outside. Or boundary transaction
systems are larger and more elaborate entities-for example kereitzus and R&D
partnerships. In the latter case the inter-firm link is not nearly as dependent on the
presence of boundary-spanning individuals such that the significance of their
mediation is comparatively minor. The personal ties often complement non-personal
ones such as reciprocal ownership arrangements and R&D partnerships.
Furthermore, the relative salience of the system hinges on the duration of links
that are maintained by individuals that make up the system. The longer the tenure,

Table 1. Examples of social capital among organizations

Mediated by Individuals (Simplex) Mediated by Systems (Multiplex)

Interlocking directorates (Pennings 1980) Business groups (Acevado et al. 1990)
Guest engineers (Dyer 1996a) Chaebol (Kim 1997)
Social register (Useem and Karabe11986) Keiretzus (Gerlach 1987)
Revolving door syndrome (Pennings, Lee Investment bank syndicates (Chung, Singh, and Lee
and Witteloostuyn 1998) 1995)
Alumni (McKinsey) Joint ventures
Double agent R&D partnerships
Gatekeeper (Tushman 1978) Guanxi-chia-jen (Tsui and Fahr 1997)
Emissary Electronic clearing house (Pennings and Harianto
52 - Corporate Social Capital and Liability

the more distinct the boundaries of the transaction system and the greater the
likelihood that its members 'go native,' i.e., acquire an identity almost different from
the firms they span. Consider boards of directors, or executive councils of Japanese
business groups who over time might become closely knit teams. Employees
originate from leading universities, where they have already formed friendship
networks, and synchronically move upward through equivalent organizational
ladders, such that the 'old boy network' remains intact from university years until
retirement. The implication is that succession patterns further strengthen the
boundary system's identity (Yoshino and Lifson 1986).7
The boundary transaction system is useful in that it points to the role of
member's social capital in producing organizational social capital. Likewise, by
recognizing that the system often evolves into a system that cannot be reduced to the
participating members, social capital might become depersonalized. The system
might become part of a business group, cartel, a joint venture, a long term licensing
agreement, or R&D partnership. Such systems are bound to become semi-
freestanding entities when three or more firms decide to participate. For example,
SEMATEC and ESPRIT are consortia of semiconductor firms that joined forces at
the behest of the US and European Union governments respectively to create what
we might call a boundary transaction system.
A key difference between a simplex and multiplex boundary system involves
the notion of trust. In a simplex system, trust is anchored in a dyad of trustor and
trustee who maintain a form of personal trust of what Simmel calls 'mutual
faithfulness.' Bradach and Eccles (1989) refer to expectations that the other side will
not behave opportunistically. It accords with the definition of trust by Mayer, Davis
and Schoorman (1995: 712)-a willingness of a party to be vulnerable to actions of
another party based on the expectations that the other party will perform a particular
action important to the trustor, irrespective of the ability to monitor or control the
other party. This definition excludes the social context of the dyad.
In multiplex systems, the social context becomes central and will in fact color
the nature of the relationships between individuals who are part of that system. The
context includes not only traditions, ties inherited from individuals who are no
longer present, contracts and financial leverage, but also forms of institutionalized
trust (Luhmann 1979; Shapiro, Sheppard and Cheraskin 1992; Zucker 1986). The
institutionalization evolves both temporally and spatially. Firms have often recurrent
contacts with other firms, and the history of their relationship provides a platform
for the current boundary system. Firms are also entrenched in larger entities, most
notably business groups. The firms that make up a business group share norms about
inter-firm transactions, have developed routines for contracting, and enjoy a group-
derived reputation that molds the dynamics of interpersonal relationships within a
boundary transaction system between two member firms. And history matters here,
too: the member firms have collectively gone through actions that resulted in shared
practices, mutual stock ownership, exclusive supplier-buyer relationships, or
investments in transaction specific assets (Dyer 1996a). The historical and spatial
context for two individuals who span their respective firms is therefore critically
important in comprehending corporate social capital.
Social Capital of Organization - 53

The fact that building up social capital requires time was nicely illustrated in the
recent difficulties between Ford and its suppliers. Ford sought to redesign its Taurus
model, while at the same time redesigning its boundary transaction system (Walton
1997). For example, the firm attempted to move from multiple, arm's length ties
with suppliers to single source relationships. Having made few investments in social
capital, its 'relational competencies' (Lorenzoni and Liparini 1997) for managing
such supplier relationships were grossly inadequate. The boundary system included
individuals such as Taurus project managers and representatives from 235 suppliers.
The project's social architecture was to embrace a Japanese-style long-term
cooperative relationship with suppliers. Yet, the culture of the system could be
described as 'You could not trust them.'
The boundary transaction system should not be confined to individuals who
gave rise to the system or were involved in its perpetuation. It ranges from dyads of
individuals to complex social, economic, and technological arrangements. It evolves
from individuals who interact frequently so that the firms become familiar with each
other. Familiarity alleviates transaction costs, improves coordination across
organizational boundaries, and reduces agency problems-in short the familiarity
that comes with organizational networks confers benefits. Familiarity also produces
group-think, cuts the firms off from important external stimuli, and renders it
increasingly inflexible. More specific benefits of social capital and the harmful
effects of social liability are discussed next.


At the onset of this chapter, social capital was mentioned as an integral part of the
organization'S intangible assets. The reference to assets suggests a rent producing
potential. However, social capital as such cannot produce rents, but it contributes to
greater rent maximization of other resources that complement social capital.
Burt (1992) points out that social capital is owned jointly by the parties to a
relationship whereas financial and human capital are the property of individuals or
firms. In other words, social capital is embedded in the positions of contacts an
organization reaches through its social networks (Lin, Ensel, and Vaughn 1981).
Second, social capital is related to rate of return in the market production function
whereas financial and human capital pertain to the actual production capability. We
should ask: What is the role of social capital in economic transaction? Under perfect
competition, social capital cannot generate any economic rents (Burt 1992). The
market however is hardly perfect and information is not costless. The member's
social capital strengthens his firm's ability to retain clients, perform market
intelligence, and learn about new technologies. This is particularly true in our
knowledge economy where many industries are characterized by abstract products
or services, whose quality and other dimensions are difficult to articulate and where
delivery of output is highly coupled with reputation (cf. Burt 1992). Clients resort to
their social contacts to screen their suppliers because assessment criteria for quality
might be hard to come by. While social capital is not part of the production function,
it has profound impact on the benefits that firms derive from their productive
capabilities. Putting it differently, social capital brings the opportunities to exploit
financial and human capital at a profit.
54 - Corporate Social Capital and Liability

In the next two sections, we belabor these implications by reviewing two

examples with rather different manifestations of interfirm ties: industrial firms that
make up business groups such as Keiretzus and Chaebol and professional services
firms that comprise the audit industry. We have hinted that these two examples
present different manifestations of a firm's external networking. Firms that belong to
a business group are typically depicted as ('rational') actors in a conglomerate-type
setting with mutual equity ownership, long term supplier-buyer transactions, and
shared directorships. The relational structure that business groups have is assumed to
furnish social capital to member firms. We impute such benefits to the firm without
confronting aggregation issues or delving into internal factions. Individuals are
merely one of the threads that make up the fabric of networks of business groups.
Thus the member-firms of business groups are depicted as integrated, unitary actors
who might benefit from their inclusion. The groups furnish interesting data on the
benefits of social capital and costs of social liabilities among firms that come close
to the stylized Allison-type Rational Actor.
In contrast, professional services firms belong to a sector that resembles a
cottage industry, where individual professionals appear to be the most salient
participants. While many professionals join a partnership and thus become co-
owners of the firm, these organizations are very flat and by dint of the
professionalization comprise members whose loyalty might be as strong to their firm
as it is towards the profession. The social capital of the firm might in fact be the
social capital of individual professionals. Even if we aggregate their social capital to
that of the firm they belong to, there always remains the issue as to whether it is the
partner, his peers as co-owners, or his firm who can make claims on the social
capital that is mediated by the professional. The professional has his own roster of
clients and might feel more loyal to those clients than to his brethren with whom he
makes up the partnership. His ties, and by implication his firm' s links, often fit the
notion of embedded ties. Arm's length transactions are incompatible with the
rendering of services, although some emotional distance with the client is often
deemed appropriate. Since partnerships often break-up, or witness an exodus of
partners, the caricature of Allison's political actor might sometimes be quite
appropriate as a general descriptor. Yet, as we will see we often have to qualify this

Business Groups
Social networks have been a pervasive feature of Asian socIetIes in general.
According to Hofstede's (1980) landmark study, Asian societies stress collectivist
values and cherish loyalty and commitment to family, organization, and community.
At the corporate level we also discern a preponderance of networking-most visibly
in business groups. Business groups include Japanese keiretzus, or their pre-war
predecessors, called zaibatzus, and Korean chaebols. These groups contain a myriad
of firms held together by ownership links, supplier-buyer relationships and mutual
guarantee for each other's bank loans. Other countries, most notably Sweden (e.g.,
Hakanson and Johanson 1993; Sundqvist 1990; Berglov 1994) and Argentina (e.g.,
Acevado et al. 1990), harbor business groups, but take on a local, idiosyncratic form.
Therefore it is prudent to limit ourselves to a relatively homogeneous class of cases
Social Capital of Organization - 55

(cf. Guillen 1997). Furthermore, some other Asian countries manifest distinct forms
of social capital among organizations; we could mention bamboo networks that are
depicted as a guanxi (relation)-based cluster of Chinese firms (cf. Tsui 1997;
Weidenbaum and Hughes 1996). In these cases, the individual as family member
performs a primary role in forging inter-firm links, and the family rather than the
firm appears to be the most salient unit of analysis. Unlike more centrally coupled
business groups in Korea and Japan, these Chinese forms of organization are
octopoid and opportunistically diversified (Tam 1990). In this section, we restrict
ourselves to keiretzus and chaebols.

Korean business groups manifest several features that set them apart from Western-
style business groups (Kim 1997). They display family ownership and management,
controlled by a powerful chair. The chair's power derives from stockholdings and
from being the father or senior family member who are heads of member companies.
Kim (1997) even refers to unquestionable filial piety and patriarchy based family
control within modem multinational firms. A founder' s descendents actively
participate in the top management of the chaebols. When the founder dies, his
descendents succeed as heir. When the founder with multiple descendents dies, 'his'
chaebol sometimes divides into several mini-chaebols as the case of Samsung
indicates. Still, the kinship and family networks link the member firms of those
Chaebols also exhibit high flexibility in mobilizing financial capital, technology
and human resources. Unlike keiretzus and zaibatzus (although the same Chinese
character is used to denote this extinct type of Japanese business group as well as
chaebol!) that are governed through consensus building and psychological
commitment, chaebols are nimble in their deployment of resources and the patriarch
can implement strategic decisions without consulting others. There is widespread
rotation of key personnel, R&D efforts are pooled across companies and transfer of
cash can be arranged through financial services firms, and the member companies
can guarantee each other's borrowings from financial institutions.
Finally, the complex set of networked firms that make up a chaebol are
exceptionally broadly diversified. Kim (1997) shows that a chaebol like Samsung
operated in light and heavy manufacturing as well as in financial and 'other' (e.g.,
construction, media, hospitality, and advertising) services. Presumably, such
diversification allowed chaebol to offset lack of high-tech skills by exploiting semi-
skilled and unskilled labor in a way that would not be feasible to a non-networked
competitor (Amsden 1989), while at the same time produce products that are price-
competitive rather than quality-competitive in the global markets. Compared to
keiretzus, chaebols are basically shaped on the basis of the founding family. The
financial institutions are less utilized to form the relationship within a chaebol's
member firms, because chaebols are blocked from owning more than 8 % of shares
in commercial banks.
Chaebol should thus not be confused with keiretzus or even with their name sake
zaibatzus although the degree of contrast is a matter of controversy. After the second
56 - Corporate Social Capital and Liability

world war, zaibatzus were dismantled but reappeared in a different form called
'keiretzus.' As a result of the transformation, the founding families of zaibatzus lost
their shares and power and thus were no longer a source of connections. The
insurance companies are at the keiretzus' apex, and from them cascades a transitive
pattern of equity cross holdings-the implication being that the insurance firms and
their executives are the ultimate center of power and influence (Nishiyama 1982).
Keiretzus' governance is much more decentralized with decision making among
firms by consensus rather than through fiat by the keiretzus' insurance firm's
executives. The zaibatzus provided a template and became mimicked by Korean
entrepreneurs and in any event evolved into a prominent form during Korea's
industrial revolution. Zaibatzus and chaebols share characteristics such as family
ownership, management by patriarch, and unrelated diversification. However, unlike
the chaebol, the zaibatzu also controlled commercial banks, giving them access to
capital markets.
Keiretzus are laterally federated with transitive stock ownership arrangements
that induce minimal interference in between-firm interactions, rather than
resembling a chaebol-like holding with a vertically arranged governance structure. s
Gerlach (1987: 128) refers to them as 'business alliances,' which he defines as the
'organization of firms into coherent groupings which link them together in
significant, complex long-term ownership and trading relationships.' They are
distinct in the manner in which they have established coordinative mechanisms to
govern their relationships. These include high level councils of executives, the
shaping of exchange networks, and the external presentation as a coherent social
unit, for example, through advertising and product development activities.
Prominent, but largely invisible in the structuring of network links is the role of
financial institutions, which unlike the chaebol are an important component of the
Japanese style alliance. The member firms are heavily indebted to the keiretzu's
main life insurance company and bank. The cross equity holdings constitute an
important link over and beyond the relationships that could be uncovered if one
were to have access to their inner circles. Unfortunately, no research exists on the
power structure within such circles, and the sort of collective decision making
processes that ensue. Thus we are also deprived from making strong conclusions
regarding the stock of social capital among keiretzus firms. These links are not
merely leverage tools, but in fact might acquire a significant symbolic meaning on
their own and complement other media of networking such as exclusive R&D
projects. The keiretzu as a somewhat hierarchical network is therefore multiplex-
debt holdings, cross-equity holdings, supplier-buyer links, and personnel bonds are
part and parcel of the connections that bind the firms into a tight and relatively
unified alliance.

Social Capital of Business Groups

Firms that are part of chaebols and keiretzus are presumed to benefit from the social
capital that ensues from their membership in these alliances (e.g., Kim 1997;
Lawrence 1991). Social capital is manifest in two ways: First, business groups
provide its member firms with access to resources from other firms. As a quasi-
holding or federation of businesses, they can furnish superior access to financial
Social Capital of Organization - 57

capital through a member financial services firm and cross guarantee each other's
bank loans. Similarly, business group specific suppliers and their Original
Equipment Manufacturers belonging to the same group display shorter lead times in
new product development because they circumvent transaction costs, for example,
by making significant asset specific investments that in the absence of a business
group context would incur significant hold-up problems (e.g., Dyer 1996a; Gerlach
1992b). The inclusion in the keiretzu reduces the outsourcing to one or at most two
suppliers, and the relationship is typically based on trust and mutuality. By way of
contrast, Toyota relies often on a single, keiretzu-anchored supplier, while US auto
manufacturers such as GM usually rely on as many as six suppliers, with whom they
interact opportunistically and at arm's length (Dyer 1996a; Nooteboom, this
volume). The suspension of the hold-up problem results also in joint R&D and in the
geographic clustering of OEM and their suppliers, thus economizing on value chain
coordination costs, transportation distance, and inter-firm transfer of tacit knowledge
(e.g., Hansen 1997; Nooteboom, this volume).
The social capital of business groups, however, is not confined to intra-group
relationships. Since their boundaries are also salient at the group level, they have
enjoyed scale advantages, not unlike those accorded fully vertically-integrated firms.
Such assertions question the saliency or distinctiveness of boundaries, and in
particular the issues associated with vertical integration, governance, and transaction
costs (Williamson 1996; Powell 1990). Even though inter-firm links are not
exclusively mediated by individuals-as we have argued they are mUltiplex, to say
the least-the links that bind them might be so strong that the focal attention often
shifts to that level of analysis when discussing social capital. They maintained levels
of flexibility in moving around human resources and other assets, and because of
superior access to cheap and unskilled labor, were able to claim cost leadership
positions in their world of multi-point competition. (Kim 1997: 180-195). Yet, on
the next higher level of analysis, these groups commanded clear benefits that
surpassed inter-firm arrangements, as reviewed by Powell (1990).

Empirical Evidence
In Korea, there is the often documented 'cozy' chaebol-government interface.
Chaebols as groups are often endowed with a good deal of social capital because of
the support they have extracted from the South Korean government. Compared to
non-chaebol firms, chaebols have had better access to state-controlled resources, and
were thus able to exploit governmental powers for their own benefit (Kim 1997).
The chaebol dominated segment of the economy grew much faster than the economy
as a whole.
The reasons that chaebols have received a great deal of governmental support
are two-fold. First, the sheer size of chaebols has made them very important for the
development-oriented Korean government. For instance, the value added by the 30
largest chaebols has been around 15 % of GNP and their sales volume has been
around 80 % of GNP (Cho, Nam, and Tung 1998). Since chaebols have been used as
a tool for the government's industrialization policy, the Korean government has
provided a great deal of favors including soft loans, import prohibition, tax breaks,
etc. Second, the relationships of elite university graduates strengthened the
58 - Corporate Social Capital and Liability

relationship between the Korean government and chaebols. People who graduated
from elite universities have occupied major positions in the Korean government,
banks, and parliament. As a result, chaebols appointed elite university graduates as
CEOs to lubricate their relationship with external entities. For instance, 62% of
CEOs of the seven largest Korean chaebols in 1985 graduated from Seoul National
University (Steers, Shin, and Ungson 1989).
There is also some provisional evidence that member firms within a chaebol or
keiretzu might encounter the adverse effects of 'over-embeddedness.' In Korea we
have the case of the Kukje chaebol and recent bankrupcies of major chaebols, while
in Japan the differential learning of keiretzu versus non-keiretzu suppliers provide
testimony to the harmful effects ('social liability') of social embeddedness. The
Kukje case emerged in February 1985 and evolved from an ordinary bankruptcy into
a scandal when the Chun government disbanded the chaebol due to 'reckless
management, and exceedingly high debt rates.' It is most relevant for our argument
because of 'nepotic management by the sons of the founder' (Kim 1989). The
bankruptcy case is somewhat ambiguous and opinions varied as to whether it was
over-embeddedness among member firms or deficient external social capital that
accounted for the disbanding of Kukje. Yang, the chaebol president, claimed
favoritism on the part of the Chun government. In any event, further research should
identify whether it was social capital at the group level or at the group-state level
that explains the demise of Kukje.
Due to the risk-sharing role of chaebols, Korean chaebols enjoyed very high
survival chances and thus only a few chaebols experienced bankruptcy. During the
period of January 1997-January 1998, however, nine chaebols among the 30 largest
chaebols experienced insolvency. The mutual guarantee of bank loans made whole
member firms rather than some of them insolvent. In some cases, the failure of one
member firm became the reason of the bankruptcy of the chaebol. Over-
embeddedness to other member firms rendered profitable and financially sound
member firms bankrupt, thus revealing the 'dark side of social capital' (Gargiulo and
Benassi, this volume).
Keiretzus in Japan also function as a tool for risk sharing among member firms
(Nakatani 1984) and thus they enjoy a lower bankruptcy rate (Suzuki and Wright
1985). However, criticism has surfaced regarding their traditionally claimed
advantages. Gerlach (1992b) sees the potential unraveling of keiretzus now that their
benefits have appeared to wane. Nobeoka and Dyer (1998) have recently completed
a survey of OEM-automotive supplier relationships and produced evidence
indicating that suppliers that diversify away from a single keiretzu based OEM are
more profitable compared with firms who are locked in a close single-source
relationship. They interpret this finding as being due to either superior bargaining
power, or to a broader exposure to technological know-how; such firms diminish
their dependence on a single OEM or they witness learning benefits in that their
know-how is likely to be more generic and less firm-specific.
Similarly, Lincoln, Ahmadjian and Mason (1997) provide evidence of Toyota
the auto manufacturer and Toyota the keiretzu member, which diversified away
from keiretzu-based automotive suppliers. These authors report that intra-keiretzu
knowledge was not only limited, but that Toyota did not even attempt to elevate its
Social Capital of Organization - 59

'internal' suppliers to the standard that would meet its needs. The implication is that,
in spite of trust and inbred capabilities, the firm begins to question the benefits of
traditional arrangements. Such precedents might lead the keiretzu on a path of
further unraveling its stale social capital and the substitution of a fresh one.
Summarizing, business group's endowment of social capital should be
differentiated into that social capital that is discernible at the group level versus that
which resides at the interface between the business group and external actors. The
beneficiary of social capital is the firm or a group of firms who are portrayed as
unitary actors, operating in their economic-political arena. The evidence so far has
focused on the social capital inherent in social structure, but more recent evidence
shows also that over-embeddedness might lead to social liability.

Audit Firms
The accounting sector presents another setting in which the costs and benefits of
social embeddedness are evident. Unlike markets with industrial firms, as is the case
with industries comprising business groups, the accounting sector produces largely
intangible and abstract services. The measurement of product quality is elusive, the
production flow is exposed to the client who is often an active co-producer of the
services rendered. The firm has some degree of hierarchy but is usually much flatter.
In fact most firms are stratified into partners (i.e., owners) and employees, some of
whom expect to join the partnership. Their close exposure to the market place and
their intense involvement with clients makes social capital a central feature of
operations and a key driver of organizational performance. This sector resembles
numerous cottage industries where personally mediated ties predominate, not unlike
the settings of garment district members (Dore 1983; Uzzi 1997a), or investment
bankers (Burt 1997).
Ironically social capital can be viewed as a substitute for objective criteria of
quality, reliability and consistency. In the absence of objective, verifiable and
measurable product attributes, clients might rely on their networks to select auditors
or to remain loyal to them even after the honeymoon period has passed (cf. Podolny
and Castellucci, this volume). The endowment of social capital is therefore a critical
resource in such sectors. Absent social capital, the firm might not extract much rent
from its human capital. Furthermore, social capital allows the firms to leverage their
human capital thus extracting more quasi rent from that asset. Social capital is not
only valuable as rent producing potential, but is also scarce and difficult to
appropriate. These aspects suggest social capital as a resource not unlike brand
equity, reputation and goodwill, and should be further explored here.
As we indicated at the onset of this essay, social capital fits Barney's (1991)
criteria of the resource-based-view of the firm. Resources that provide a competitive
advantage should be valuable, rare, hard to imitate, and imperfectly substitutable.
Applying these conditions to accounting firms and other professional service
sectors, it appears therefore obvious that the social capital of an audit firm forms a
major source of competitive advantage in this 'knowledge' sector. Social capital of
audit firms has a rent-producing potential, in that it is valuable and scarce (product
market imperfectness) as well as imperfectly tradable (factor market imperfectness).
Araujo and Easton (this volume) employ a similar list when they conceptualize
60 - Corporate Social Capital and Liability

social capital through a 'relational' lens. Let us review these aspects of social capital
in closer detail in order to reveal their role in explaining the benefits of embedded

As far as the value argument is concerned, a substantial number of studies in
sociology have shown that social ties transfer influence and information (e.g., Burt
1992, 1997; Coleman, Katz, and Menzel 1966). At the individual level, the benefit
of having supportive relations has been welI established. Supportive relations
contribute to getting a job (Granovetter 1995), high compensation (Boxman, De
Graaf, and Flap 1991), and promotion (Burt 1992). We argue that this argument
pertains to the (audit) firm level as well. Burt and Ronchi (1990) and Burt (1992)
applied the notion of social capital to organizations. Burt (1992: 9) pointed out that
'the social capital of people aggregates into the social capital of organizations.'
Social capital amassed in the organization's members is among the firm's most
valuable productive assets (Burt and Ronchi 1990). Unlike the setting of business
groups, in this sector we can define an organization's social capital as the aggregate
of the firm members' social capital. An individual member's social capital is
captured by his connectedness with client sectors.
Why would audit firms with social capital enjoy competitive advantages and
higher survival chances? That is, what is the role of social capital in the economic
transaction of providing audit services? Under perfect competition, social capital
cannot generate any economic rent (Burt 1992). However, the market for auditing
services is hardly perfect, and information about audit services is not costIess. The
owner's social capital strengthens his firm's ability to retain and attract clients. This
is even more true in the audit industry, where information with respect to qualities of
professionals is hardly perfect (cf. Burt 1992; Polodny and Castelluci, this volume).
Clients resort to their social contacts to screen their service providers, because
assessment criteria for auditing quality are hard to come by. Crucial contacts include
those that involve the client sectors that an audit firm serves. There are three reasons
why network ties with client sectors may well facilitate the building and retention of
First, people tend to rely on their current social relations to alleviate transaction
cost (Ben-Porath 1980). A stranger who does not anticipate an enduring exchange
relationship, has an incentive to behave opportunistically. To curb this malfeasance,
ill-acquainted exchange partners typically rely on elaborate, explicit, and
comprehensive contracts. These contracts, however, are difficult to write and hard to
enforce (Williamson 1975). Mutual trust between the actors, developed through
repetitive exchanges, obviates the need for writing explicit contracts. If the creation
of trustworthy social relations were costless, however, the existing network ties
would not confer benefits to those who nurtured them. In reality, individuals and
organizations have to invest substantial time and energy in forging durable relations
with others (Burt 1992). Variations in networking among firms should then
contribute to differences in the firms' ability to attract clients. Second, trustworthy
relations produce information benefits for the linked actors (Burt 1992). Information
is not spread evenly across all actors. Rather, its access is contingent upon social
Social Capital of Organization - 61

contacts (Coleman et al. 1966; Granovetter 1985). An actor cannot have access to all
relevant information, nor can he process and screen all important information
single-handedly. Being embedded in a network of relations allows a particular actor
to economize on information retrieval. Second hand information, at least, serves to
signal something to be looked into more carefully (Burt 1992). Personal contacts
also make it possible for the involved actors to acquire the information earlier than
others. Third, trustworthy relations enhance the possibility for an actor to refer his
contact person ( for example, an auditor, physician or management consultant) to a
third party (i.e., 'tertius'). Burt (1992: 14) puts the benefit this way: 'You can only be
in a limited number of places within a limited amount of time. Personal contacts get
your name mentioned at the right time in the right place so that opportunities are
presented to you.' The counterpart in a dyadic relation can playa role as a liaison to
link the social actor to third parties.

The argument as to the scarcity issue is, again, specific to the CPA profession. The
CPA profession is there to attest financial outlets of organizations. In effect, this was
the very reason for the origination of the profession. In away, this is comparable to
other public professions. For example, police officers are trained to perform their
public, and legally protected, role of preventing and bringing action against
violations of the civil order. In a similar vein, CPAs are expected to prevent and
bring action against violations of the 'financial order.' Therefore, CPAs are trained to
perform their public attesting role-this is the core of any CPA education program.
This very nature of the profession implies that the majority of CPAs are employed in
public practice, working within audit firms rather than client organizations. Only a
minority is attached to internal control jobs within client organizations. Hence,
social ties that come with current (or previous) partners or associates with previous
(or current) employment outside the audit industry - i.e., through jobs in
governmental bodies or private enterprises - are not abundant. For example, in 1920
roughly 80 % of Dutch CPAs worked in public practice. In the period from the
1960s up until the 1980s, this percentage dropped to slightly above 50 %. Hence,
there is much room for audit firm heterogeneity in this respect, both in time as well
as over time.

Apart from market imperfection (resource value and scarcity), nontradability is
needed to guarantee the sustainability of rent appropriation. Social capital is
tradeable, however, though all but perfectly. Within audit firms, an individual CPA
handles a set of client accounts. That is, from the perspective of the client there is a
double tie to the audit service supplier-i.e., to both the audit firm and the individual
auditor. For one, client loyalty to the audit firm is rather high. This is particularly
true for large companies, which rarely switch from one audit firm to the other
(Langendijk 1990). Among small and medium-sized client firms, audit firm
switching may well be common, though. Additionally, however, a client's financial
reports are attested by an individual CPA. This introduces a tie to the individual
auditor, too. In many cases, the auditor' s position involves confidentiality and trust.
62 - Corporate Social Capital and Liability

In a way, the auditor develops into a mediator who plays an advisory role in a wide
array of financial and even non-financial issues. So, social ties are partly linked to
the audit firm, and partly to the individual auditor. This implies that by moving to
another firm, an auditor only depreciates part of this social network, because client
sector ties are both an integral part of the firm as well as linked to the trust
relationship with the individual CPA. Of course, the partner-associate distinction is
relevant from the observation that ownership is associated with limited mobility.
Finally, we should mention that during the last half century partnership
contracts have further diminished the portability of social capital. In both the US.
Europe and elsewhere, partnership agreements typically contain a clause that blocks
partners from taking clients with them in the event they leave the firm. Needless to
say, such contractual constraints bolster the non-tradability assumption of a firm's
social capital. Such clauses have also become standard since the second world war
and diminish the mobility of a partner's roster of clients.
In sum, a firm of which partners are tied with potential clients is better
positioned to build clientele since a potential client can 1) actually become the
firm's client, 2) provide valuable information about potential markets, and 3) refer
the firm to other potential clients. These aspects should strengthen a professional
service firm's survival chances

An Empirical Test
An empirical study of the Dutch accounting industry over a period of 110 years
(1880-1990) was used to test the proposition that social capital diminishes the
likelihood of firms getting dissolved. Social capital was proxied by various
measures. For example, 'partner from client sectors' was the proportion of partners
who worked in client sectors (i.e., other industries or governmental agencies). They
are assumed to have more valuable network ties with potential clients than partners
without job experience in client sectors. When departing partners find employment
in client sectors, they are likely to have an affiliation that can utilize their
professional knowledge. 'Controller' and 'chief financial officer' are examples. As a
result, they are likely to be in a position to choose a professional service provider.
Because they have strong incentive to take advantage of their social capital, they are
likely to choose the professional service firm they worked for (Maister 1993; Smigel
1969). To reflect this effect, the study included a 'partner to client sectors' variable.
This is the proportion of partners who left the firm within the previous ten years in
order to work for other industries or governmental agencies. A ten-year span was
adopted for two reasons. First, the strength of network ties may decrease over time,
as the departed partners develop new network ties. Second, the departed partners are
ultimately bound to retire from the business world and thus no longer provide
economic opportunities to the firm. Note that these proxies of social capital derive
from the mobility of professionals who move through a revolving door between two
firms. Much of the social capital literature assumes stationary individuals who link
two or more organizations through overlapping membership, for example,
interlocking directorates.
To test the hypothesis regarding social capital and dissolution, a hazard analysis
Social Capital of Organization - 63

Table 2. Complementary log-log regression of firm dissolution

Variables Model 2
b (s.e.)
Intercept -.450 (.934)
Partners 'from' client sectors -.090*** (.033)
Partners 'to' client sectors -.0\3*** (.001)
Heterogeneity in partners' origin -.061 (.062)
Heterogeneity in departed partners' destination .024 (.076)
Partners' industry-specific human capital (Graduate school -.138*** (.040)
Partners' industry-specific human capital (Industry tenure) -.\06** (.050)
Partners' industry-specific human capital (Industry-tenure)2 .145*** (.032)
Partners' firm-specific human capital -.236** (.107)
Partners' firm-specific human capital 2 .226*** (.034)

Log-likelihood (Degrees of Freedom) -2060 (39)

chi-square compared with previous model (d.f.) 118*** (9)
Notes: *p < .10, **p < .05, and ***p < .01 (two-tailed test)
Data: 1851 firms, 8696 firm-intervals, and 1164 firm failures .
Regulatory, historical, industry level (e.g., density), firm level (e.g., size, age) and control
variables not displayed (compare Pennings, Lee, and van Witteloostuyn 1998). Model 2
includes control variables, but not the variables involving associates' human and social capital.

was conducted on these firms, while controlling for numerous other variables (e.g.,
industry level variables such as density, size distribution, history and regulation; and
firm level variables such as firm age and size). The rent producing potential of
human capital is conditional on the firm possessing social capital. Further details are
provided in Pennings, Lee and Witteloostuyn (1998). A partial display of the results
is provided in Table 2.
The results were supportive of the hypotheses. In Table 2 we present the results
involving the human and social capital of owners, i.e. the partners without showing
the simultaneous effects of numerous control and other variables, including those
that are associated with firm and industry characteristics. Consistent with the
hypothesis, all coefficients of the social capital proxies were negative, indicating
that a firm's social capital statistically significantly decreases firm mortality. The
effect was statistically significant for two classes of social capital: ties that derive
from the recruitment of professionals out of the accounting firm's client sectors, as
well as ties that are associated with a firm's 'alumni' who after the tenure in the firm
have moved to client sector firms . The heterogeneity in bundles of social ties, as
derived from the interfirm mobility of professionals did not statistically significantly
affect the firms. When we add the proxies of human and social capital involving
associates to the model of Table 2, it was found that associates' social capital does
not seem to benefit their firm. In short, this study provided some important findings
regarding the beneficial effects of social capital.
64 - Corporate Social Capital and Liability

What Further Implications Regarding Social Capital?

This chapter illustrates the benefits and drawbacks of social capital that are either
mediated by individuals or become formed through an array of linking vehicles such
as cross-stockholdings and long-term buyer-supplier relationships. We have
suggested that the model of the firm conditions our conceptualization and
operationalization of social capital and the consequences associated with social
capital. Firms are conceived of as unitary actors that interact with other actors (e.g.,
peer firms in business groups), or they can be conceived of as a community of
practices and aggregates of individuals with their distinct objectives and unique
agendas (e.g., professional services firms). Allison's (1971) labels of rational and
political actor correspond with these stylized forms of organization. In the former
case, social capital inheres in the multiplex arrangements that bind a firm to other
actors. In the latter case, we focus on individuals and their ties that aggregate to
organizational social capital. We then set out to review the benefits of social capital
as a distinct organizational (intangible) asset.
Mediated by individuals, social capital nonetheless can be viewed as an
organizational property. The individuals might be stationary (as illustrated by the
linking pin (Likert 1961) or double agents) or they might migrate between firms (as
illustrated by the revolving door syndrome). The relative inclusion of the individual
defines its functionality for information and knowledge transmission: the personal
needs to be available for external linking, yet requires also sufficient proximity to
internal members and groups who can convert the flow of knowledge and other
resources into some competitive advantage.
Individuals can also mediate social capital in the case of business groups. In
fact, some of the pertinent literature has focused on individuals as transmitters of
knowledge between firms they span-for example, so called guest engineers who
are employed by the OEM or its supplier and are assigned to work in the partners'
site, or civil servants who have been recruited by a chaebol firm and join their ranks.
For example, in the above Kukje bankruptcy, it has been suggested that the chaebol
management shunned participation in semi-public sectors such as the I1hae
Foundation, thus depriving themselves from individually mediated social capital.
The Pusan based chaebol neglected to maintain part of its boundary transaction
system. What sets business groups apart from partnerships, among others, is that
business group links are typically mUltiplex, comprising both personal and
impersonal means for maintaining durable links.
In spite of such differences, this chapter has indicated that network
embeddedness can have both positive (social capital) and negative (social liability)
consequences. The links that bind provide access to competitively critical resources,
but they can also be so binding that they are stultifying and rather harmful. The case
of Kukje illustrate the deleterious effects of embedded ness that becomes fractured as
a result of governmental interventions. The inclination of Toyota to reduce its
embeddedness within its keiretzu signals a desire to increase the flow of novel
information that current links cannot furnish; its conventional supplier links might
be too limited in contributing potentially innovative ideas. The negative first order
effect and positive second order effect of social capital on performance in the
Social Capital of Organization - 65

apparel industry might be the most robust finding to date regarding the paradox of
embeddedness (Uzzi 1997a).
Uzzi (1997a, this volume) makes the important observation that embeddedness
is a two-edged sword. Embeddedness ranges from 'under-embedded,' via 'integrated'
to 'over-embedded networks.' As was shown, this distinction hinges largely on
whether links are 'arm's length'(i.e., contacts based on selfish, profit seeking
behavior) versus 'embedded' (i.e., contacts based on trust and mutual intimacy). A
firm's network that comprises largely arm's length links does not confer much
advantage in knowledge transfer, coordination, or strategic alignment. Conversely, a
firm that is strongly entrenched in embedded networks might become so insular that
it suspends exposure to markets and technologies that reside outside its immediate
It appears that these distinctions do not readily map on the two contrasting cases
we presented in this chapter. The partnerships in a professional services sector fit the
conceptual distinctions between arm's length and embeddedness, together with their
functionality such as trust, tacitness of knowledge being transferred, and mutual
adjustment (Thompson 1967) as coordination mode. At face value, partnerships are
internally personalized and anchored in trust, and so we would expect some of the
relationships to be among professionals and their clients. Uzzi's (1997a) case
involves similar Gemeinschaft-like firms, i.e., small entrepreneurial firms, mom and
pop, a trade making up a cottage industry-in short, organizations in which face-to-
face relationships predominate and which often become extended externally. The
apparel world resembles the Chinese 'bamboo network' (Tsui 1997) and Dore's
(1983) description of the Japanese textile industry, which he labels as 'cottage
industry' and in which goodwill becomes the central feature in describing the
prevailing trust and mutuality. The network ties are largely mediated by individuals.
How do we map these descriptions onto the social capital of firms in business
groups that tend to be mUltiplex? Are such links more Gesellschaft-like in their
appearance and functionality? What sort of processes can we envision in a boundary
transaction system in which personal ties complement contracts, equity cross-
holdings, and traditions that outlive their instigators? We should ask such questions
particularly when the individuals in the boundary transacti0n system 'do not go
native,' and continue to link up with people and groups in the firms they span,
together with other elements that define their inter-firm context. The issue is
germane to our earlier review of the firm as a layered entity in which the boundary
spanning system resides largely in the more peripheral bands. Such networks abound
with actors possessing 'structural autonomy' (Burt 1992) and creating opportunities
for opportunism, information asymmetry, and knowledge hoarding--opportunities
which Uzzi considers antithetical to embedded ties.
The implication of these observations is to recognize the two faces of
organizations and to develop divergent frameworks for capturing the performance
implications of network embeddedness. Without forcing us onto a meso-level of
research, by artificially integrating face-to-face and small group dynamics with large
scale firm-interface arrangements, we might develop a middle range theory of social
capital that fits the specific questions we might ask. Whether organizations have at
least two faces, or whether we invoke two cognitive models of organizations might
66 - Corporate Social Capital and Liability

be an issue left to philosophers and epistomologists. Empirically, we might envision

a continuum in which organizations range from highly cohesive, well bounded
aggregates that are tightly coupled and present few if any intrafrrm hurdles for
coordination, knowledge sharing, and strategic positioning. We can also envision
organizations that are loosely coupled, with permeable boundaries and few isolating
mechanisms, barely holding themselves together and maneuvering on the brink of
dissolution. In either case, the firm is part of a larger context. How they position
themselves onto this continuum, and what image we impose on them remains a
never-ending challenge. The research on social capital will shed further light on how
they negotiate their embeddedness, and what sort of advantages and shortfalls they
derive from that social structure.

We appreciate the comments from Jon Brookfield, Jeff Dyer, Giovanni Gavetti, Jim Lincoln, Lori
Rosenkopf, and Brian Uzzi.

1. Embedded ties could have two (if not three) rather divergent meanings: I) ties that are reinforced by
mutual feelings of attachment, reciprocity, and trust; and 2) ties that are a link within a larger set of links
and nodes. Since Uzzi's work is confined to dyads, the first meaning applies. When members of a dyad
become affected by third parties who envelop their tie, as in Burt (1992), the second meaning applies. In
both cases, the concern is with a focal person. If one moves to an even higher level of analysis 3), as for
example the internet, transactions among textile traders in 15th century Florence and Flanders, or
community power structures, then the network takes primacy over the ties between individuals who are
embedded in those networks. A person's or firm's 'centrality' conveys relative access to other actors in the
network such that a focal actor's social capital hinges partly on the direct and indirect ties that the tied
partners possess (e.g., Levine 1972). Empirically the effect of centrality on firm behavior or performance
has not been studied adequately (an exception is Freeman, this volume).
2. Note that the rational model of the firm does not presume anything about its embeddedness here. In
either the rational or political scenario, we do not assume organizations to behave as if they are atomized
from the impact of their relations with other organizations, or from the past history of these relations. If
we were to extend methodological individualism to the embeddedness of firms, we would not be able to
furnish an adequate account of how firms' actions combine up to the level of the value chain, markets or
institutions. We only make the analytical distinction based on the relative saliency of aggregation when
examining social capital as a firm-specific asset.
Hence, our reluctance to include Allison's second model, the 'organizational actor model' in our review.
In the extreme, over-socialized individuals would reduce to mentally programmed automatons who
mechanically replicate the routines that the organizational socialization process has imprinted onto them.
As role incumbent, they would have no discretion to embellish their position or protect personal interests,
nor could they be construed as the personal authors of their social network.
3. Some examples might illustrate the issues at hand. Firms are tied to each other through trade
associations, business groups, consortia, cartels, joint ventures, and directors who sit not only on their
board but on the boards of other organizations as well. They are locked into licensing agreements and
long term supplier-buyer arrangements, and might have made significant investments in specific inter-
firm relationships. The presence of such links and their benefits seem obvious, when that capital is treated
as firm level or individual level phenomena.
For example, Boeing's 747 aircraft requires the input from numerous contractors and sub-
contractors-only certain chunks of the cockpit and wings are developed and produced by Boeing. Such
inter-firm transactions result in long term links that become independent of the members who forged
them originally. Many firms occupy positions in the value chain with interdependencies so dense that one
might consider the value chain to be a more salient unit of action than the firms that exist within the value
chain. A simple illustration from the computer industry might further illustrate this observation.
During the main-frame computer era, it was common for firms like mM and Hitachi to control all
the steps in the value chain, from silicon, computer platform, system software, application to distribution
and service. The firm was the value chain, and competition between corporations matched competition
Social Capital of Organization - 67

between value chains. In the late nineties, we observe a fragmented horizontal competition between finns,
but vertically dense complementarities have surfaced. Microsoft competes with Apple and Unix, but is
symbiotically linked with upstream PC manufacturers and their suppliers, such as Intel. Downstream, the
finns relate to distribution and service finns such as computer stores and mail order firms. Microsoft has
been a shrewd exploiter of network externalities: the various technologies require complementary
products, lead to the fonnation of virtuous cycles such as software developers writing more Microsoft
Windows applications, and when these become available, more customers adopting Microsoft Windows.
Increasingly all firms in the value chain become 'locked-in' (or locked-out!) resulting in a complex string
of links that are straddled around a dominant computer design (e.g., Yoffie 1996). In such a value chain,
links are often de-personalized and it is the organizations that become the salient unit of the network. The
ties in such networks are critical for the finns involved as their products and technologies become heavily
intertwined with those of others.
Much of the social capital literature has an individual slant (e.g., Burt 1997) and finn attributes have
often been examined as an individual manifestation. Burt's (1997) recent study examines investment
banks but really focuses on its traders and the 'structural holes' that benefit the size of their perfonnance
based bonuses. One might also focus on their banks' tombstones and the social capital that could be
inferred from them. Coleman's (1988) classic example involves the tight social circle of diamond traders
in New York whose smooth and paperless transactions hinge on the social ties that they maintain with
other traders. The trust that is sustained within such a network results in a substantial reduction of
transaction costs. Likewise, he (Coleman 1988) shows that children whose parents know other parents
and teachers are better embedded in their school community and show lower dropping-out rate. Finally,
Uzzi (1997a) recounts the linkages among individuals who make up the New York apparel industry. In
such instances, the issue of aggregation and presumption of finn as a unitary actor is rather moot: the
entrepreneur is the finn . In these and many other contributions, social capital is a resource that belongs to
the networking or interacting individuals and that might affect the venture with which the embedded
individual is associated.
4. By the same token, an individual who is neutral to the bridging between two finns cannot easily be
incorporated in the organization's social capital. Referee, arbitrator, or mediator roles are sharply
different from those we associate with ambassador, spy, or guest engineer. The fonner's neutrality might
depreciate or sanitize whatever infonnation or knowledge the 'middle-man' furnishes to the linked
organizations. His neutrality also precludes intimacy and creates social distance. We assume that
organizations have discrete bundles of knowledge and infonnation whose rents will be augmented by the
development of 'proprietary' social capital.
5. Sherer (1995) identifies three major types of employment relationships. The first is the employment
relation coupled with ownership. It includes employees who share the risk of organization via various
incentive systems which link their earnings to the perfonnance of the organization. Employees in that
relation constitute the core group in our analysis. The second is the traditionally described employment
relation in which employees receive a fixed amount of earnings, provide a fixed length of time, and
perfonn work based on the direction from the supervisor or job description. Employees in these types are
designated regular group in the present discussion. The third embodies relationships that involve
temporary employment or contracting out. Employees in this type fonn the temporary or marginal group.
Note that with the rise of temporary employment agencies, outsourcing and sub-contracting, this latter
group has acquired huge proportions. Analogous distinctions have been made by Jensen and Meckling
(1976) and Milgrom and Roberts (1992).
6. The classification was suggested by Jon Brookfield.
7. For example, Toshiba and Tokyo Power maintain close buyer-seller relationships; they both draw
graduates from Tokyo University who get promoted in their respective companies, and they move in
tandem, their roles might change but their mutuality stays intact. The demography of the system co-
evolves with that of the respective organizations. Such evolutionary arrangements ensure network
continuity throughout the firms' history.
8. The tenn transitive cross-equity holding refers to a string of keiretzus finns between which
ownership is mutual yet unequal. Nishiyama (1982) reports the pattern of large block holdings in the
Sumitomo Business Group, with Sumimoto (S.) Life Insurance owning a larger percent of shares in S.
Bank, S. Metal, S. Chemical, S. Electric, etc. than vice versa; it augments its power over these finns
because these finns in tum own shares in each other, such that cumulatively, S. Life Insurance scores
highest on the 'comprehensive power index.'
A Relational Resource

Perspective on Social Capital

Luis Araujo
Geoff Easton

This chapter reviews the notion of social capital from a resource based perspective.
We argue that the notion of social capital relies on a metaphorical mapping of
features associated with economic notions of capital or assets into the social domain.
We start from the notion that not all economic resources can be classified as assets
in the way the term is deployed within legal and accounting language, and argue that
social capital shares many features with other less understood and intangible
resources. By employing a framework to examine the multifaceted and relational
dimension of resources, we examine in detail the entailments of the social capital
metaphor and relate to current applications within the business and management
literature. We conclude by reflecting on the characteristics of social capital as an
economic resource and caution against the dangers of engaging in facile
prescriptions based on a cursory understanding of the logic of accumulation and use
of social capital.

The notions of embedded ness (Granovetter 1985), social capital (Bourdieu 1986;
Coleman 1988, 1990), and social resources (Lin 1990) have, in recent times,
contributed to a rekindling of interest in the interaction between the economy and
society. At the heart of this revival is an attempt to trace the mutual influences
between economic exchange and the social structures in which the economy is
The notion of embeddedness relies on the insight that economic life is shaped
and constrained by norms, social networks, institutions, and a variety of motives
A Relational Resource Perspective on Social Capital - 69

other than the unconstrained pursuit of self-interest. The revival of the interest in the
social has also got to do with the increasing awareness that modern economies,
while relying on impersonal forms of exchange and complex forms of contracting
with third-party enforcement, cannot dispense with other forms of support in the
guise of moral, social rules or codes of conduct (Platteau 1994ab).
The emergence of market economies does not diminish the need for social
solidarity and trust. On the contrary, as Macneil (1986: 592) argues, market
economies have acute problems regarding social solidarity. The embeddedness of
economic exchange in social structures very often dictates complex legal structures
remote from, though essential to, the exchange relations themselves. Thus markets
cannot be regarded as a spontaneous order or a primitive state of nature, but a
convergent network of actors and institutions mixing different forms of exchange
and where order is generated through translation processes and rules that are
reproduced across exchanges and over time.
Market order is partially generated by institution building, to establish and
enforce sanction systems and solve coordination problems involving the risk of free-
ridership and dilemmas of collective action. But, as Bates (1988) argues, formal
rules enforced through third parties can also be subjected to free-riding, and the role
of institutions as impartial rule makers and enforcers can be questioned. The
problem cannot be resolved through appeal to a further tier of institutions to monitor
the performance of the first tier. In short, order can emerge only in the presence of
both institutions promoting and enforcing formal rules, and informal norms such as
a generalized morality that draws on a society'S social fabric and culture. These
informal norms can thus act as substitute for or a reinforcement of formal rules and
control mechanisms, with the consequence that coercive enforcement of formal
norms becomes either redundant or of secondary importance. 1
The objective of this chapter is to revisit some of the issues concerning the ways
social structures impact on economic exchange. In particular, we are concerned with
the ways the notion of social capital, residing both in concrete, interpersonal
relationships inside and outside formal organizations as well as in wider social
structures, can be deployed to break down some of the artificial divisions between
the economy and society. The structure of the chapter is as follows: in the first part
we look at the notions of embedded ness and social capital and the way they have
been used to understand the coordination of socioeconomic life. In the second part
of this chapter we use a framework we developed to dimensionalize economic
resources (Easton and Araujo 1996), to look at the characteristics of social capital as
an economic resource. We conclude with some speculations on the role of social
capital in the coordination of socioeconomic life.

The Notion of Embeddedness

The key issue that arises in discussions of embeddedness and that is germane to the
notion of social capital is the extent and form of the embedded ness that we might
expect to find in a modern economy. Below we review briefly some of the more
recent offerings to give a flavor of the debate.
The embedded ness argument is given short shrift in traditional neoclassical,
undersocialized conceptions of human action and also in neoinstitutionalist
70 - Corporate Social Capital and Liability

arguments that, although acknowledging the role of social networks and private
orders borne out of repeated social contacts, still regard the role of abstract, formal
rules as solely responsible for market order (Platteau 1994a). Greifs (1994) analysis
of the contrasting solutions adopted by Maghribi and Genoese traders in the eleventh
century to trade expansion demonstrates the advantages and limitations of
embedded ness as a mechanism for governing economic life. The collectivist system
adopted by Maghribi traders, where order was enforced through moral sanctions,
worked well in the case of intraeconomy agency relationships, but was incapable of
supporting intereconomy relationships and of allowing for the division of labour
necessary to take advantage of new trade opportunities. By contrast, the Genoese
introduced formal enforcement institutions to support impersonal forms of economic
exchange and promote further division of labour, thus enabling their society to
capture the efficiency gains stemming from the expansion of trade.
Hardin (1993: 510) regards the thick relationships that the embeddedness
argument prescribes as yielding only a part of the knowledge we have of others. But,
of course, one might learn from the experience of others, through reputational
effects and a variety of other indirect means. As Hardin (1996: 31) argues, there are
two modal categories of controls operating in society. There are geographical
associates-the group of friends, family, and associates with whom one is inevitably
bound up in repeated interactions and long-term relationships. And there are the
elaborate large-scale controls associated with institutions, such as the legal system,
relying on formal rules and coercive enforcement. But between these two modal
categories there are a variety of mixed devices such as broad social norms, mixing
elements of both modal categories, that provide important elements of social control
(Hardin 1996).2
In his oft quoted critique of conceptions of human action in economics and
sociology, Granovetter (1985) rejects both notions of undersocialized actors as
behaving atomistically and oversocialized views of human action, where actors
simply follow a script attached to the intersection of the social categories to which
they belong. Instead, Granovetter (1985: 490) revives Polanyi's (1957) notion of
embeddedness and stresses 'the role of concrete personal relations and structures (or
'networks') of such relations in generating trust and avoiding malfeasance.'
Granovetter's argument revolves around the notion that the production of trust in
economic life is mainly accounted for by concrete social relations rather than
institutional arrangements or norms of generalized morality.
He further argues, however, that the existence of strong social relationships may
contribute both to the production of trust and trustworthy behavior as well as,
perversely, mistrust and malfeasance. In short, embedded ness can both contribute to
the resolution and the collective dilemma implied by the Hobbesian position as well
as introduce the possibility of disruption on a larger scale than the one that is
possible in a truly atomized, state-of-nature social situation (Grano vetter 1985: 493).
Zukin and DiMaggio (1990: 15) elaborate on the notion of embedded ness by
defining it as the contingent nature of economic action with respect to cognition,
culture, social structure, and political institutions. Cognitive embedded ness is
defined as the equivalent of bounded rationality, the set of heuristics and biases that
pervade all forms of reasoning. Cultural embedded ness refers to yet more limitations
A Relational Resource Perspective on Social Capital - 71

on economic rationality imposed by the constraints of shared values and

understandings in shaping economic goals and strategies. Structural embedded ness
is for Zukin and DiMaggio (1990: 18) what embedded ness tout court is for
Granovetter-the contextualization of economic exchange in ongoing patterns of
social relationships. Finally, political embeddedness refers to the power struggles in
which different types of actors (e.g., business firms, institutional actors, and the
state) vie for the power to shape the rules that govern economic life.3
Uzzi (l996a, 1997a) elaborated the concept of structural embeddedness
describing it as a specific logic of exchange or coordination in which trust, borne out
of repeated interactions and the prospect of a continuing relationship, pushes the
logic of calculativeness and monitoring to a secondary role. In this mode of
coordination, thick relationships provide more fine-grained, tacit, and holistic
information transfer with a significant and positive impact on problem solving and
conflict-resolution arrangements as well as innovation. 4 Uzzi's conclusions are
largely unsurprising given the plethora of studies over the last twenty years on
buyer-supplier relationships in Europe (Hakansson 1982; Axelsson and Easton
1992; HAkansson and Snehota 1995), the U.S. (Helper 1991), and Japan (Sako 1991;
Smitka 1991 ; Nishiguchi 1994). Furthermore, the literature on flexible specialization
and industrial districts has long since underlined the role of embeddedness in
promoting flexibility, innovation, and adaptability to changing demands as
alternatives to mass production of standardized goods (Lazerson 1988, 1995).
One central conclusion that can be drawn from this brief review is that the form
and extent of embeddedness seen in an economy is strongly influenced by the
discipline and the school of thought the writer subscribes to. While this phenomenon
is a commonplace, as any sociologist of knowledge would acknowledge, what is less
usual is that one line of attack on the issue is the result of the borrowing of a concept
from one of these disciplines for use in another.
Economics has been described as the study of the allocation of resources,
sociology as the study of human aggregations and their behavior. Clearly, these
definitions already imply an overlap. Resources cannot be allocated in an economic
system without the intervention of human agency. Nor can the behavior of social
systems be unaffected by the resources they create, consume, and exchange, directly
or indirectly. However the crucial link here is that of resource. What has happened is
that sociologists have appropriated the concept of resource, or more narrowly
capital, from economics and have used it to capture the notion of the embeddedness
of economic behavior, itself described in terms of resources, in social structures by
way of the notion of social capital. While the basic idea has been around for a long
time, it is only more recently that it has become popular.

Social Capital
The notion of social capital is an intriguing one and raises interesting possibilities as
a counterintuitive metaphor, mapping an economic domain where the notion of
capital is well established and tying accounting and legal conventions to a social
domain, where accounting for value, investment, depreciation, and so on poses a
number of difficulties. One of the functions of metaphors in theory building is an
exploratory one (Easton and Araujo 1993). This is not much removed from an
72 - Corporate Social Capital and Liability

explanatory function, but it emphasizes research directions and possibilities rather

than assumes a comprehensive mapping of the base into the target domain. Before
proceeding to explore the relationship between the base and target domains, we
focus briefly on the notion of social capital that has been used in the social sciences
and the business literature. Portes and Sensenbrenner (1993), while recognizing the
value of Granovetter' s embedded ness argument, make a case for focusing on the
more manageable concept of social capital as developed by Pierre Bourdieu in the
Francophone world and by James Coleman in the Anglo-Saxon literature.
In Bourdieu's hands the notion of social capital is deployed in a rather broad
and undifferentiated sense. Bourdieu (1986: 249) defines social capital as 'the
aggregate of the actual or potential resources that are linked to a possession of a
durable network of more or less institutionalized relationships of mutual
acquaintance and recognition ... a 'credential' which entitles them to credit, in the
various senses of the word.' Further on, Bourdieu characterizes these connections or
obligations as the product of investment strategies aimed at establishing or
reproducing social relationships that can be leveraged either in the short or long run.
Different forms of capital (e.g., cultural or political) derive their forms from the
fields in which they are deployed, but the notion of economic capital remains the
master metaphor for all other forms of capital.
Coleman (1988) starts from the rationalist assumption that each actor has
control over certain resources and that social capital is one type of resource available
to an actor. Coleman (1990: 302) distinguishes between physical, human, and social
capital as three separate forms of capital, all of them having a potential to facilitate
action, or as Coleman puts it 'social capital is productive, making possible the
achievement of certain ends that would not be attainable in its absence.'5 If physical
capital is regarded as wholly tangible, being embodied in observable physical forms,
and human capital is less tangible but still embodied in the skills and knowledge
possessed by an individual, social capital is even less tangible and nonlocalizable for
it is a property of social relationships and social structures.
On closer inspection, Coleman's notion of social capital encompasses both
properties of dyadic relationships and the social structures in which these
relationships are embedded. Social capital takes the following forms : obligations,
expectations and trustworthiness of structures, information channels, norms, and
effective sanctions. So, for example, when in a dyadic relationship one of the parties
does a favor for the other, there is an expectation of reciprocity at some unspecified
future date. Coleman (1990: 306) refers to this situation as one of the parties issuing
a credit slip that can be called on at a future date. But other forms of social capital
are clearly properties of social structure that impinge on and constrain specific,
dyadic relationships. An example of this would be a moral norm prescribing the
foregoing of self-interest in favor of collective welfare.6
Coleman (1990: 317) also argues that the public good aspect of most forms of
social capital is, perhaps, its most distinctive feature vis a vis other forms of capital.
As in Olson' s (1965) collective action dilemma, self-interested individuals free-ride
or seek to reap the benefits of public goods while evading the costs of participation
in their production, and so rational actors will consequently underinvest in the
production of these public goods.7 Many of the benefits flowing from actions that
A Relational Resource Perspective on Social Capital - 73

create social capital can be enjoyed by people other than the ones who have
contributed to those activities in the first place. In short, those who produce them
cannot necessarily appropriate the benefits flowing from investing in the creation of
social capital.
Smart (1993: 393) elaborates on the problems concerning the use of the term
capital and offers a few interesting avenues to preserve the distinctions Bourdieu
makes while clarifying the interrelationship between different forms of capital.
Smart notes that social capital by its very nature is vague and unmeasurable: it lacks
a currency and a space of calculation where debits and credits can be accumulated
and compared. An obligation, for example, becomes concrete only once it is
liquidated, and until then there is no certainty that will ever be reciprocated. For
Smart (1993: 393), if social capital cannot be possessed but can be converted into
other forms of capital, then it is entirely contingent on the reproduction of the social
structures in which it is embedded. If social debts-such as obligations-can be
recovered through enforcement by third parties then, according to Smart, we are
talking about economic and not social capital. 8
Smart relies on a similar logic to argue that social capital is a resource that
resides in dyadic, specific social ties and that more generalized resources such as
honor or reputation that are valued within society or subgroups within it, are best
characterised as symbolic capital. Smart goes on to distinguish between Bourdieu's
notion of capital-in-general and the notion of power. Whereas power is seen as
resting on the authority to command the actions of others, capital in the sense
Bourdieu uses it, is the ability to induce others to act in one's interests through the
leverage of resources available to the agent. Smart (1993: 394) distinguishes
between the different forms of capital invoked by Bourdieu in the following way:
Economic capital involves ownership of objects, but property ownership entails claims
that others may not interfere with your property without your permission, and exclusive
ownership may be used to induce others to act in particular ways (such as hiring or firing
them). Cultural capital is a claim to having the ability to engage in certain types of
practices, and in the strongest forms it accords a monopoly over such practices (for
instance, medical doctors or accountants). Symbolic capital involves claims by the
possessor that he or she be treated in particular ways by classes of others. Social capital
consists of claims to reciprocation and solidarity from particular others. What is
fundamental to social capital. however. is that explicit claims are normally excluded
from the performances within which they are made, so that power over the action of
others is radically distinct from exercises of power utilizing the discourse and apparatus
of command. (emphasis added)
Smart uses the example of gift giving and guanxi in China as an instance in where
explicit recognition of instrumental goals is excluded from the performance, and
incompetent performances results in loss of face and dissipation of the very outcome
that the performance intended to achieve. Gift exchange must, by its very nature, be
devoid of any explicit reference to a calculative logic, lest they be devalued as failed
gift performances or understood as a bribe. In essence, Smart's reinterpretation of
Bourdieu's social capital consists of locating it within concrete, dyadic relationships
and insisting on the absence or accomplished concealment of explicit claims by the
possessors of social capital when attempting to cash in their credits as well as the
74 - Corporate Social Capital and Liability

nonexistence of third-party enforcement to reclaim bad debts. Although we are less

keen on the fine distinction that Smart adopts between social and symbolic capital,
his definition has at least the advantage of relating social capital to contextual
performances that have to be interpreted in the light of experience and further cues.
What we conclude from this discussion is that while there is agreement that
there are different forms of social capital, there is disagreement as to what
conceptualizations of these forms might be most useful. One of the ways we can
explore the notion of social capital is by examining in more detail how the base
domain of the metaphor (economic capital, or more generally economic resources)
map into the object domain (social interaction). In the next section we briefly review
the notion of economic resource before proceeding to examine the properties of
social capital as an economic resource.

The Notion of Economic Resources

The notion of resource has recently been deployed extensively within the strategic
management literature under the guise of the resource-based theory of the firm.
These developments can be traced back to the pioneering work of Penrose (1959)
and pursued in different guises since Wernerfelt (1984).
Resources, in the traditional language of business strategy, are strengths, and
there is little attempt here to suggest how and why they become so and how that
might relate to one another. Despite its growing popularity, the resource-based
theory of the firm has come under attack for its lack of a clear conceptualization of
what constitutes a resource. In the absence of a clear definition, we are left with a
tautology: firms must discover which resources contribute most to their competitive
advantage and develop or acquire more of these (Porter 1991).
This chapter draws on a specific notion of economic resources, inspired by the
industrial networks research program into interorganizational relationships-see
Axelsson and Easton (1992) and Hakansson and Snehota (1995) for recent
examples. We define economic resource as any entity that can be deployed by an
actor that is capable of continuing independent existence, has futurity, and can, or
may, meet an economic need. In this sense, an economic resource is a stock that can
be drawn on either through ownership and control or through indirect access.
Resources stand in contrast to activities. Activities comprise current actions and
have no direct continuity. Resources have a durable character, even if they are the
products of past activities, and acquire their meaning and value when mobilized
within specific activity structures.
We make a further distinction between the generic notion of economic resources
and the terms asset or capital as used in management accounting theory. We reserve
the use of the term asset or capital to describe a resource whose ownership, control,
and market valuation can be determined. Thus assets are simply a subset of
economic resources protected by property rights enforceable by third parties, whose
valuation has some claims to be measurable which, in turn implies the construction
of calculable spaces and technologies able to enforce standardized measures of value
(Miller 1994).
An obvious corollary to our distinction between resources and assets is that we
are not constrained by legal and accounting definitions of the firm. We draw on the
A Relational Resource Perspective on Social Capital - 75

early work of Chamberlain (1968) to emphasize the distinction between the firm as
an entity defined for legal and accounting purposes and the firm as an entity for
coordinating a set of business activities, a key aspect of the industrial networks
approach. Chamberlain employs a broad notion of asset and argues that most of the
assets that constitute the firm's instruments of action are not the ones described on
its balance sheet. In other words, Chamberlain does not assume that a firm's
capability for strategic action resides within the boundaries of what it owns and
controls. Instead the firm's capabilities are seen as embodied in evolving networks
of interdependence both within the firm and with aspects of its environment. A
firm's strategic capability depends on two factors: 1) its capacity to generate
resources from the its current operations and 2) its capacity to mobilize support and
resources from entities and institutions within its environment. More recently,
Barney (1991) argues that a wide variety of the firm's resources may be complex to
the extent that they reside in relationships among people and may be socially
constructed. Among these Barney includes the interpersonal relations among
managers, a firm's culture, and the firm's reputation among suppliers and customers.
And, from an industrial networks perspective, we would add social bonds developed
in the course of economic exchange relationships between two organizations
(Hakansson and Snehota 1995).9
The above discussion has some relevance to the use of the concept of capital-in-
general and, in particular, the distinctions between different forms of capital.
Economic capital, in the cursory way most authors in the social capital literature
seem to understand it, is only a small subset of all the economic resources that are
necessary to conduct economic life. Social capital patently lacks most of the
characteristics that define economic capital. In particular, the absence of accounting
conventions, property rights enforceable through third parties and a system of
exchange devalue the metaphor somewhat. But, on the other hand, it shares many
other features with other economic resources that cannot be qualified as capital, in
the way defined above.
In the following section we attempt to compare the notion of social capital with
that of an economic resource by relying on the framework adapted from Easton and
Araujo (1996). The adaptation consists in changing the focus from resources as an
organization-based phenomenon to resources as a property of intra and
interorganizational, individual-based dyads and social networks.

Social Capital as an Economic Resource

This framework dimensionalizes resources in four separate categories relating to
their existence, evaluation, relationship to actors, and relationship to other activities
and resources. These dimensions are further subdivided into separate categories each
(see Figure 1). Pennings and Lee (this volume) employ a similar list in their
discussion of the benefits of embedded ties for audit firms.
76 - Corporate Social Capital and Liability

Existence Depreciation

Relationships to actors Accessibility
• Evaluability
Relationships to other activities and resources
Figure 1. Dimensions of resources (adapted from Easton and Araujo 1996)

The Existence of Social Capital

Starting with the existence group, the processes involved in the creation,
depreciation, and durability of social capital are difficult to understand, and there is
no obvious causal mechanisms that relate investment to a logic of accumulation of
social capital. The same is true of other intangible resources such as organizational
competencies in producing innovative products, for example.
Coleman (1990: 317-318) argues that much social capital arises or disappears as
a by-product of other activities and without anyone's willing it into or out of being.
Furthermore, some of the conditions that may foster the creation of social capital can
also contribute to its destruction. The closure of social networks, for example, where
judgements of trustworthiness may depend on intersecting patterns of relationships
and transitive judgments (A trusts C, because A trusts Band B trusts C) may lead
both to inflationary and deflationary spirals of trust placement in a community
(Coleman 1990: 318). Stability in social structures may become an important
precondition for the creation and maintenance of social capital, although strangely
Coleman makes an exception for formal organizations where social capital is
assumed to be vested in role occupancy rather than on the identity of the occupant. 10
Finally, Coleman (1990: 321) argues that all factors that help to minimize social
interaction and increase individuals' self-reliance, such as affluence or public aid,
tend to contribute to the depreciation of the stock of social capital, through
bypassing the very activities-e.g., the exchange of obligations and favors-through
which social capital is produced and maintained. Like other economic resources
such as reputations or brand names, social capital is depreciated through the lack of
use, and only through use can its stock be maintained and rejuvenated.
Coleman's account of the creation, destruction, and maintenance of social
capital undervalues instrumental actions directed at creating social capital and elides
some of the dilemmas we alluded to earlier-namely the relationship between social
capital as a property of dyads, small-knit face-to-face groups, or wider social
structures. For example, Burt's (1992, 1997) notion of structural holes describes
how social capital is created and leveraged as a result of brokerage opportunities in a
social network. Uzzi's (1996a, 1997a) account of the New York apparel industry
A Relational Resource Perspective on Social Capital - 77

documents how social capital is created by one-sided investments in business

relationships anticipating future reciprocity.
Pennings and Lee and Lazega (this volume) discuss the relationship of dyad-
specific social capital to wider structures at intra- and interorganizational levels.
Putnam (I993a), in his celebrated work on modern Italy, elaborates on this point in
the case of trust at the societal level. lI For Putnam (1993a: 171-176) the
generalization of trust from closed social networks to the wider society can be traced
back to two sources: norms of generalized reciprocity and networks of civic
engagement. Norms of generalized reciprocity, or diffuse reciprocity to use
Keohane's (1986) expression, are seen as a catalyst for the development of social
capital. As in Gouldner's (1960) account of reciprocity, social capital is built not
solely on the basis of exchanges of obligations that are perfectly balanced but on
mechanisms that induce further debts and credits of obligations and avoid the
expectation that all exchanges will ever return to a zero state in which none of the
parties is indebted to the other.
Keohane (1986) makes an important connection between the norm of specific
reciprocity in dyadic relationships, where the exchange of obligations and duties is
roughly equivalent and performed in a strictly delimited sequence, and the norm of
diffused reciprocity, where equivalence is less precise and sequence of events less
clearly bounded. The repayment of debts and obligations in specific reciprocity
situations may lead actors to take a broader view of their interests and engage in
diffuse reciprocity. Conversely, the decay of diffuse reciprocity may lead actors to
revert to exchanges narrowly bounded by specific reciprocity. As Keohane (1986:
25) puts it: 'specific and diffuse reciprocity are closely interrelated. They can be
located on a continuum, although the relationships between them are as much
dialectical as linear.'12
In short, the creation and maintenance of social capital as an economic resource
is often a by-product of distributed activities and actor orientations that may be
reinforced or hindered by the institutional context in which these activities are
embedded. Whereas the role of institutions and formal rules has been underlined in
helping to supersede collective action dilemmas, by providing a stable background
of enforceable norms and duties and thus lowering the risks involved in engaging in
cooperative relationships, less has been made of their role in destroying social
Recently, Pi Ides (1996) has argued that the intervention of the state through the
provision and enforcement of legislation can destroy social capital through an
insufficient appreciation of the role of social capital that underwrites the successful
enforcement of laws. Pi Ides (1996) advances an argument on the role of the state in
contributing to the destruction of social capital that can be extended to all attempts at
formalizing rules. There are three ways in which social capital can be destroyed
through state intervention: 1) indirectly, by attacking the structural foundations on
which social capital can be created and maintained-e.g., by prescribing spaces in
which social interaction mayor may not take place, 2) directly, by attacking norms
of reciprocity through efforts to rationalise conventions, and 3) indirectly, by
attempting to incorporate social norms into legislation without taking due account of
the role of remedial flexibility in the enforcement of social norms.
78 - Corporate Social Capital and Liability

The Evaluation of Social Capital

In this section, we attempt to relate the notion of social capital to four dimensions
associated with the evaluation of economic resources: valuation, evaluability,
scarcity, and value (positive/negative) (Easton and Araujo 1996). All economic
resources are generally held to possess a value, even if that value cannot easily be
expressed in the metric of money. More often than not, it is the intangible and more
difficult to evaluate resources that constitute the most valuable resources a firm
possesses (Itami 1987). Furthermore. the value of a resource is intrinsically tied to
the context in which it is used. or to put it differently, to describe the content and
value of a resource is also to describe its context of use.
Social capital is usually hailed as an important resource for promoting both
effective government and prosperous economies. In particular, trust has been singled
out as a key resource in promoting economic growth and competitiveness and as
having a positive impact in intraorganizational relationships (Breton and Wintrobe
1982; Miller 1992) as well as in interorganizational contexts-from buyer-supplier
relationships (Barney and Hansen 1994; Nooteboom 1996) to industrial districts
(Lorenz 1993) or the relationship between capital market institutions (Neu 1991).
In general. as Gambetta (1988a: 221) observes: 'societies that rely heavily on the
use of force are likely to be less efficient, more costly, more unpleasant than those
where trust is maintained by other means. In the former. resources tend to be
diverted away from economic undertakings and spent in coercion. surveillance, and
information gathering and less incentive is found to engage in cooperative activities.'
Social capital can thus be regarded as a valuable. infrastructural resource to
economic life but one that cannot be evaluated easily. Sabel (1995) contrasts two
opposite views on this issue, one inspired by Hayek and the other by Durkheim.
Whereas the Hayekian view stresses the distributed and partial nature of knowledge
in society. and the way norms evolve as a spontaneous. unintended outcome
(Bianchi 1994), the Durkheimian perspective stresses the role of institutions in
curbing the freedom of rational. selfish agents to act as they see fit. From a
Durkheimian perspective, the economy can perform well only if is embedded in a
well-integrated society and that society is willing and able to impose normative
constraints on the pursuit of self-interest (Streeck 1997: 199).
The role of social capital as an economic resource is clearly less important under
a Hayekian perspective. For Hayek (1978) the market with the price mechanism,
sets of property rights. and enforceable rules is the best example of an evolved
institution. Social institutions are instrumental, problem-solving mechanisms that
evolve spontaneously under the pressure to find ways to coordinate the activities of
individuals facing a complex world equipped with limited knowledge. Under a
Durkheimian perspective the value of social <;apital to economic life is much clearer.
For example. social constraints on the pursuit of self-interest in the form of moral
norms or formal laws can reinforce trust and thus increase the efficiency of
economic transactions by reassuring potentially suspicious parties of continued
adherence to norms of reciprocity and normal business practice. As Streeck (1997:
202) puts it: 'Credible information that the other side has noneconomic in addition to
A Relational Resource Perspective on Social Capital - 79

economic reasons not to defect accelerates and consolidates the growth of trustful
relations. '
But if a Durkheimian perspective exalts the value of social capital in promoting
economic efficiency, it also warns about the dangers of attempting to design social
institutions with the sole purpose of economic efficiency in mind. Streeck (1997:
217) warns that 'the kind of social embedding good economic performance requires
can be built only for reasons other than good economic performance, enabling it to
support rational-economic action by containing it.'
These considerations lead us to reflect on two other dimensions of the
evaluation of social capital: scarcity and value (positive/negative). The issue of
scarcity of social capital is important and related to the issue of valuation. As a
resource, most forms of social capital are not scarce in the sense that it is a stock
depleted through use. On the contrary, social capital depreciates through lack of use
and appreciates through extensive use and imbalanced exchanges. For Putnam
(1993a: 178-179) both reciprocity/trust and dependence/exploitation can become
stable equilibria in particular communities depending on initial conditions-namely,
stocks of social capital and paths of evolution. As in North (1990), path dependence
is deemed responsible for the self-reinforcing nature of a particular equilibrium.
Distrust, for example, once set in has the capacity to become self-fulfilling and lock
a particular community into a low-level equilibrium of authoritarian government,
patron-client relationships or putting them at the mercy of criminal organizations
(Gambetta 1988a).
Sabel and Zeitlin (1997) effectively reject the notion of path-dependence and
claim instead that actors are able through reflexivity and choice to transform the
conditions under which they live and to be less discriminating between the two
equilibria Putnam describes. Economies often contain what Sabel and Zeitlin (1997)
describe as moral borderlands where moral rules are only partly observed or
significantly relaxed. In short, the cycles of trust and mistrust that Putnam describes
as being the resultant of initial endowments of social capital and path-dependent
evolution may be less sharply defined and delimited in historical epochs and equally
dependent on stocks of social capital.
The issue of the scarcity of social capital as an initial condition to foster
economic development has been hotly debated for some time. Recently, Evans
(1996) reflects on the role of social capital in promoting economic development and
suggests that endowments of social capital cannot be seen as the major obstacle to
foster development projects in the Third World. In most Third World communities
prior endowments of social capital, in the form of social networks, norms of
solidarity and trust at the micro-level are seen as adequate, but the difficulty lies in
scaling up such capital to generate solidarity and diffuse norms at a politically and
economically efficacious level.
Lastly, we must consider the valence of social capital as a resource. Throughout
most of this chapter we have implied that social capital is an asset to political and
economic life, rather than a liability. But, as Portes and Sensenbrenner (1993)
remind us, the same social mechanisms that account for useful resources
appropriable at the individual, dyadic, or community level may also set important
limits to action and economic development.
80 - Corporate Social Capital and Liability

Gargiulo and Benassi (this volume) illustrate how social capital can, in the face
of changing circumstances, turn into social liability constraining opportunities for
organizational change. Recent research on entrepreneurship and immigrant
communities in the U.S. documents both the negative and positive aspects of social
capital and embeddedness. Portes and Sensenbrenner (1993) argue that the greater
the social capital produced by solidarity and community ties, the more likely that
particularistic demands will be imposed on successful entrepreneurs, thus limiting
the possibilities of individual expression and the expansion of opportunities outside
the community boundaries. Waldinger (1995), in his study of immigrant
communities in the New York construction industry, argues that in this instance, the
embedded ness of economic behavior in ongoing social relations among a myriad of
social actors impedes access to outsiders. Embeddedness contributes an exclusion
effect from social networks, breeding a preference for established players with track
records. However, the overlapping of economic and ethnic ties has a further
undesirable effect, since outsiders also fall outside those networks that define the
industrial community.
Lastly, Gambetta (l988b) reminds us that there are situations where the public
interest might be better served if trust and the social capital built through exchange
of obligations and favors is collapsed rather than reinforced. Baker and Faulkner's
(1993) study of social networks involved in collusive practices in the American
heavy electrical equipment industry demonstrates the point.

The Relationship of Social Capital to Actors

The third group of dimensions to classify economic resources is labeled relationship
to actors. In this case, actors might be individuals, groups, business firms, or other
collective actors such as business interest associations. Three dimensions
characterize the relationship of a resource to actors: controllability or the extent to
which actors can appropriate and control a resource, accessibility or conversely the
existence of barriers to use a resource and tradeability, or the ability of actors to
convert resources through exchange processes.
Starting with controllability, we recall that social capital can be regarded as a
public good in the way Coleman (1990) defines it. Public goods are usually
characterized by jointness of supply and nonexcludability of benefits (Olson 1965).
But clearly this is not always the case with social capital. For example, Portes and
Sensenbrenner (1993) describe an informal financial system operating in the Cuban
immigrant community in Miami in the 1960s, whereby Cuban bankers were
prepared to provide 'character loans' to newly arrived refugees, based on the
personal reputation of the recipient in Cuba. This system worked well until a new
wave of immigrants, no longer known to the Cuban banking community in Miami,
arrived in 1973 after which character loans were discontinued.
Hechter (1987) discriminates between public and collective goods, on the basis
of the degree to which benefits are excludable. Most forms of social capital such as
trust cannot be characterized as public goods, in the sense that its benefits are
distributed homogeneously within a population. One might describe generalized
trusting within a community as a form of social capital but trust itself, is a specific
property of a dyadic relationship: A trusts B to do X (Hardin 1993). My ability to
A Relational Resource Perspective on Social Capital - 81

deepen trust within a dyadic relationship--A trusts B in a range of situations, rather

than in a specific instance X--or to extend it to other relationships, is a learned
response. As Hardin (1993: 508) puts it: 'Trust has to be learned like any other
generalization. Insofar as my trust is a generalization in the face of new persons, this
merely means that the capacity to trust, the optimistic Bayesian estimate of
trustworthiness, is learned perhaps from long experience.'
But my propensity to trust is also dependent on factors outside my own
experience-namely, my family upbringing and the community in which I have
been brought up. The amount of generalised trusting I have encountered will
undoubtedly affect my estimates of the benefits of trusting and others'
trustworthiness. Whereas Hardin is right to contend that trust is not a form of human
capital, it is also true that the generalized trusting that is so critical to foster trust in
dyadic relationships could itself be seen as a form of social capital. My ability to
trust someone is partly dependent on my experience with similar people and
possibly also the experience of people I trust with that third party. Thus, Uzzi's
(1996a) study of the New York apparel economy found that, as in third-party
referral networks, the stock of social capital existent in a business relationship was
often used in a new, connected relationship by applying expectations and norms
from an existing relationship.
In short, social capital cannot be owned, controlled, or appropriated by an actor.
But that doesn't mean to say that social capital is a simply a resource uniformly
distributed and awash in the social structure of a particular community. It is at once
located in concrete, identifiable dyads and as a generalized resource that actors can
selectively draw for their social interactions.
Accessibility is an associated dimension to controllability. It denotes the ability
of actors to access a resource even when they are unable to secure control over it.
Information, for example, is one type of economic resource that presents specific
problems in terms of control and access. Access to social capital requires active
participation in the social networks and engagement in specific relationships.
Krackhardt (1992), Lazega and Lebeaux (1995), and Lazega (this volume) provide
good examples of how in formal organizations social capital is differentially
distributed among different social networks and how individuals might leverage
their social capital differently depending on specific circumstances. Krackhardt
(1992) is concerned with the strength of strong ties or what he describes as phi/os
relationships that have are based on long-standing friendships and affection. In a
casestudy, exploring an unsuccessful attempt to unionize an electronics firm in
California, Krackhardt attributes the failure to unionize the plant to the power of
individuals, based on their friendship networks, in opposing the move. Whereas the
advice networks reflected routine and workflow knowledge and influence, the phi/os
network became more important in a crisis context, when advice was sought from
those one trusted (friends).
Lazega and Lebeaux's (1995) study of a law firm uncovered partly overlapping
social networks based on different factors (friendship, advice, coworker) and
showed that actors leverage their social capital rather selectively, depending on the
particulars of a situation. Lazega and Lebeaux (1995) research focused on how a
focal actor used an intermediary to influence a target actor in a range of situations.
82 - Corporate Social Capital and Liability

The results of the research demonstrated how focal actors drew selectively on their
social capital by using different social networks for different purposes and were
concerned about limiting the negative effects of leveraging that social capital.
However, this research also showed how the focal actors were often forcing
intermediaries to draw on their social capital in less discriminating and selective
Smart' s (1993) study of the use of guanxi in China is at once a good description
of how social capital is produced and reproduced within close-knit networks of
relationships, and also a good example of how it takes time and investment in
relationships to build up social capital in a particular network. As knowledge of the
rules of guanxi in China are difficult for an outsider to comprehend, many Hong
Kong investors use social connections as intermediaries to contact local brokers with
reputations for being able to solve problems (Smart 1993: 403).
Lastly, social capital is not a tradeable resource in the sense that it can be easily
exchanged for other resources. Although it is tempting to use the language of debits
and credits in the exchange of obligations, favors or trust, these exchanges are
characterized by the absence or accomplished concealment of explicit claims by the
possessors of social capital when attempting to cash in their credits as well as the
nonexistence of third-party enforcement to reclaim bad debts. The logic of social
capital is thus the logic of making calculativeness, self-interest, profit, accumulation,
and so on taboo (Bourdieu 1994).
So one party might build social capital in a particular relationship by a gift
presentation or trusting gesture, but there is no guarantee that the debt will ever be
repaid or, if repaid, the exchange will be equitable. But, to the extent that one invests
in gift giving and disinterested gestures, this will have positive outcomes via
building up status and reputation in a particular community or group. If
reciprocation is not direct, benefits can still be accrued by others' recognition of
one's generous disposition. But as Smart (1993: 396) remarks, we have had
experiences of unreciprocated gifts and invitations or noted the absence of
acknowledgment of social debts.
In economic life, the notion of gift presentation and exchange of obligations is
more nuanced and less clear cut than in the examples given above. Often, the
presentation of gifts in the social exchanges that accompany economic exchange are
part of specific, particularistic relationships between roleoccupants (e.g., salesman-
purchasing agent). The social capital built in these social relationships may be
recognized as institutional as well as personal social capital. However, this form of
social capital may only be recognised when individuals leave roles or take with them
a range of relationships (e.g. , customers) to a different organization. 13
In practice the picture may be more complicated since organizations are related
not simply through economic exchange but also through business-interest
associations, joint suppliers or customers, and their members' activities in
professional associations, industry forum, and technical standards committees
(Hakansson and Snehota 1995). Studies of information transfer and social networks
demonstrate how social capital is created and consumed in a variety of industry
contexts. Saxenian' s (1994) study of the evolution of the computer and
semiconductor industries in Silicon Valley and Route 128 ascribes the comparative
A Relational Resource Perspective on Social Capital - 83

success of Silicon VaHey to informal and densely patterned communication

networks between communities of practice, based on professional aHegiances and
reciprocity of information exchanged, that provide an efficient and rapid source of
In short, social capital as a resource in economic life is characterized by a
continuum ranging from areas where the absence of calculativeness and self-interest
is paramount to areas where the tradability of favors, obligations, and reciprocity in
information exchange is more readily accepted. In all cases though, the borderlands
between specific reciprocity and equity in the exchange, and performances where
instrumental ends and calculativeness are either absent or successfully concealed,
are contingent performances subject to continuous negotiations within concrete
relationships, rather than fixed and socially prescribed rules.

Relationship to Other Resources and Activities

The first dimension in this set is that of integrity. It measures the extent to which
resources are simple or compound; that already comprises a mix of resources or else
representing some fundamental unit of resource that cannot be subdivided. Social
capital provides an excellent example of a compound resource that is impossible to
disentangle from the context in which it is produced and consumed. As Smart (1993 :
393) argues social capital is social in the sense that it cannot be possessed and its
existence is contingent on the reproduction of the concrete social relationships
within which it resides. At the same time, social capital in the form of obligations or
connections constitutes the very raw material through which social relationships
within groups or communities are reproduced and maintained.
The compound nature of social capital is intrinsically linked to the nature of its
creation and maintenance. The creation of some forms of social capital such as trust
can be regarded as fortuitous by-products of experiences over which the individual
may have little control or even did not undertake (Hardin 1993: 525). An individual
may grow up trusting and reproducing trust in his or her social relationships, as a
result of growing up in a supportive environment and learning through repeated
experience that trusting pays off handsomely. Similarly, Putnam's (1993a) argument
is that social capital is created and reproduced in communities where associations
proliferate, membership overlaps, and multiplex relationships are formed. 14
A dimension widely acknowledged to be important in understanding resources
is that of versatility. Resources may be versatile or specific in nature. In this context
versatility refers to the ways in which a resource may be combined with other
resources or activities. In practice, versatility is unequally distributed among
resources and, in any case, is not regarded as an intrinsic property of resources but a
function of their use-namely, in the context of exchange activities.
Social capital provides an excellent example of a resource that may be both
specific and versatile, depending on the context of use and its form. For example,
the existence of an obligation in a dyadic relationship is narrowly confined to the
context of that relationship and the ways in which it can be repaid forms part of the
atmosphere and tacit understandings prevailing in that relationship. Similarly,
placing trust in somebody to do something is often a highly particularistic judgment
that is embedded within the joint history and the atmosphere of a relationship.
84 - Corporate Social Capital and Liability

However, belonging to a particular community, group, or clique and drawing on

the social capital accorded to members may be a highly valuable and versatile
resource. Examples of uses to which social capital acquired in this manner include
getting a job, landing a contract, or getting a client's trust (Waldinger 1995).15 In
particular, social capital built across overlapping relationships between the same
individuals (e.g., as buyer and seller, friends, professional colleagues) may provide a
highly versatile resource that can be leveraged across a range of contexts. Coleman
(1990) argues that social capital is often the by-product of membership in voluntary
associations and the social capital thus formed can be harnessed for other collective
action ends. Taylor and Singleton (1993) claim that the endogenous solution of
collective action dilemmas is greatly facilitated by the existence of social capital
lodged in multiplex relationships.
Similarly, clear-cut typifications of an individual as member of a group or
organization, may shut off the very same opportunities it affords generously to
insiders. Thus Waldinger' s (1995) study of the New York construction industry and
its ethnic niches showed that black contractors often had no choice but to restrict
themselves to public works contracts since all the other contracts were stitched up in
social circles and activities (e.g., golfing, boating) from which they were excluded.
The third dimension in this set is that of complementarity. Some resources are
easier to combine with other resources, and with activities, than others are.
Resources may be competitive as well as complementary even within the same firm.
Complementarity is a measure of the processes of combination. It is concerned with
the ease with which combinations with a specific resource may, in general, be
carried out. The complementarity of human capital and physical capital, for
example, is well documented in the resource-based theory of the firm, starting with
the pioneering work of Penrose (1959).
However, the complementarity of social capital with both physical capital and
human capital has received less attention. Coleman (1988) uses family upbringing
and relationships within the community in which the family is located as examples
of how social capital influences the creation of human capital. Waldinger (1995:
560) argues that Coleman' s examples only illustrate the indirect effects of social
capital on the development of human capital and that a more compelling case could
be made of how social capital is constitutive of all human capital. Learning through
socialization and participation in a community of practice such as the building
workers that Waldinger studied, are good examples of how the formation of human
capital is enmeshed with social capital. Ostrom (1994) provides a rare example of a
study where the interaction of social and physical capital is explicitly considered in
collective action problems.
Burt (1992) provides the most forceful argument for the belief that social capital
is both complementary to other forms of capital as well as being capable of being
leveraged for the benefit of an individual or organization. For Burt (1992) in firms
that provide services, for example, there are people valued for their competencies in
providing expertise and ability to deliver a quality service and those (the rainmakers)
who are valued for their ability to deliver and handle clients. The former represents
financial and human capital, whereas the latter represent social capital. In short,
different forms of capital are highly complementary and divisible on the basis of a
A Relational Resource Perspective on Social Capital - 85

sharp division of labor and endowment of competencies even if, as Burt states,
'social capital is the final arbiter of competitive success.' In a world where players
can easily be matched on their endowments of human and financial capital, social
capital narrows down the pool of individuals who can compete successfully for new
opportunities. However, Burt (1992, 1997) privileges the notion of structural holes
or a relationship of nonredundancy between two contacts, and thus, in his view,
social capital is effectively reduced to the instrumental use of social relationships for
the pursuit of entrepreneurial opportunities to transform network structures to one's
own advantage.
The final dimension in this set is understandability. It is a measure of how easy
it is to comprehend the nature of a resource or even to recognize a resource at all.
Some resources are easily recognized as such and are well understood. Most forms
of physical capital come into this category. They are tangible, controllable, and
easily evaluated. Other resources, human capital included, do not share these
properties. They are intangible and cannot easily be valued. As mentioned
previously, organizational competencies such as routines fall into this category.
Exchange relationships and managerial skills are examples of this kind of resource.
If a resource can readily be understood, then it is more likely that the ways in
which it can be created and maintained as well as combined with other resources
will be well understood and hence exploited successfully. Virtually all forms of
social capital fare badly on this score. Coleman (1990: 312-313) argues that there
are few cases where social capital is well understood as a resource and can be
created as a direct result of investment of actors who have the aim of receiving a
return from their investment. The creation of formal organizations and the associated
authority structures with well defined obligations and expectations associated with
role occupancy is presented as a case of investment in social capital, in the same
way investment in human capital is associated with the appointment of individuals
to specific roles. But Coleman's formulation seems to conflate notions of power,
vested in authority structures, and social capital as the ability to induce others to act
in one's interests through the leverage of resources available to the agent. Whereas
power can ultimately be seen as resting on the authority to command the actions of
others and enforce sanctions for noncompliance, social capital tends to rely on
implicit rather than explicit claims to reciprocation and cannot resort to formal
sanctions on recalcitrant targets.
In hierarchies, as Miller (1992) cogently argues, managers do not spend their
time writing and enforcing contracts defining obligations and expectations
associated with specific roles anymore than employees spend their time maximizing
self-interest within the constraints imposed by those contracts. Norms of co-
operation and trusting relationships are as critical to the functioning of an
organization as a supplement, but not necessarily as a substitute for incentive
systems based on formal contracts. In short, notions of power and social capital are
as crucial to the functioning of hierarchies as private and public orders are vital to
the functioning of markets.
In summary, the character of most forms of social capital makes it a poorly
understood resource. Its creation is often dependent on by-products of other
activities, mostly escaping the direct control of actors, its leverage is dependent on
86 - Corporate Social Capital and Liability

contextual performances, it cannot be easily transferred from context to context and

its linkages with other resources cannot easily be disentangled. Moreover, the
concealed or implicit nature of leveraging social capital, often attached to the history
and atmosphere of particularistic relationships, makes it even less understandable as
an economic resource.

The increasing attractiveness and popularity of the notion of social capital can be
understood by reference to the appealing logic of the capital-in-general metaphor
and for the possibilities it brings to breach some of the artificial divides between the
economy and society. This appeal rests as we have argued, in the often insufficiently
understood mechanisms of investment, accumulation, and benefits accruing from the
use of capital. In this sense, as others have noted (see, e.g., Tarrow 1996) social
capital has an appealing normative component. Creating, maintaining, and growing
social capital, so we are told, is the key to healthy societies, high-performing
organizations, and prosperous economies. In this sense, social capital can be seen as
important resource since its existence obviates the need to allocate other resources to
the formalization of rules, coercion, surveillance, and information gathering to
supplement private norms.
On the other hand, social capital provides a vehicle to breach the divide between
the social and economic worlds and its attributes make it an all-encompassing
notion, connecting dyads to wider social structures and back again. Thus social
capital-although conceived as a property of wider social structures in the form of
organizational structures, occupational communities, trade or civic associations, and
so on or based on categories such as ethnic groups-is manifested through concrete
outcomes in ordinary social practices and relationships. In this sense, social capital
can provide an antidote to more generic notions of embeddedness, understood
simply as a static backcloth of social structures encapsulating economic exchange
but themselves immune from reciprocal influences.
The argument in this chapter has been that neither the normative nor the
structural attractiveness of social capital as a notion can mask the dangers of
engaging in too cursory an application of the capital-in-general metaphor to the
study of the interaction of social structures and economic action. Social capital
understood as a resource that can impinge on economic action is perhaps not unique
in some of its attributes, but its heterogeneous manifestations, intangibility if not
purposeful invisibility, its compound nature, and poorly understood logics of
accumulation and use make it a counterintuitive metaphor.
In this chapter we have attempted to be more specific about the nature of
economic resources and subject the social capital metaphor to a more systematic
appraisal of its attributes as a resource impinging on economic action. In so doing,
we effectively moved the notion of social capital away from the notion of capital-in-
general and instead stressed its intangible, dynamic, and relational nature. This
allows us to move beyond notions of social capital as an abstract property of social
structures to explore its concrete outcomes and contextual performances and link
these to the logic of production and reproduction of those social structures where
social capital is deemed to reside. This last point is perhaps worth stressing anew,
A Relational Resource Perspective on Social Capital - 87

even if a full discussion of this point is outside the scope of this chapter social
capital consists of claims to reciprocation and solidarity contained in concrete
episodes and contextual performances. What governs these episodes, however, are
that explicit, calculative claims are normally excluded from the performances within
which they are made and thus the leverage of social capital has to be read and
interpreted anew in every single episode.
Finally, we wish to issue a cautionary note to ourselves as much as to others,
regarding the need to resist the facile prescriptions that inevitably spillover from too
many of the romantic accounts of the role of social capital in socioeconomic life.
Even if we begin to understand a little better the notion of social capital, the
operation of policy instruments directed at building, maintaining, or destroying
social capital remain as slippery, open-ended and uncertain as any other policy
instrument. 16 Hopefully, this volume will contribute to advancing our understanding
of these issues.

I. Recent work in the sociolegal field has recently taken an interest in the interaction between the law
and infonnal social nonns in law making and enforcement (see, e.g., Pildes 1996; Posner 1996).
2. On this topic see also the belated discovery of the existence of a variety of blends of trust and
contract in buyer-supplier relationships in some sociolegal and economics literature (e.g., Burchell and
Wilkinson 1997).
3. See F1igstein (1996) for a comprehensive treatment of the role of political embeddedness in the
evolution of markets.
4. See also Podolny (1994) and Uzzi and Gillespie (this volume) on this topic.
5. Ostrom (1994: 527-528) defines social capital simply as the arrangement of human resources to
improve flows of future income for at least some of the individuals involved in the production of that
capital. Similarly, Portes and Sensenbrenner (1993: 1323) define social capital as the relevant expec-
tations concerning economic action within a collectivity, affecting the goal and goal-seeking behaviour of
its members.
6. For a parallel and insightful distinction between social capital as lodged in personalised, dyadic
relationships and depersonalized, institutional relationships, see Pennings and Lee (this volume). See also
Knoke (this volume).
7. For a similar argument on social capital as a public good see Putnam (1993).
8. On the role of third parties in exchange see Nooteboom (this volume).
9. See also Pennings and Lee (this volume) on the importance of relationships that span organizational
boundaries for corporate social capital.
10. See Pennings and Lee (this volume). See also Burt (1997) for an argument on how social capital is
becoming increasingly important in organizational life and managerial perfonnance.
II. For critical reviews of Putnam's perspective on the impact of social capital on political and
economic life see Levi (1996), Tarrow (1996), and Kenworthy (1997).
12. See Lazega (this volume) on the notion of the finn as a multiplex, generalized exchange system and
on the role of reciprocity in this exchange system. Scharpf (1993: 153-154) makes a related point
concerning generalized trust whose existence 'pressuposes a generalized willingness to cooperate even in
constellations where cooperation is not advantageous and is easily destroyed by the pursuit of self-interest
at the partner's expense. But where it exists, generalized trust is enonnously advantageous. It will enable
rational actors to enter into vulnerable positions, and to engage in high-risk (and potentially high gain)
mixed motive transactions under conditions of incomplete infonnation.'
13. See Pennings and Lee (this volume) for a more comprehensive discussion of these issues.
14. The importance of multiplex relationships for the emergence and maintenance of social capital is
highlighted in several chapters in this volume-see, in particular Pennings and Lee, Knoke, and Lazega.
15. See also Flap and Boxman (this volume).
16. See also Leenders and Gabbay (this volume).
Social Capital by Design:
Structures, Strategies,

and Institutional Context

Wayne E. Baker
David Obstfeld

We examine social entrepreneurship from a structural perspective, distinguishing
between two structures of social capital and their associated entrepreneurial
strategies: structural holes and the 'disunion' strategy versus social cohesiveness and
the 'union' strategy. These two strategies represent alternative ways social
entrepreneurs access and mobilize the resources inherent in the structure of a social
network. The disunion strategist exploits structural holes between alters by keeping
them apart; the union strategist creates value by bringing together disconnected
alters. The frequency, legitimacy, and success of each strategy depends on the
'design' of the institutional context in which social entrepreneurs operate. Disunion
strategies tend to occur in organizations and markets characterized by sparse,
disconnected, and differentiated networks, coupled with competitive rules of
exchange, opportunism, and an individualist orientation; union strategies tend to
occur in organizations and markets characterized by dense, connected, and
undifferentiated networks, coupled with cooperative rules of exchange, norms of
reciprocity, and a collectivist orientation. We illustrate the distribution of triadic
strategies in a specific institutional context by taking a triads census of alliances in
the global automobile industry and testing the structural hypothesis about the use of
disunion and union strategies.

Ever since Schumpeter (1934: 156) identified entrepreneurship as a 'vehicle of
continual reorganization of the economic system,' entrepreneurship has been
recognized as playing a key role in catalyzing change, promoting innovation, and
Social Capital by Design - 89

enhancing productivity in the economy at large. For example, the continuous

process of 'breaking away' from existing firms to create new ones is a prime engine
of the growth and economic development of cities (Jacobs 1965, 1970). The role of
entrepreneurs as intermediaries between corporate actors (firms, associations,
governmental bodies) is well known (e.g., Coleman 1990: 180-188). However,
entrepreneurship is also recognized as a creative activity that occurs inside
organizations, an activity that is critical to their survival and health (e.g., Burgelman
1983a; Kanter 1983a). Entrepreneurial action within organizations is called intra-
preneurship (Pinchot 1985), corporate venturing, or corporate entrepreneurship
(Burgelman 1983a, 1983b).
Structural sociologists shifted the study of entrepreneurial behavior away from a
focus on traditional entrepreneurial activities, such as using economic capital to start
new ventures, to the analysis of the strategic use of 'social capital' both inside and
between organizations (e.g., Burt 1992; Bourdieu and Wacquant 1992; Coleman
1988, 1990). Our chapter follows in this structural tradition. We offer a new
theoretical distinction between two structures of social capital and their associated
entrepreneurial strategies-structural holes and the 'disunion' strategy versus social
cohesiveness and the 'union' strategy. These two strategies represent alternative
ways social entrepreneurs access and mobilize the resources residing in a social
We use the phrase 'social capital by design' in two senses that together capture
the essence of our argument. First, social capital is related to the 'design' of an
institutional context. A context characterized by sparse, differentiated, and
disconnected networks, for example, yields social capital in the form of structural
holes. A context characterized by dense, integrated, and connected networks, in
contrast, yields social capital in the form of social cohesiveness. Second, social
capital can be created by intention, that is, by the strategic moves of an individual
entrepreneur or the deliberate manipulation of organizational and interorganizational
structures. Change agents, for example, can change the structure of social capital
(and the entrepreneurial strategies used to access it) by altering the design of the
institutional context.


The Structural Approach to Entrepreneurship
Most attempts to understand and promote entrepreneurship have examined
entrepreneurs and their behaviors at the individual level, striving to define the key
traits and characteristics of successful versus unsuccessful entrepreneurs (Gartner
1989; Low and MacMillan 1988). For example, McClelland (1967) argued that the
'need for achievement' is a key psychological characteristic of the successful
entrepreneur, but empirical research has not supported a link between the need for
achievement and, say, the decision to start a new business (Sexton and Bowman
1985). Locus of control and propensity for taking risks have been proposed as
possible distinguishing characteristics, but research has not provided much empirical
support of these hypotheses (e.g., Brockhaus 1982; Sexton and Bowman 1985;
Gasse 1982). Tolerance of ambiguity does appear to distinguish entrepreneurs from
managers (e.g., Schere 1982; Sexton and Bowman 1985). In general, the concerted
90 - Corporate Social Capital and Liability

attempt over the past decades to build a personality profile of successful

entrepreneurs has not yielded insights into the unique personalities of entrepreneurs.
Based on their review, Low and MacMillan (1988: 148) conclude that, '... at
a... fundamental level, it can be argued that the wide variations among entrepreneurs
make any attempt to develop a standard psychological profile futile. One is struck by
the appropriateness of Gartner's (1985) observation that 'there is as much difference
among entrepreneurs as between entrepreneurs and non-entrepreneurs."
The failure of personality research to identify key characteristics suggests that
the individual level of analysis may be inappropriate for understanding
entrepreneurial action (Gartner 1989). These efforts suffer from the neglect of social
structure and the complex relationships between the individual, corporate actor, and
environment (Martinelli 1994). Others have proposed a behavioral focus for
entrepreneurial research, emphasizing the processes associated with entrepreneurial
behaviors (e.g., Gartner 1989, 1990). The structural approach, which we use here, is
consistent with such a behavioral focus (e.g., Aldrich and Zimmer 1986; Burt 1992;
Krackhardt 1995; see also, Freeman, this volume, for a similar perspective).
The entry of structuralists into the study of entrepreneurial behavior shifted the
focus away from traditional entrepreneurial behaviors, such as starting new
businesses or founding firms, to the examination of the strategic creation and use of
'social capital' both inside organizations and among organizations (e.g., Burt 1992;
Bourdieu and Wacquant 1992; Coleman 1988, 1990). 'The central proposition of
social capital theory,' summarize Nahapiet and Ghoshal (1998: 243), 'is that
networks of relationships constitute a valuable resource for the conduct of social
affairs .. .. ' We use the term 'social entrepreneur' to describe individuals or corporate
actors who access and mobilize the social capital inherent in organizational
networks, as opposed to the 'traditional entrepreneur' who uses economic capital to
create new businesses and firms (of course, traditional entrepreneurial activities
often involve the use of both social and economic capital). The social processes we
discuss apply to activities both inside organizations and between organizations (what
we call 'institutional contexts' below). We focus on 'corporate social capital'-social
capital that resides inside, between, or among organizations-in contrast to other
locations of social capital, such as families and communities (see, e.g., Coleman
1988; Putnam 1995a).
The definition of social capital, along with the role and activities of the social
entrepreneur, have been the subjects of considerable debate. l Clarifying the
dimensions of social capital is a high research priority (Putnam 1995a). Nahapiet
and Ghoshal (1998), for example, make useful distinctions between three
dimensions of social capital: the 'structural dimension' (the configuration of social
networks), the 'cognitive dimension' (shared systems of meaning, narratives,
language), and the 'relational dimension' (norms, trust, obligations). Our structural
view of social entrepreneurship emphasizes the first dimension, exploring the basic
structures of social capital (structural holes versus social cohesiveness), their
corresponding entrepreneurial strategies (disunion versus union), and the
relationship between institutional context and strategy. Our distinctions, as we
elaborate below, help to clarify the debate and resolve some of the confusion about
the structural approach to social capital and social entrepreneurship. Though we
Social Capital by Design - 91

emphasize the structural dimension of social capital, we acknowledge the

importance of the cognitive and relational dimensions.

Two Structures of Social Capital

The triad of social actors, composed of 'ego' and two 'alters,' is the basic unit of
analysis in social entrepreneurship. Structures larger than the triad are possible, of
course, and often occur; however, they are based on the triad as the fundamental
building block. For example, 'communities of trust' are generalizations of the three-
actor structure (Coleman 1990: 188-189). Social actors can be persons or corporate
actors: people acting as individuals, people acting as agents or representatives,
organizational subunits (such as teams or departments), organizations, and even
governmental bodies, states, and nations. (For example, the triad is a basic unit used
in analysis of geopolitical relations.) Given our focus on corporate social capital,
however, we are interested primarily in two types of social actors: individuals who
are members of organizations and organizations themselves. Our concepts apply to
both types.
The structural basis of entrepreneurial action was suggested by Simmel (1950:
154-162), who stressed the importance of the 'third element' in group dynamics.
Simmel argued that the introduction of a third party fundamentally alters the social
dynamics of dyadic ties (see also Nooteboom, this volume). Of particular interest is
the triad type Simmel (1950) called tertius gaudens-'the third who enjoys' benefits
by his or her position between two disconnected parties. These two parties, because
of their unfamiliarity with each other, can be manipulated to the third party's
benefit. Simmel's tertius gaudens is the basis of Burt's (1992) influential theory of
structural holes. Burt argues that a sparse egocentric network with few redundancies
(few members of the network know each other) is a social structure rich in structural
holes. A hole exists between two people (alters) if they are not connected to each
other but share a tie with a common third party (ego). This structural arrangement
puts the third party in the role of the tertius gaudens who can take advantage of the
two disconnected persons (or corporate actors) for private gain. Burt (1992) found,
for example, that managers in a large corporation who have networks rich in
structural holes were promoted faster and at earlier ages, compared with otherwise
similar managers whose networks lacked structural holes. Burt (1992) also found
that firms with interorganizational networks rich in structural holes earned a higher
rate of profit, compared to firms without these structural advantages.
Some argue that Burt's (1992) structural holes theory is an alternative to, rather
than an example of, a theory of social capital (e.g., Walker, Kogut, and Shan 1997:
112). These arguments emphasize the 'relational' dimension of social capital
(Nahapiet and Ghoshal 1998), which is absent or undeveloped in Burt's theory.
From the 'relational' view, social capital exists in a relationship between two people
(or two corporate actors) if they develop personal bonds, attachments, and trust. '[A]
close working group of [graduate students] working on the same problems constitute
social capital for each of them for his [sic] graduate training' (Coleman 1990: 170).
This 'relational' view also stresses the restraints on opportunism maintained by social
capital that allow cooperation take place (Walker, Kogut, and Shan 1997; see also,
Granovetter 1985).
92 - Corporate Social Capital and Liability

Rather than pitting structural holes theory against a relational definition of social
capital, we believe it is more theoreticalll productive to consider these as alternative
views of social capital (Obstfeld 1997) and to concentrate instead on delineating
and analyzing their characteristic social structures. Burt's (1992) theory emphasizes
only one of several structures of social capital. Coleman's (1988: 98) original
definition is broad enough to encompass other structures: 'Social capital is defined
by its function . It is not a single entity but a variety of different entities, with two
elements in common: they all consist of some aspect of social structures, and they
facilitate certain actions of actors-whether persons or corporate actors-within the
structure. Like other forms of capital, social capital is productive, making possible
the achievement of certain ends that in its absence would not be possible.' Similarly,
Bourdieu's (Bourdieu and Wacquant 1992) definition is broad enough to include
multiple structures of social capital. For Bourdieu, social capital is 'the sum of the
resources, actual or virtual, that accrue to an individual or a group by virtue of
possessing a durable network of more or less institutionalized relationships of
mutual acquaintance and recognition' (Bourdieu and Wacquant 1992: 119).
Social capital 'inheres in the structure of relations between and among actors'
Coleman (1988: 98). Structural holes theory emphasizes a structure of social capital
characterized by sparse networks and few redundancies. For Burt (1992), social
capital resides in the patterned absence of ties. This view is consistent with the
argument that social structure is defined more by the patterned absence than
presence of ties (White, Boorman, and Breiger 1976). What is the alternative? We
argue that social capital also inheres in the structure of social networks as the
patterned presence of ties. The alternative to the holes view of social capital is what
we call social cohesiveness, where the structure of social capital is characterized by
dense networks and mUltiple redundancies. This structure of social capital
corresponds to another of Simmel's triad types, the third party who acts as a
mediator or 'non-partisan' to create or preserve group unity: The non-partisan either
produces the concord of two colliding parties, whereby he withdraws after making
the effort of creating direct contact between the unconnected or quarreling elements;
or he functions as an arbiter who balances, as it were, their contradictory claims
against one another and eliminates what is incompatible in them' (Simmel 1950:
146-147).3 Also see Nooteboom (this volume), who describes six roles of the third
party (the so-called 'go-between').
Common examples of this structure of social capital include real estate
brokerage, literary agency, and political mediation. 4 For example, some Washington
lobbyists specialize in the introduction of corporate actors to public officials (such as
executive agency officials or congressional representatives) (Coleman 1990: 180-
182). Coleman (1990: 180) calls the third parties in these social structures
'intermediaries in trust.' The third-party intermediary is able to bring together the
other two parties because each one trusts the intermediary. This structure of social
capital is pervasive in society; as Coleman (1990: 184) describes:
This form of intermediary exists in all areas of social life. For example, professors write
letters of recommendation to prospective employers about students, and persons seeking
a job or a loan list other persons who will recommend them. The acceptance of a
recommendation by a prospective employer or creditor is a placement of trust in the
Social Capital by Design - 93

judgment of the intermediary, which allows a placement of trust in the ability of the
prospective trustee to perform as expected. If the latter defaults, then the trustor's trust in
the intermediary's judgment is reduced.
A social actor can advertise its position in the social cohesiveness structure of social
capital as a valuable resource. Consider, for example, the language used by
Mayfield, a venture capitalist firm, in its promotional brochure; as quoted by
Freeman (this volume): 'Because of our long association with a large number of
successful companies and entrepreneurs, a relationship with Mayfield is highly
regarded. It can enhance the credibility of a young company with potential
customers, vendors and employees, and with other financial institutions.' Similarly,
professional service firms, such as advertising agencies and investment banks, 'sell'
access to the social capital inherent in their networks. Indeed, advertising agencies
that occupy a central position in the market are likely to be kept by their corporate
clients, indicating the value clients place on this structure of social capital (Baker,
Faulkner, and Fisher 1998).
The level of trust in the social cohesiveness structure is probably higher, on
average, than the level of trust in the tertius gaudens arrangement. This is one reason
why some critics argue that structural holes theory is not a theory of social capital
(see above). Because the tertius gaudens exploits the structural hole between two
alters, the level of trust is presumed to be low. This is not necessarily so. Each alter
may trust ego, even though the alters are unaware of each other's existence; in other
cases, the alters may prefer to remain out of contact with each other, relying instead
on their trust in ego. Moreover, the issue of trust in the social cohesiveness structure
is not as unambiguous as it might seem. Sometimes an 'intermediary in trust' runs
the risk that he or she will be circumvented or 'cut out' by opportunistic alters. For
example, the risk that the principals in a real estate transaction might consummate
the deal in secret, saving the commission owed to the broker, is so high that the
standard legal contract between a real estate broker and seller contains protections
against such actions. 5
The distinguishing feature between the two structures of social capital is not
trust, but the answer to this question: What does the social entrepreneur do with the
gap between alters? While one social entrepreneur exploits the structural hole,
keeping alters apart, another social entrepreneur may choose to close the gap,
bringing together the two alters. These actions represent the two basic
entrepreneurial strategies for accessing and mobilizing the social capital inherent in
social networks. We next describe and illustrate these strategies, followed by our
analysis of the relationship between entrepreneurial strategies and institutional

Two Entrepreneurial Strategies

Strategy is used here to refer to a social entrepreneur's plan of action for using
network structure to access and mobilize social capital. Burt's (1992)
conceptualization of social capital as the structured absence of ties favors what we
call 'disunion strategies.' In this strategy, the social entrepreneur generates 'profit' by
taking advantage of the disconnection of the two parties. The disunion strategy is
94 - Corporate Social Capital and Liability

Alter I Alter 1

Ego Ego

Alter 2 Alter 2

Disunion Union
Figure 1. lIlustration of disunion and union strategies

illustrated in Figure 1. As shown, ego (the tertius gaudens) is linked to two

disconnected alters. The hole is represented by the blank space between alter 1 and
alter 2, and enables ego to play the two alters against each other or to secure a
valued resource from one and provide it to the other, extracting a profit in the
exchange. In such a case, ego benefits from the absence of a connection between the
two alters, may act to keep the alters apart, and at the very least, chooses not to
introduce the two alters. For disunion strategies, value is created by the exploitation
of these structural holes; value, therefore, is a 'private good'-the benefits of social
entrepreneurship accrue to the third party (ego).
The view of social capital as social cohesiveness leads to the alternative
entrepreneurial strategy we call 'union strategy'-illustrated in Figure 1. In this case,
ego (Simmel's non-partisan or arbiter) is linked to two alters who are disconnected
or in conflict. Ego 'closes' the gap between alters by bringing them together or by
resolving their differences. This suggests a sharing or exchange of resources. The
union strategy produces what Coleman (1988: 107) calls 'closure'-a social structure
that creates the conditions for the enforcement of norms through 'sanctions that can
monitor and guide behavior.' The combination of norms and trust that emerge under
these structural conditions facilitates additional use ofthe union strategy.

Disunion Strategies In Action

Disunion strategies are common in situations where the formal differentiation of the
organization presents the social entrepreneur with many temptations to play one
person (or department) against another person (or department). Burt's (1992)
original study of managers and structural holes was conducted in one such
organization. Consider, for example, an entrepreneur with ties to a person in the
sales department who collects current information about customer needs, and to
another person in the marketing department who is desperate for customer input for
a new product. The entrepreneur can position him or herself to sales as someone
with influence over the company's product development process and to marketing as
a source of new information (Obstfeld 1997). With this disunion strategy, the
entrepreneur seeks to benefit without ever introducing the two alters, a move that
would eliminate the advantageous position. A similar case is the classic role of the
boundary spanner who links two disjoint groups (Friedman and Podolny 1992).
Competitive markets are principal locations of disunion strategies at the
interorganizationallevel. Simmel (1950) notes that the market is a prime example of
Social Capital by Design - 95

the tertius gaudens strategy writ large. Here, disunion strategies are characterized as
'rivalry,' where two or more sellers vie 'for opportunities of exchange' with a buyer
(Weber 1978: 63; Swedberg 1994: 271). For example, Coca-Cola Company
maintains relationships with six different advertising agencies (Baker, Faulkner, and
Fisher 1998: 149), playing one advertising agency against the other in an elaborate
disunion strategy. The practice of competitive bidding is based on the disunion
strategy, where multiple sellers are pitted against each other. For example, disunion
strategies are evident in the garment industry studied by Uzzi (1996a) in cases where
dress manufacturers (buyers) select contractors (sellers) on the basis of price alone
to effect discrete, nonrecurring exchanges. Of course, sellers in an industry suffering
from intense competition caused by the buyers' relentless use of the disunion
strategy may become motivated to collude, employing an illegal union strategy to
counterbalance the power of buyers (Baker and Faulkner 1993).
Disunion logic drives the avoidance of ties in a competitive market. Competing
companies will not use the same supplier because doing so would put the supplier in
the structural position of the tertius gaudens. For example, General Motors avoids
using the investment bank Goldman Sachs because its chief American rival, Ford
Motor Company, uses Goldman as its main bank (Baker 1990). Similarly, corporate
competitors avoid using the same advertising agencies, citing 'conflict of interest' as
their rationale (Baker, Faulkner, and Fisher 1998).
The social structure of disunion strategies can be fluid, particularly in dynamic
markets. When AT&T and China began negotiations to install an undersea cable
system to provide a telecommunication link between China and the U.S., the
entrepreneurial leverage associated with the disunion strategy shifted from one actor
to another. AT&T initially approached the negotiation as the owner and operator of
international marine cables, enjoying an advantageous negotiating position (Glain
1997). Deregulation and other changes in the telecommunication industry, however,
unleashed a variety of competitors (such as Baby Bell SBC Communications,
Nynex, Britain's Cable & Wireless, and Japan' s Kokusai Denshin Denwa Co.) that
the Chinese invited into the negotiations. One of China's key negotiators indicated,
'Our general policy is to not engage in projects that exclude other parties. We want
to engage as many companies as possible on an equal basis' (quoted in Glain 1997).
This approach nullified AT&T's disunion strategy by leveraging confidential
information provided by the competing companies and the disconnections between
them. Ultimately, the Chinese insisted on a 14-member consortium that included all
of AT&T's major rivals.
Finally, direct exploitation of disconnected parties may represent only a fraction
of the activities associated with disunion strategies. Some disunion strategies require
constant effort and vigilance (perhaps even subterfuge) to keep alters apart and to
maintain the structural conditions of disconnection, secrecy, and concealment. Price-
fixing conspiracies, for example, require conspirators to deliberately maintain the
ignorance of their corporate customers (Baker and Faulkner 1993). Some corporate
actors attempt to prevent competitors' actions or the development of new regulations
that would reduce or close the structural holes in markets, fighting to maintain the
structural conditions that favor disunion strategies. For example, the strategic
alliance of IBM and Apple was opposed by many competitors (Baker 1994a) who
96 - Corporate Social Capital and Liability

viewed the potential union as a threat that would reduce their ability to compete for
customers-that is, to operate using the disunion strategy.

Union Strategies In Action

The formal differentiation of an organization does not always lead to disunion
strategies. For example, Kanter (1983a: 141) describes the union strategy used by a
corporate entrepreneur who created an internal alliance between sales, service, and
product development:
He first wrote a memo to all of the sales people in his area, copying the district managers
for service and products.... He then held a series of sales meetings, inviting commercial
and service staff too .... [He] explained and reexplained the benefits of cooperation across
the sales/service/products boundaries to people from each function. (Ashkenas et al.
1995: 17)

Similarly, Burgelman (1991) depicts how a corporate entrepreneur at Intel

collaborated with two other product champions to develop RISC processor
technology and to line up a customer base in advance of Intel's entry into the new
market. Ashkenas et a!. (1995) describe union strategies used to transform General
Electric's Retailer Financial Services into a 'boundaryless' organization. For
example, a systems manager employed union logic to streamline a business process:
'Nastasi then brought together a group of systems, marketing, finance, and customer
service people and challenged them to complete new customer conversions in a
matter of days, not weeks.' In many cases, union strategies are used by senior
managers targeting major organizational change. Union strategies are also initiated
by relatively low-ranking employees with more modest objectives.
Union strategies exhibit variation in official support. General Electric CEO Jack
Welch, for example, officially sponsored GE's Corporate Executive Council to
stimulate the exchange of information and collaboration (Baker 1994a,b). In
contrast, Hutt, Reingen, and Ronchetto (1988) describe the emergence of a new
product through the collaboration of actors from multiple divisions along with key
customers-all acting outside prescribed corporate guidelines. Similarly, the
successful Intel intrapreneurs described above went against explicit corporate policy
and had to disguise the RISC project as just another 'co-processor' until sufficient
progress had been made (Burgelman 1991).
Union strategies also occur in markets. Some union strategies involve the
introduction of initially unconnected suppliers or customers via common third
parties. Starr and MacMillan (1990: 86) describe a case of an entrepreneur who
introduces two suppliers:
A corporate entrepreneur saw an opportunity to connect two of his major suppliers while
developing a new medical products business. One manufacturer, in the south, had cheap
labor costs and good employee morale, but was nonetheless losing money due to lack of
business. The other, in the north, had high labor costs and an employee shortage. By
introducing these two manufacturers the venture manager reduced the costs of one and
increased the sales of the other.
Uzzi (1996a: 679) describes a similar union strategy used in the New York garment
Social Capital by Design - 97

One CEO explained how [a] tie formed between him and a manufacturer named 'Diana.'
He said that his contact with Diana began when Norman, a close business friend of his
and Diana's, asked him 'to help Diana out' in a time of need (cut her fabric at a special
price and time), even though he had no prior contact with her.... [The CEO said] 'So why
did I help her out? Because Norman asked, 'Help her out.'
The formation of strategic alliances-joint ventures, technology sharing,
marketing arrangements, product development, and others-{;an result from union
strategies. For example, Coming introduces alliance partners to each other, fostering
creation of ties between them. AT&T Global Information Solutions convenes an
annual conference in which its strategic allies meet in 'alliance fests' used to generate
new alliances and associations (Baker 1994b). Digital Equipment Corporation
(DEC) invites its alliance partners to a four-day conference to share information and
foster alliance formation (Gulati 1995a). Similarly, the monthly meetings of the 128
Venture Group were convened by an entrepreneur and venture capitalist to establish
a place where venture capitalists, entrepreneurs, consultants, and management team
candidates could meet and explore collaborations (Nohria 1992b).
The introduction of disconnected parties is only a fraction of the activities
associated with union strategies. Union activities include investing in already
established ties, cultivating ongoing collaborations, and maintaining the general
structural conditions that facilitate union strategies. Some union strategies are self-
sustaining, such as those associated with the trading groups in the industrial districts
of north central Italy and southwestern Germany (Powell 1990). In other cases,
however, such as real estate brokerage, the repeated cultivation of new alters is
necessary to stay in business. Of course, the real estate broker with a good reputation
and lots of contacts enjoys the benefits of union strategies in reverse, as satisfied
customers refer new alters to the broker.
These examples illustrate the general principle that union logic drives the
selection of ties, whereas disunion logic drives the avoidance of ties (as described
above). Disunion strategies proscribe the use of common third parties, such as
suppliers, while union strategies prescribe the use of common third parties, such as
alliance partners. These strategies may exist side by side in the same institutional
context, which is a topic we take up in the following section.

So far, we have not been very specific about the context in which social
entrepreneurs operate. This was intentional, so that we could be clear about the
distinctions between the two structures of social capital and their corresponding
strategies. However, the relationship of structures and strategies to institutional
context is critical. First, we argue that the nature, level, and forms of social capital-
and therefore the strategies social entrepreneurs employ~epend on the structure
and culture of the institutional context. This relationship holds in the institutional
context of an organization as well as that of organizational fields, business sectors,
industries, and markets. Second, we argue that the frequency, legitimacy, and
success of an entrepreneurial strategy depends on its 'fit' or compatibility with the
institutional context in which it is used.
98 - Corporate Social Capital and Liability

Table 1. Institutional context and entrepreneurial strategies

Institutional Context Disunion Strategy Union Strategy

Structural Conditions
size large size small size
density sparse networks dense networks
connectivity disconnected networks integrated networks
formal differentiation many formal boundaries few formal boundaries
Cultural Conditions
rules of exchange competition cooperation
norms opportunism. distrust reciprocity. trust
orientation individualist orientation collectivist orientation

Nahapiet and Ghoshal (1998) argue that the 'firm' is a better institutional setting
than the market for the development of high levels of social capital; because the firm
is a 'social community' (Kogut and Zander 1996: 503), it enjoys an 'organizational
advantage' over markets. However, some firms are organized and operated as
'markets,' and some markets are organized and operated as 'firms' (Eccles 1981;
Eccles and White 1988; Stinchcombe 1985).7 Therefore, we make the bold
assumption that the characteristics of 'institutional context' can be specified in such a
way that they apply to both firms and markets. An advantage of our approach is the
generalization of the concepts of corporate social capital across institutional levels-
that is, within firms and between firms. The relationship between institutional
context and strategy is summarized in Table 1.

Structural Conditions
By structural conditions, we refer to the 'network configuration' of an institutional
context, consistent with Nahapiet and Ghoshal's (1998) definition of the 'structural'
dimension of social capital. Standard network measures can be used to represent a
configuration. Nahapiet and Ghoshal (1998) propose density, connectivity. and
hierarchy. We substitute 'formal differentiation' for hierarchy. because an
organization can be differentiated along three dimensions-spatial. horizontal, and
vertical. Markets, too, are organized along these three dimensions, as economic
geographers have documented. Formal divisions such as these, whether in firms or
markets. create structural holes. We add 'size,' since the number of social actors
influences the fragmentation of an institutional setting, and along with it, the number
and extent of structural holes.
The structural conditions associated with disunion strategies are large size,
sparse and disconnected networks, and many formal boundaries. The design of most
large-scale, traditional organizations favors disunion strategies because it creates so
many structural holes between departments, across levels, and between spatially
separated operations. Formal differentiation, for example, hampers collaboration
across boundaries (Lawrence and Lorsch 1986). Large size disfavors the integrated
'network' organizational design (Baker 1992a). Similarly, markets with many players
are more fragmented than markets with few players, producing many structural
holes. The alert tertius gaudens in a fragmented market generates profit by
Social Capital by Design - 99

arbitraging across these gaps in the social structure of trading (Baker 1984), just as
the tertius gaudens in a fragmented firm generates profit by exploiting the
disconnections inside the organization (Burt 1992).
The structural conditions favoring union strategies are small size, dense and
integrated networks, and low differentiation (see Table 1). Union strategies are
common in network organizations, for example, because these organizations are
characterized by flexibility, lateral ties, and a high degree of integration across low
formal boundaries (Baker 1992a). A dense network of customers, producers, and
suppliers provides a structural basis for cooperation (Perrow 1992), such as the
dense social networks of firms, local universities, community colleges, research
institutes, financial institutions, trade associations, and regional governments in
Silicon Valley (Saxenian 1991, 1994).
The biotechnology industry features dense and well-connected networks of
interfirm cooperation as well (Walker, Kogut, and Shan 1997; in this volume see
also Stuart, Smith-Doerr et al. and Omta and Van Rossum). These structural
conditions are conducive to the formation of new alliances based on the union
strategy. For example, the likelihood that two previously unallied firms will form an
alliance increases with the number of third-party ties they have in common (Gulati
1995a). Similarly, the greater the number of research and development alliances and
other types of collaborations a biotechnology firm has at a given time, the more
diverse its future portfolio of ties will become (Powell, Koput, and Smith-Doerr

Cultural Conditions
Cultural conditions can be defined in many ways. Indeed, the definition of culture
itself is a subject of considerable debate and rival interpretations (DiMaggio 1994;
Scott 1995). Rather than trying to resolve this debate, we propose a simple definition
of cultural conditions-the institutionalized rules of exchange, norms, and social
orientation in an institutional context-and focus primarily on the relationship
between strategy and cultural context. s
Rules of exchange are shared social understandings about 'who can transact with
whom and the conditions under which transactions are carried out' (Fligstein 1996:
658). For example, exclusivity (sole-source) is a rule of exchange governing buyer-
seller relationships in the advertising industry (Baker, Faulkner, and Fisher 1998:
151). Competitive rules of exchange prohibit union triads. The avoidance of
common suppliers is a good example (Baker 1990). Disunion strategies are
legitimate and successful in firms operated as 'markets' (Eccles and White 1988). In
settings such as markets operated as 'firms' (Stinchcombe 1985), however,
cooperative rules of exchange discourage disunion strategies (which would be
interpreted as self-serving, opportunistic behaviors). Similarly, organizations in
small-firm networks share information, establish long-term relationships, and
support each other's efforts, in opposition to the competitive rules of exchange in
classic markets (Perrow 1992). The search for greater efficiencies through closer ties
between customers and suppliers can displace traditional rules of exchange that pit
suppliers against one another. New developments in just-in-time (JIT) inventory
control, for example, call for close coordination between customers and their
100 - Corporate Social Capital and Liability

suppliers. Honeywell orchestrates close cooperation between its internal buyers and
five suppliers who would ordinarily compete for business (Bleakley 1995).
Norms are shared expectations that regulate behavior (DiMaggio 1994),
including choice of entrepreneurial strategy. Norms of openness, teamwork, trust,
and reciprocity favor union over disunion strategies. For example, Putnam (1993a)
argues that norms of reciprocity in the industrial districts of northern Italy foster
interfirm cooperation and limit opportunistic behavior-that is, the use of disunion
strategies. Similarly, norms of reciprocity support the union strategies evident in
such business groups as the Japanese kieretsu and Korean chaebol.
Social orientation refers to the distinction between individualism and
collectivism (Triandis 1995: 2). Individualism, for example, is 'a social pattern that
consists of loosely linked individuals who view themselves as independent of
collectives; are primarily motivated by their own preferences, needs, rights, and the
contracts they have established with others; and emphasize rational analyses of the
advantages and disadvantages of associating with others' (Triandis 1995: 2).
Individualism favors disunion strategies; collectivism favors union strategies. For
example, collectivist practices such as cross-functional teams, multi-level
management networks, group-level reward systems, and team-building programs,
facilitate union strategies within organizations (Baker 1994a). Recent research has
shown that a collectivist orientation often increases the odds of alliance formation
(Dickson and Weaver 1997). Private economic associations and political
organizations foster cooperation in the industrial districts of northern Italy (1993a).
The robust collaboration found in these industrial districts is supported by technical
colleges, vocational training, supportive banks, and extended kinship ties (Powell

Distribution of Strategies
We argue that the frequency, legitimacy, and success of an entrepreneurial strategy
depends on its 'fit' with the 'design' of the institutional context in which social
entrepreneurs operate. This suggests the following proposition: The ratio of disunion
to union strategies varies according to the structure and culture of the institutional
context. At one extreme, disunion strategies dominate in settings characterized by
sparse, disconnected, and differentiated networks, along with competitive rules of
exchange, opportunism, and an individualist orientation; at the other extreme, union
strategies dominate in settings characterized by dense, connected, and
undifferentiated networks, coupled with cooperative rules of exchange, reciprocity,
and a collectivist orientation. A mix of strategies occurs in an institutional context
located between these two extremes.
The distribution of strategies in a given context can be determined by a triads
census (Wasserman and Faust 1994: 556-602). The pattern of strategic alliances in
the global automobile industry 9 illustrates one empirical distribution (Baker 1992b),
though the triads approach can be applied in any institutional context. For strategic
alliances, a triads census includes only four possible triad isomorphism classes: 1)
the null triad, composed of three unallied firms; 2) the dyad, composed of three
firms of which only two are connected by an alliance; 3) the disunion triad,

Table 2. Triads census of alliances in the world automobile industry

Null Triad Dyad Disunion Triad Union Triad
[003]" [102] [201] [300] Significanceb
Type of Alliance observed expected observed expected observed expected observed expected 'tD 'tu

All Alliances 5616 5510 1814 2010 314 241 26 9 5.972 5.540
joint venture 7045 7024 681 722 43 24 I 0 4.463 1.538
manufacturing/assembly 6631 6607 1057 1103 78 60 4 1 2.767 2.939
technology sharing 6877 6861 844 873 46 36 3 0 1.981 3.752
supplier relationship 6529 6513 1158 1186 77 70 6 1 .976 4.079
marketing/distribution 7030 7024 712 722 26 24 2 .555 3.573
equity investment 7257 7256 502 503 10 11 I 0 -.278 3.466
aNumbers in square brackets [MAN] refer to triad isomorphism classes using standard labeling, where M = number of mutual dyads, A =number of CIl
asymmetric dyads, and N = number of null dyads. on
btau test statistic from TRIADS (Walker and Wasserman 1987): 'tD = disunion triads; 'tu = union triads. E
Source: Baker (1992b). Data from Wards International, 1985 (N = 37 automobile companies). (')


102 - Corporate Social Capital and Liability

composed of two alliance dyads; and, 4) the union triad, composed of three alliance
dyads. 10 The first class, the null triad, represents a classic competitive situation-the
complete absence of strategic alliances among three firms. The second class, the
dyad, represents an isolated alliance between two firms . The third class, the disunion
triad, represents the tertius gaudens arrangement (as illustrated in Figure 1). The
fourth class, the union triad, represents the social cohesiveness structure of social
capital (see Figure 1).
The empirical distribution obtained in the triads census of the automobile
industry is shown in Table 2. For our purposes, we focus here on the two triad types
of particular interest, disunion and union triads. Both triad types occur much more
often than expected by chance alone, considering all alliance types combined. 11
However, if we examine the triads census for each type of alliance, we find an
interesting pattern: 1) Disunion triads (but not union triads) occur more often than
by chance in joint ventures; 2) Disunion and union triads occur more often than by
chance in two types of alliances-technology sharing and manufacturing/assembly;
3) Union triads (but not disunion triads) occur more often than by chance in three
types of alliance-supplier ties, marketing/distribution, and equity investments.
This pattern suggests that the structure of social capital varies by alliance type.
The structure of joint ventures implies that social capital is accessed and mobilized
by exploiting structural holes, using the tertius gaudens strategy. The structure of
supplier ties, marketing/distribution, and equity investments implies that social
capital is accessed and mobilized by closing structural holes. The mixed structure of
technology sharing and manufacturing/assembly suggests the presence of both types
of social capital. However, the statistically significant use of both disunion and
union strategies may indicate a competition of strategies in which neither dominates.
The general relationship between entrepreneurial strategies and institutional
context is illustrated in Figure 2. This stylized representation implies the possibility
of change or movement, of the transformation of an institutional context and its
corresponding strategies. The intersection of the curves in Figure 2 represents a
balance of disunion and union strategies. This point is an unstable state in which
neither strategy dominates, such as the mixed structure of triadic strategies for
technology sharing and manufacturing/assembly in the automobile industry (Table
2). If so, then the automobile industry may fall back on earlier, simpler, and
overlearned strategies (arms' -length competition), as people and organizations are
prone when faced with uncertainty and ambiguity (Weick 1995: 102).
For an organization, the point of intersection in Figure 2 indicates a particularly
risky stage in an organization's transition from one institutional design to another.
For example, a firm attempting to improve collaboration and cooperation must foster
both the structural and cultural conditions that favor the union strategy (moving
from left to right in Figure 2). Some consultants claim that a hierarchy can be
converted into a network organization simply by adding links (Lipnack and Stamps
1994: 72; see, also, Mueller 1986), but structural change is not enough. Failure to
change both structural and cultural conditions endangers a change effort. For
example, the effort to transform Industrial Computer and Control Group failed
because change agents altered only organizational structure (Nohria and Berkley
Social Capital by Design - 103



.- .- -- --
/ ""
- - Disunion
SO/50 - - - - Union


Small, dense, integrated Large, sparse, disconnected

networks; cooperation, networks; competition,
trust, collectivism Institutional Context opportunism, individualism

Figure 2. Entrepreneurial strategies by institutional context

1995). By replacing hierarchy with a network design, they may have induced more
union strategies, but by itself this structural change was not enough. Without a
corresponding cultural change, the effort to foster cross-divisional collaboration was
doomed. Employees did not fundamentally change their strategies for action (Nohria
and Berkley 1995). The change effort may have collapsed at or near the point of
intersection in Figure 2.

Our chapter attempts to specify and clarify dimensions and structures of social
capital. We offer a set of concepts that encompasses different views of the structural
sources of social capital, the basic strategies used to access social capital, and the
relationship between institutional contexts and strategies. We argue that social
capital 'inheres in the structure of relations between and among actors' (Coleman
1988: 98) in two fundamental ways-one based on the patterned absence of ties, the
other on the patterned presence of ties. Structural holes theory (Burt 1992)
emphasizes the absence of ties, where social capital resides in a social structure
characterized by sparse networks and few redundancies. We maintain that social
capital also inheres in social structure as the presence of ties, which we call social
104 - Corporate Social Capital and Liability

cohesiveness, where social capital is found In dense networks with multiple

Each structure of social capital represents opportunities to access and mobilize
the resources inherent in a social network. The first structure calls for a 'disunion'
strategy that exploits the structural holes between alters by keeping them apart. The
second structure calls for a 'union' strategy that creates value by bringing alters
together, closing the hole between them. Our qualitative and quantitative examples
provide ample evidence of the use of these strategies in a wide range of institutional
contexts, both inside and between organizations. The concepts of two types of social
capital (structural holes versus social cohesiveness) and two types of entrepreneurial
strategies (disunion and union) can be generalized and applied to understand the
structure and use of social capital in many different institutional settings.
The actual distribution of strategies in a particular institutional context depends
on the 'design' of the institutional context in which social entrepreneurs operate. An
entrepreneurial strategy that 'fits' the structure and culture of a given institutional
context occurs more frequently, enjoys greater legitimacy, and will be more
successful in the long run, compared to a strategy that does not fit. Disunion
strategies dominate in organizations and markets characterized by sparse,
disconnected, and differentiated networks, coupled with competitive rules of
exchange, opportunism, and an individualist orientation; union strategies dominate
in organizations and markets characterized by dense, connected, and
undifferentiated networks, coupled with cooperative rules of exchange, norms of
reciprocity, and a collectivist orientation.
Of course, a mix of triadic strategies falls between the pure disunion and union
extremes. Thus, one avenue of additional research is to explore further the precise
connection between institutional context and entrepreneurial strategy. This line of
work calls for a demography of social entrepreneurial relationships (see, also, Baker,
Faulkner, and Fisher 1998: 173). The statistical methods available for taking a triads
census (Walker and Wasserman 1987), which we used in our analysis of the
distribution of strategies in the global automobile industry (Baker 1992b), can be
used to determine the distribution of strategies in any organization, market, or
organizational field. In addition, organizational change can be tracked by measuring
shifts in the distribution of strategies over time, using methods designed to model
longitudinal change in networks (e.g., Leenders 1995a, 1996; Wasserman and Faust
1994). By coupling new and better measurement methods (e.g., Borgatti 1997; Han
and Breiger, this volume; Doreian, this volume) with more precise concepts of social
capital and their corresponding entrepreneurial strategies, it is possible to take
another step forward in the exploration of corporate social capital.
We are grateful to the editors of this volume for their helpful comments and suggestions. Direct
correspondence concerning this chapter to Wayne Baker, University of Michigan Business School, 701
Tappan Street, Ann Aibor, MI 48109 (wayneb@umich.edu).

I. The debate takes place in published works (see citations herein) as well as in infonnal discussions
on the Internet, such as the recent interchanges on social capital in SOCNET between January and June,
1997. We cite the published literature to support our points, but we also acknowledge the contributions to
the debate in SOCNET made by Xavier De Souza Briggs, Robert Putnam, Barry Wellman, and others.
Social Capital by Design - 105

2. This point was also made in SOCNET by Barry Wellman and Robert Putnam (January 1997).
3. Simme1 (1950) describes another triad type, 'divide and conquer,' in addition to the tertius gaudens
and 'non-partisan' types. In 'divide and conquer,' the third party deliberately introduces conflict between
the other two. This type is not our concern in this chapter.
4. Real estate brokerage and literary agency qualify as this type because the ego is not a principal in a
transaction, but instead brings together the two principals (alters) who consummate the transaction. In the
first case, the real estate broker (ego) matches a buyer and seller (alters) who exchange property for
money; in the second, a literary agent (ego) matches a publisher and author (alters) who exchange 'the
property' (manuscript) for royalties. Some types of brokerage do not qualify as this type. For example, in
a wholesaler arrangement, the broker is a principal in one transaction (buying goods from a manufacturer)
and in a second transaction as well (selling the goods to the consumer).
5. Consider, for example, the protections contained in the standard 'listing agreement' (officially, the
'cooperative selling contract') used in the state of D1inois to define the legal obligations of the seller to the
real estate broker: 'SELLER SHALL: Cooperate fully with Broker; refer all inquiries to Broker; conduct
all negotiations through Broker; ... and pay a real estate brokerage commission of X % of the sale price; if
1) Broker provides a purchaser ready, willing, and able to purchase in accordance with this Contract; or
2) if the property is sold, exchange, gifted, or optioned by Broker or by or through any other person
including the Seller during the period of this contract; or 3) if it is sold directly or indirectly within six (6)
months after termination of this contract to a purchaser to whom it was offered during the tenn thereof.'
6. It is possible, of course, that the individual pursuit of entrepreneurial profit as a private good
produces value at the collective level. This line of reasoning is consistent with traditional management
and economic theories. For example, the tournament model of mobility assumes that competition among
managers yields better ideas, more satisfied customers, and greater shareholder value. The theory of the
market similarly assumes that individual striving maximizes social welfare. Burt (1992) does not
emphasize the public-goods aspect of social entrepreneurship, though he mentions the possibility: An
entrepreneur is 'a person who generates profit from being between others. A nonprofit player, pursuing
entrepreneurial opportunities just for the pleasure of being the one who brings others together to build
value, could choose to reinvest it all' and strengthen existing relationships (Burt 1992: 34-35). Of course,
social capital can be considered a public good in and of itself. Putnam argues, for example, that
Coleman's (1988) original concept of social capital incorporates such a view. He adds, however, that the
views of social capital as a private and public good are complementary (see discussions in SOCNET; see,
also, Putnam 1993a, 1995; Knoke this volume).
7. Nahapiet and Ghoshal (1998: 261) briefly mention that some interorganizational networks may
develop an institutional context that is conducive to high levels of social capital, and point to this as an
avenue of future research.
8. We acknowledge, but do not elaborate here, the importance of the 'constitutive' as opposed to the
'regulatory' view of culture (DiMaggio 1994). In addition, we acknowledge but do not discuss the state as
part of the institutional context (Fligstein 1996).
9. We obtained data for the triads analysis from a special issue of Ward's Automotive International
(1986). This issue reported the alliances known to exist among 37 major automobile companies, classified
by the six alliance types reported in Table 2. We created a square, binary, symmetric matrix for each
alliance type, where each company is assigned a row and corresponding column, and an entry indicates
the presence or absence (0,1) of an alliance between two companies. Thus, we created six 37 X 37
matrices. We analyzed the triads structure of each matrix using TRIADS (Wassennan and Walker, 1987).
These data are treated as symmetric because alliances are naturally mutual ties, and we adjusted the
weighting vector in TRIADS accordingly.
10. A triads census includes sixteen isomorphism classes when asymmetric ties are considered
(Wassennan and Faust 1994: 566). Since alliances are mutual ties, however, there are only four classes of
interest: triad types 003 (null), 102 (dyad), 201 (disunion), and 300 (union ). These three-digit numbers
[MAN) refer to triad isomorphism classes using standard labeling, where M =number of mutual dyads, A
= =
number of asymmetric dyads, and N number of null dyads.
II. Structural hypotheses are tested using the TRIADS program (Walker and Wassennan 1987) in
which the empirical distribution of triads is compared against a theoretical distribution of triads assuming
random assignment of ties. The weighting vector used in these tests reflects the number of disunion
configurations contained in the disunion triad, and the number of union configurations contained in the
union triad. The test statistic, tau, is interpreted using the standard nonnal distribution (Wassennan and
Faust 1994: 595). We use a significance level of .05 in this analysis.
Corporate Social Capital and Liability:
a Conditional Approach to Three

Consequences of Corporate Social

Dan TaImud

This chapter defines and illuminates three aspects of corporate social capital which
are created by different aspects of corporate social structure. The chapter also shows
that corporate structure which generates social capital may become later a liability.
The chapter briefly reviews the literature regarding the contingent value of corporate
structure in creating competitive social capital. Then two other kinds of corporate
social capital, political and cognitive, are illustrated via a case study of an Israeli
Armament firm. Finally, the chapter briefly discusses the relations between the three
kinds of corporate social capital for business policy and strategy as well as for future
studies on corporate advantage.


Certain corporate social structures provide comparative advantage to the firm. The
term 'corporate social capital' is a heuristic device, assists us in comprehending how
corporate social structure benefits certain business organizations at the expense of
others. Corporate social capital is any positive, goal-specific outcome, originating
from corporate social structure. 1 It includes any means of corporate control or
business leverage, embedded in social relations, thus assisting a firm in promoting
internal assets and resources. 2 Excess profit or economic rent, which is an outcome
of the operation of corporate social capital, is depicted as 'any advantage or surplus
created by nature or social structure over a certain period of time' (Sorenson 1996:
1344). Corporate advantage may be explained then by the differential capacity of
firms to create, promote and take advantage of corporate social capital (Nahapiet
and GhoshaI1998).3
Corporate Social Capital and Liability- 107

I argue that one can differentiate analytically between three kinds of corporate
social capital, each derived from a particular dimension of corporate social structure
(Zukin and DiMaggio 1990) [see Table I]: I) Competitive corporate capital is as a
function of a firm's structural location in imperfect competition (Burt 1992). The
term expresses the extent to which the social organization of the competition, as
measured via transaction networks, systematically benefits certain market players at
the expense of others (Burt 1992; Talmud 1992, 1994; Talmud and Mesch 1997).
For example, a particular form of competitive corporate capital is the manner by
which suppliers or buyers are dependent on a player in an exchange system. 4
Another example is the ability of a firm to engage in exclusive relations with sub-
contractors, thus increasing its capacity for complex adaptation and economy of
speed (Uzzi 1997a).
By contrast, 2) political corporate capital indicates the manner in which the
political organization of the economy, institutional and legal regulation of corporate
behavior, and the collective struggles over economic options extend the strategic
possibilities of corporate executives. By virtue of membership in larger political
networks, certain firms have more institutional leverage and can buy political
protection from state and local institutions. For example, firms with direct
institutional ties to the political center may gain more collective resources than such
lacking those ties (Talmud 1992; Talmud and Mesch 1997).
In their strategic considerations, firms attempt to influence and take advantage
of legal arrangements, political ties, and state policies at various levels (Zukin and
DiMaggio 1990: 20-21; Talmud 1992; Talmud and Mesch 1997). Political
arrangements have uneven effects on market players and therefore may be defined
as political corporate capital. This is measured by strong linkage to the
institutionalized political center, as well as by institutional regulation and political
arrangements favoring the focal firm or industry. However, political arrangement are
dynamic and high reliance on political capital in one period may be detrimental in
another time. Continuing benefits from political corporate capital compel corporate
players to calculate and rearrange their corporate capital for sudden shifts in the
political sphere or to politically-directed changes in the corporate task environment. 5
Paradoxically, those corporations rich in political corporate capital which have
learned to rely on institutionalized political arrangements are among those firms,
highly prone to crisis and mismanagement following transformation of political
agencies. These firms often enjoy rewards such as lower tax rates, tariff protection,
interest-free loans, business tips, favorable environmental conditions, and direct and
indirect state subsidies (Talmud 1992). In economic terms, firms that concentrated
their efforts on using political corporate capital are engaged in rent-seeking conduct,
or in 'directly unproductive, profit-seeking behavior' (Bhagwati 1982). Therefore,
their strategic repertoire is not geared toward efficiency but toward an effective
usage of their political leverage.
Finally, only recently the organizational literature has characterized a less
known form of corporate social capital: 3) cognitive corporate capital.6 While
competitive and political corporate capital are instrumental for the external economy
of the firm, cognitive corporate capital is chiefly relevant for the internal economy
of business organization. It is defined as the extent to which a firm can facilitate
108 - Corporate Social Capital and Liability

advantageous business action by using internal networks which reconstruct shared

understanding, disciplinary power, mental models, organizational identity, tacit
knowledge, corporate norms, the ability for corporate foresight, reflexivity, and self-
awareness. 7 Cognitive corporate capital then is facilitated by relations, formal and
informal, inside the firm (Podolny and Baron 1997; Nahapiet and Ghoshal 1998). It
assists the organization in accomplishing goals, to speed production process, to
improve quality according to clients' requirements, to 'move things ahead' and to
fortify its competitiveness, effectiveness, or efficiency (Uzzi 1996a).
The importance of cognitive corporate capital cannot be overstated. Several
perspectives have currently acknowledged this unique form of corporate social
capital. In sociology and psychology, organizational self-command has been
perceived as an integral part of 'relational management' and 'knowledge
management' of the firm as a social community.8 Resource-based business strategy
has shifted organizational theory from an emphasis on value appropriation to the
exploitation of the firm's social structure for value creation. Similarly, Human
Resources Management have recently put more weight on employees' long term
relational contracts and 'employability' (Ghoshal and Moran 1996).
These cognitive assets can be constructed and reproduced because any corporate
firm is first and foremost a social community, a nexus of social relations (Nahapiet
and Ghoshal 1998). Intra-organizational social networks thus yield and convey
organizational memory and shared narratives and norms. In fact, social networks can
create or modify trust in the organizational system (Kramer, Brewer, and Hanna
1996). Social networks also convey intangible corporate assets such as socially-
produced tacit knowledge, which is particularly useful when and product complexity
and quality are at stake (Polanyi 1958; Dore 1983; Jones, Hesterley, and Borgatti
1997).9 Moreover, cognitive corporate capital, like any other social capital
discussed in this chapter, is a firm-level phenomenon. It is endowed in corporate
social relations, and not in individual players, and thus the firm can take advantage
of these nested relations (Monge, Cozzens, and Contractor 1991). Cognitive
corporate capital then may speed a firm's reaction to market cycles, enhancing
complex adaptation and reducing learning cost and time-to-market (Uzzi 1997a). It
may, furthermore, promotes collaborative production of social knowledge capacity,
which is especially relevant for the construction and transfer of socially tacit
knowledge (Nahapiet and Ghoshal 1998). It may also increase learning speed and
effectiveness, and promote smooth solutions to problem solving (Dore 1983; Uzzi
1996a, 1997a; Nahapiet and Ghoshal 1998). Still, cognitive corporate capital is a
firm-level virtual asset which decreases the economic cost of monitoring, learning,
adapting, and also diminishes agency problems. Like any other form of corporate
social capital, it has diminishing marginal utility, and thus could be accumulated up
to a certain point (Uzzi 1997a; Talmud and Mesch 1997). If external conditions
significantly change, the very same corporate social structure producing cognitive
corporate capital in one circumstance, may in another instance create a corporate
liability, thus inhibiting the firm's adjustment to new environmental circumstances.
Because cognitive corporate capital is an intangible corporate, relation-specific
asset, it cannot easily be imitated by competitors (Barney 1991). It is transmitted via
Corporate Social Capital and Liability- 109

Table 1. Three types of corporate social capital

Social Structure Producing Structural Embeddedness: Corporate Network Position
Corporate Social Capital
Examples Non-redundant Market Ties (Burt 1992)
- Rewards Profit squeezing capacity (Burt 1992; Talmud 1994); sustained
industry leadership (Talmud and Mesch 1997)
Social Structure Producing Political Economy and Institutional Regulation
Corporate Social Capital
Examples State regulation; political institutional ties (Talmud and Mesch 1997)
- Rewards State subsidies (Talmud 1992b)

Social Structure Producing Normative and Cognitive Embeddedness: Shared narratives,
Corporate Social Capital normative role expectations and commitment, corporate identity, vision
and discourse enacted via intra-organizational social networks.
- Examples 'Self-command capital' (Lindenberg 1993); 'Strong culture' (Kotter and
Heskett 1992); tacit knowledge (Polanyi 1958); 'combinational
capability' (Nahapiet and Ghoshal 1998)
- Rewards Corporate innovation (Monge, Cozzens, and Contractor 1991);
complex adaptation, 'economy of speed' (Uzzi 19%a; 1997a);
provision of product quality (Dore 1983)

a complicated web of social relations. The creation, configuration and maintenance

of cognitive corporate capital is costly, consuming time and corporate resources.
An organization needs to maintain a careful balance, therefore, between various
modes of corporate social capital, but in a similar vein to those business
organizations relying upon political corporate capital, firms which have
advantageous cognitive corporate capital tend to remain stationary. Corporate
networks conveying identities and normative expectations are inclined to become
inflexible over time (White 1992). Accordingly, ideal cognitive capital should not
prevent a firm from being flexible.
All these three forms of corporate social capital have a few common attributes:

• They are not a property of any individual agent;

• They can be accumulated over time;
• They are relation-specific and cannot easily be transformed from network to
• Therefore they are expensive, bearing investment-specific sunk cost;
• If they are not properly maintained, they tend to fade out or become
irrelevant to the extent that a firm is located in a turbulent and uncertain
environment, and to the degree that the corporate social structure producing
corporate social capital is rigid, they tend over time to turn from a corporate
capital into a corporate liability.
110 - Corporate Social Capital and Liability

There are also some significant differences between the three forms:

a. Competitive corporate structure creating corporate social capital is the most

measurable as precisely operationalized by network models of competitive
advantage (cf. Burt 1992).
b. Political corporate capital is more vague, composed of many sub-forms, and is
molded corporate actors' ties as well as by historical and institutional
c. Cognitive corporate capital is the most 'sticky' and vague form. It is a product of
both past and present relational composition inside the firm, and of interface
with other agents, such as competitors, consultants, suppliers and clients.

The following part has three sections, each of which addresses a particular
dimension of the corporate social capital. The first section summarizes the literature
on the contingent sources of competitive corporate capital. The next section
demonstrates how social structure producing political corporate capital becomes a
corporate liability. Finally, the chapter discusses to inability of a manufacturing firm
to renovate its cognitive corporate capital.


Recent literature on corporate social capital and competitive advantage has stressed
the role of weak ties and sparse social networks in determining corporate
performance (Burt 1992).\0 Burt (1992) and Talmud (1992, 1994) found that firms
using 'structural holes,' spreading their ties with unconnected market segments are
more profitable than those connecting to 'redundant market areas.'
In contrast, other studies on corporate social capital have found significant
impact of strong ties and dense networks in creating opportunities. New and
sensitive business information and opportunities are enhanced through cohesive
contacts (Aldrich and Zimmer 1986; Gilad, Kaish, and Ronen 1989). Moreover,
Krackhardt (1992), Podolny (1994) and Gabbay (1995, 1997) showed the
importance of enclosed networks for managing corporate uncertainty.
It seems, accordingly, that there is no such thing as a universal optimal network
structure (dense or sparse); as Coleman phrased it 'social relationships that constitute
social capital for one kind of productive activity may be impediments for another'
(Coleman 1994: 177).
Furthermore, the effect of strong ties seems to be a conditional one. Burt (1992)
and Han (1993) revealed that, contrary to the logic behind the theories of resource
dependence and structural holes, those American firms having high recourse
dependence on the government survive at higher rates. Also, Israeli firms that are
'locked in' by political ownership are more profitable (Talmud 1992), and are more
stable over time (Talmud and Mesch 1997). In uncertain conditions, or where
identity and role expectations are important for performance, embedding economic
transactions in strong relations, or constructing the economic deal in terms of strong
ties, could serve as remedies for survival (Dore 1983; Podolny 1994; Jones,
Hesterly, and Borgatti 1997; Podolny and Castellucci, this volume).
Corporate Social Capital and Liability- 111

Organizations develop indirect corporate leverage via social networks to

manage their dependence, while protecting their advantage in strategic positions
(Galaskiewicz 1985; Baker 1990). Juggling between contrasting business pressures,
corporate firms attempt to maintain their access to reliable and non- standard
information, while also striving to reduce trade dependency and uncertainty. Thus,
corporations are bound into repeated transactions, while also preserving other
opportunities at hand (Powell 1990; Podolny 1994). Uzzi (I996a, 1997a) found in
his study of the New York apparel industry that organizational failure increases to
the degree that the focal firm's network tends to be composed of either all arm's-
length ties (i.e. 'under-embedded,' disconnected, non-redundant contractors), or all
embedded ties (i.e. 'over-embedded,' dense, semi-integrated contractors). Uzzi
(I997a) showed that organizational survival is positively associated with a 'mixed
model' network, which provides a combined advantage. On the one hand, market
embeddedness provide benefits such as trust, joint problem-solving arrangements,
complex adaptation, reduced bargaining and monitoring costs. On the other hand,
arm's-length contacts provide the firms new and novel information outside the
immediate ties (Uzzi 1996a, 1997a). Similarly, Talmud and Mesch (1997) have
demonstrated that the probability of the leading Israeli firms to sustained their
industrial leadership is a function of the 'mix embedded ness' type of relations.
Similarly, Uzzi (1996a, 1997a) claims that each type of controlled embedded ness
'yields positive returns up to a certain point,' and 'embedded networks offer a
competitive form of organizing, but possess their own pitfalls.' Competitive
pressures create complex, dissimilar, and distinct network forms, rather than a singly
optimal embeddedness strategy (Uzzi 1997a). As a result, 'a paradox appears:
optimal networks are not composed of either all embedded ties or arm's-length ties
but integrate the two' (Uzzi 1996a: 694). In sum, the conditional approach to
competitive corporate social structure argues that the corporate value behind
network forms is contingent upon contextual features of the environment. II Next, I
will apply this insight to political and cognitive corporate capital.


Political Corporate Capital: The Demise of Corporate Leverage
This section illuminates how political corporate capital diminishes due to a
combination of organizational rigidity and changing environmental conditions.
In many cases, competitive corporate capital is not sufficient to understand the
construction and maintenance of corporate social capital. In most countries,
corporate political embedded ness of the market, institutional regulation, and
governmental restrictions on competition augment firms' market positions and
elevating barriers to entry (Levacic 1987: 118-138; Talmud 1992). In other words,
political corporate capital assists a firm in its capacity for 'price making.'
Ta'as (The Hebrew acronym for the Israeli Military Industry) was incorporated
in 1947 as an institutionalized organ of armament. It operated first in the Jewish
underground against British rule and later on for the Israeli Defense Forces, which
had been its chief client over the years. Ta'as became then a state agency which
gradually transformed into a state-owned enterprise with very low autonomy and
without financial reserves or structural leverage. State regulation and laws prevented
112 - Corporate Social Capital and Liability

private investors to compete with Ta'as in the local market, but also to from loaning
finances to the company. Ta'as' business policy and strategy was effected much
more by political considerations than by sheer economic issues (see also Samuel
1998). This situation has facilitated a monopoly position for many Israeli Defense
Industry firms. Yet, the emergence of the newly globalized weapons market, the end
of the cold war, and the peace process in the Middle East brought about a radical
change in the political embeddedness of the weapon's industry, as demonstrated in
the case of Ta'as. If political corporate capital can be measured by a premium paid
for (or excessive revenue earned by) a given investment in political ties, then-
because it is a relation-specific investment-dearly, any drastic modification of the
political structure can alter the fate of the firm, from a revenue to a loss.
Until the 1980s, the firm could use the dramatically leaking bucket of the State
of Israel's defense budget. Accordingly, its business strategy was typified by the
overarching principle of autarky and freedom from functional constraints.
Structurally, it had one main buyer (monopsony), but Ta'as was that buyer's main
supplier as well (monopoly). These structure of bi-Iateral monopoly had been
maintained and fortified by the firm's identification with the heroic Jewish
underground under the British Mandate in Palestine, to be superseded later on by its
strong cultural affiliation with the dominant security symbols of the State of Israel.
Already heavily invested in political corporate capital, the firm rationally preferred a
strategy emphasizing effectiveness over efficiency. As a result, the organization
forged an ambitious business strategy: it tended to 'make' rather than cheaply 'buy'
various components of its product; it stressed, moreover, activities of research and
development function and employed a disproportionately large, permanent and
expensive workforce (increasing from 1000 employees in 1948 to 14,500 in 1985;
Talmud and Yanovitzky 1998). Maintaining its political corporate capital. the
organization possessed an excessive production capacity. accompanied by useless
vertical integration along many of product lines including logistics, maintenance and
support systems (Kleimann 1992; Talmud and Yanovitzky 1998) to overcome
prospective shortage or demand fluctuation. Additionally, the firm diversified into
about 500 different products in various market areas, at the expense of achieving
competitive advantage by focusing on core advantage. Ta'as chose, in fact, to
focusing on technological push (imposing scientific innovation and engineering
excellence) rather than by market pull (imposing product's fit to customers'
preferences) (Samuel 1998; Talmud and Yanovitsky 1998). Additionally,
competition in the global market was considered to be only secondary to the focus
on the local Israeli market. Many governmental restrictions on military equipment
export were compensated by state agencies. Still, the industry was booming until
the 1980s, as a result of its political corporate capital. In the 'heroic' period of
1947-1985, Ta'as was socially conceived as strategically important, and thus
enjoyed close institutional linkage with the Ministry of Defense, which was formally
its owner and main buyer (Talmud and Yanovitzky 1998). The fact that Ta'as
lacked competitive corporate capital (its transaction network was concentrated on a
single buyer for any single product) was compensated for by the higher value of its
political corporate capital. which was deemed instrumental for corporate solvency.
Corporate Social Capital and Liability- 113

Following 1985, there have been increasingly severe cuts in the Defense budget.
The State of Israel thus has become more sensitive to armament prices. In addition,
global prices have decreased, as suppliers' competition in the world market has
intensified. The sharp modification of Ta'as' political corporate capital created a
series of acute structural crises, accompanied by a severe financial toll, layoffs, rapid
managerial turnover, and cultural shocks to employees. Moreover, it exposed the
structural weakness of the firm's transaction networks and its high dependency on
suppliers and customers (Talmud and Yanovitzky 1998). Ta'as' political corporate
capital has been eroded because it was based merely on a social ties with political
agents which were no longer relevant.
Additionally, it is important to note that political and competitive corporate
capital may also be linked. The case of Ta'as shows that the political structure
affects the industry competitive structure, and thus provide regulated competitive
leverage for the firm. Moreover, state regulation and laws prevented private
investors to compete with Ta'as in the local market, but at the same time enforced
the company to lean on political backup. Barriers-to-entry and barriers to exit were
fixed by the political organization of the industry. The competitive structure was
therefore strongly affected by the political web which Ta'as was a part of. Upon
state lessening its grasp on Ta'as later on, other kinds of structures could have been
nurtured and capital could be drawn from them. Indeed, this was a prerequisite for
survival and solvency. The state ceased to be highly dependent on the firm, as the
global market emerged, and the sense of closure of politicians, public officials and
CEOs has been also damaged because of the eroding symbolic importance of the
security sector (Talmud and Yanovitzky 1998). Nonetheless, the firm was unable to
embark on developing other forms of compensating corporate capital (neither
competitive nor political), partially because of the transformation of its cognitive
corporate capital.

Cognitive Corporate Capital: From Corporate Superiority to Corporate

In its formative period, Ta'as enjoyed a relatively safe, prestigious and certain
environment with a comparatively assured clientele (Evron 1992). The organization
was composed of relatively stable intra-organizational networks, producing
consensus regarding the corporate mission and working procedures. The firm's
managers were production-oriented, lacking market orientation and marketing
skills. \3 The firm, leveraged by its political corporate capital, emphasized production
capacity and product quality rather than marketability. The formal and informal
networks transmitted this shared language throughout the organization, and the kind
of novel knowledge developed inside the firm in the 'heroic period' (especially
regarding ammunition), reflected this socially accepted value. The fact that the
corporate vision was to assist the Israeli Defense Force regardless of cost brought
about a shared sense of commitment to a special 'community-organization.'
Ta'as consisted of four basic structural elements deemed necessary for the
creation of its cognitive corporate capital: 1) shared symbols and narratives, 2) a
sense of closure, being a part of the 'security sector,' collective history, 3) repeated
interactions between experts, and 4) interdependence between customers and experts
114 - Corporate Social Capital and Liability

(Nahapiet and Ghoshal 1998; Talmud and Yanovitzky 1998). Consequently, many
risky innovations, devised inside the firm by engineers and technicians-some of
them lacking formal training-were made possible by mitigating socially-connected
trust within the organization using technical experts' commitment and tacit know-
how (see also Kramer, Brewer, and Hanna 1996).
By contrast, in the post-1985 'competitive era,' Ta'as has attempted remedial
measures. Nevertheless, these measures have been ineffective. As I will show, Ta'as
social structure~eemed necessary for creating its older cognitive corporate
capital-has become inappropriate for the new environment, thus resulting in
corporate liability.
Following the radical changes in environmental conditions, the firm undertook a
partially successful restructuring program including decentralization, the
establishment of relatively autonomous profit centers, implanting export orientation,
introduction of (failed) products' modification for the civil markets, layoffs, and
cutting administrative costs.
Moreover, the new organizational vision, imposed by new Chairs of the Board,
CEOs and General Managers, which emphasized market competition, stood in sharp
contrast to the old one, emphasizing priceless product quality, conveyed via existing
intra-organizational networks. The management did not trouble to pass on its new
vision via social networks since its image of the firm was formal and hierarchical.
Yet relations inside the rank and file were persistent, thus assuring a highly shared
resistance to change. In other words, the newly introduced uncertainty hampered the
'psychological contract' and the trustworthy social relations inside the firm,14 and
was not accompanied by management's proper awareness of informal processes
inside the corporate units. Furthermore, uncertainty and a lack of trust were
magnified by contradictory messages relayed from the government to the firm, and
by high turnover rates of staff and line managers, including the CEO's. This, in turn,
diminished social cohesion inside the organization, and restricted the ability of the
new managers to set up effective ties with unit managers and other influential
players. Consequently, this impaired Ta'as' capacity to implant new-fashioned
corporate identity, fresh strategic vision, and more important, trust in the corporate
headquarters (Talmud and Yanovitzky 1998).
The literature on corporate social capital underscores the fact that socially
embedded exchange may prevent agency problems and opportunistic behavior
which resulting in sub-optimal performance (cf. Podolny 1994; Jones, Hesterly, and
Borgatti 1997). Yet the lack of social linkage and trust between the new CEO's and
the unit managers created a dual agency problem, originated by relational
discontinuity inside the firm. This discontinuity, in turn, was created by a lack of
durability in identity-making ties (Podolny and Baron 1997), which is required
especially in times of crisis (Mishira 1996; Webb 1996). It was also found that both
the owner (the State) and the corporate directors were not receiving reliable
information on the business and financial operation of Ta'as from their respective
subordinates. As a result, they could not make relevant decisions and could not
install monitoring devices regarding the fate of various profit centers within the firm
(Talmud and Yanovitzky 1998). The corporate ship, consequently, began to drown.
Corporate Social Capital and Liability- 115

In a highly diversified corporation mitigating a dual-level crisis (at the industry

level and at the firm level), suffering from a lack of internal linkage and trust,
bureaucratic formal control is not as effective a device as that of corporate social
control transmitted via social networks. The new management ignored the veterans'
social networks. Layoffs did not solve the problem, as recruitment was virtually
non-existence. Thus the old, durable, identity-formation networks inside the firm
have created, in fact, corporate paralysis (Crozier 1960). In Ta'as, the 'old guards' of
the veteran employees and unit managers gave credence to norms such as
'effectiveness,' 'quality,' 'security,' while the new corporate managers were more
attuned to catchwords such as 'efficiency,' 'market testing,' and 'cost-benefit
analysis.' Hence, a lack of social linkage accompanied by a deficit in shared
narratives and corporate trust inside the firm has diminished the organizational
ability to create and foster cognitive corporate capital (Nahapiet and Ghoshal
1998).15 The erosion in the cultural position of Ta'as, marked by the shift from the
'heroic period' into the 'competitive era,' has eroded to a certain extent the
employee's self-image and sense of belonging to a close-knit 'special community,'
solely dedicated to Israel security's needs; a community inducing special codes of
conduct, whose members are trustworthy.16
Moreover, a lack of effective ties, and resulting lack of social trust, have
prevented the profit-centers' managers from taking risky decisions. Furthermore, the
importance of social networks in blocking corporate vitality is demonstrated by the
finding that even though individual managers' financial compensation was
moderately associated with their center's profitability, they lack the cognitive
corporate capital to facilitate such profitable action. They were unable to take
initiatives and risks due to the lack of trust within their corporate constituencies
(Talmud and Yanovitzky 1998). Consequently, informal social control prevented
managers from initiating business ventures and from excelling in their profit centers'
performance. Because the relations inside the firm were constructed in the pre-1985
period and were significantly sustained without corporate ability to modify them,
Ta'as lost its ability to create cognitive corporate capital. Moreover, the very social
structure which had created Ta'as' cognitive corporate capital and was beneficial to
its purposive action, become detrimental to the organization, impeding it from
attaining its corporate goals, turning a once-beneficial attribute into a corporate


The case of Ta'as corroborates Gargiulo and Benassi's conclusion (this volume) that
strong reliance on cohesive networks, which had been beneficiary to the firm,
prevented its executive officers from managing their corporate networks'
composition, and therefore compromising Ta'as' adaptability.
To the extent that a firm has structurally equivalent competitors in a market
niche (Burt and Talmud 1993), it needs to develop other compensating forms of
corporate assets to promote its comparative advantage. Burt et al. (1994)
demonstrate that for those corporations weakly situated in the exchange network,
'strong organizational culture' is moderately associated with profitability. One can
conclude, then, that where competitive corporate structure cannot be advantageous,
116 - Corporate Social Capital and Liability

imposing unmanageable constraints on the firm, it is especially useful for its

managers to develop other kinds of more manageable corporate social capital,
facilitated by other dimensions of social structure.
Nevertheless, investing in corporate social capital is not risk-free. Precisely
because corporate social capital is relation-specific, it is highly costly. The keyword
in minimizing these costs is balance. A firm must 1) balance its investment in three
modes of corporate social capital; and 2) carefully determine how to construct each
investment so its sunk cost would be more sensitive and flexible to changing
conditions. The case of Ta'as reveals that concentrated, over-excessive investment
in one kind of corporate social capital may later impede corporate goal attainment.
This inference is consistent with Uzzi's conclusion that 'a firm's structural location ...
can significantly blind it to the important effects of the larger network structure'
(Uzzi 1997a).
Political corporate capital and cognitive corporate capital may be, in principle,
more flexible, indeed even more elastic, and thus may seem to be more change-
prone, because they are not merely technologically-driven. In other words, they are
more affected by the social construction of the firm (Berger and Luckman 1967;
Cicourel 1973; Nahapiet and Ghoshal 1998). Yet the paradox is that the social
structure creating political and cognitive corporate capital is one of the most rigid
dimensions of the firm. Moreover, political and cognitive corporate capital are
costly because they are relation-specific and indivisible. The litrature on corporate
social capital clearly indicates that value is created through exchange and by
combining assets. This process takes both time and relations-specific investment
enabling business organizations to develop, nurture and exploit political and
cognitive corporate structures (Nahapiet and Ghoshal 1998: 257-259). Yet the
'sticky' nature of the social structures producing them induces the alternative
hypothesis that to the extent that a firm holds crucial investments in corporate
capital, its barriers-to-exit are costly as well. The above-mentioned case of Ta'as
indicates that sometimes these two rather neglected aspects of corporate social
capital can be highly rigid precisely because they are socially situated.


The growing number of managerial perspectives on knowledge-based firms, flexible
specialization and the high-technology sector have raised the awareness of all three
aspects of corporate social capital among managers, business scholars and
practitioners. I would hypothesize that firms investing in cognitive corporate capital
(embedded in intra-organizational relations) may be found among:
a. Mature technological business, having lengthy learning curve;
b. Service industries who focus on a market niche (Porter 1980), wishing to edge
out their competitors (Burt and Talmud 1993);
c. Potential entrants developing competence-destroying technology (Tushman and
Anderson 1986).

I also would hypothesize that 1) to the degree that the content of relations
revolves around identity formation, shared narratives and tacit norms, the relational
management is (after evolution) more rigid, and 2) to the extent that a network
Corporate Social Capital and LiabiJity- 117

circulates strategic options, entrepreneurial opportunities and material resources, its

relational management is more flexible. I would put forward the proposition,
accordingly, that one can order the degree of managerial flexibility of each form of
corporate capital as the follows: competitive corporate capital is the most flexible,
then political, and the last in order is cognitive corporate capital. Future research
should examine relational management and possible tensions between these three
forms of corporate social capital.

I would like to thank Paul Ritterband and the two volume editors for their valuable suggestions.

I. This definition is slightly different from those of Coleman's (1994: 170) and Nahapiet and Ghoshal
(1998: 243). In this chapter I use the term corporate social capital in general, and for simplicity, I refer to
each kind of corporate social capital as 'corporate capital' (i.e. 'cognitive corporate capital' means
'cognitive corporate social capital').
2. The term 'control' is used here to refer to what Alfred Marshall calls 'the external economy' of the
firm, including its relational management, and its 'internal organizational economy,' including governance
structure (see also Simon 1991).
3. I limit my discussion here only to kinds of corporate social capital-tangible or virtual-which
benefit corporate performance.
4. Competitive social capital is a network-oriented conceptualization of economic action (see Gabbay
1997: 17-20).
5. On the difficulty in shifting social capital see Gargiulo and Benassi (this volume).
6. My use of the term 'cognitive corporate capital' is akin to the 'cognitive dimension' of social capital
used by Nahapiet and Ghoshal (1998). It is a generic term, and I do not differentiate here between sub-
forms of cognitive corporate capital (intellectual corporate capital, cultural or symbolic corporate capital),
as all of these sub-forms are intangible assets of the firm, embodied by social relations (Barney 1991 ;
Nahapiet and Ghoshal 1998).
7. This is a structural form of social capital (Gabbay 1997: chapter I), which is akin to Coleman's
closure theory of social capital (1988), and to Lindenberg' s definition of 'self-command capital'
(Lindenberg 1993).
8. See, for example, Weick (1979). See Moran and Ghoshal (1996) and Nahapiet and Ghoshal (1998)
for its relevance to organizational advantage. A useful manifestation of the self-awareness of firm-based
knowledge is the emergence of ERP systems. The sociological version of rational choice terms it 'self-
command capital' (Lindenberg 1993), while economic theory labels it 'self control,' involving two
contradictory forces in the organization: farsighted planner and myopic doer (Thaler and Shefrin 1981).
Yet economic theory seldom treats its problem setting in terms of relational management, apart from
Williamson (1994).
9. On the relations between networks and shared narratives, see especially White (1992).
10. As I limit my summary here to the firm-level of analysis, I exclude the rich literature relating to
individual and market levels of analysis. I specifically treated these levels elsewhere (Talmud 1992, 1994;
Talmud and Mesch 1997; Izraeli and Talmud 1997, 1998; Talmud and Izraeli 1998).
11. For more on the conditional nature of social capital, see Han and Brieger (this volume).
12. A complete, in-depth description of the case study is in Talmud and Yanovitzky (1998). I include
here only a few examples drawn from the case in order to illuminate my analytical substance.
13. Their marketing was performed solely by an Armament Export Unit at the Ministry of Defense.
14. All Ta'as' employees were tenured and unionized. In 1997. Ta'as outsourced its computation center
to a British-owned company which hired Ta'as former employees.
15. On the relationship between personal relations inside the organization and the creation of cognitive
and even intellectual corporate capital, see Nahapiet and Ghoshal (1998).
16. On the importance of a sense of community to the creation of corporate capital see Nahapiet and
Ghoshal (1998: 258).
Dimensions of Corporate Social Capital:

Toward Models and Measures
Shin-Kap Han
Ronald L. Breiger

Despite an emerging consensus on the importance of corporate social capital, little
work has been done on the analytical problem of which aspects, precisely, of a
corporate network might be identified as manifesting the concept. Where in a
specific configuration of network ties is the corporate social capital located? Is
network capital a unitary phenomenon or are there various ways to conceptualize it?
In addressing these questions, we formulate models for corporate networks that
produce counts for the expected number of ties between each pair of actors on the
basis of sets of parameters which are themselves measures of network capital. The
model we prefer decomposes a network into separable dimensions comprising
status, volume, and proximity. We apply the models to a network of 'doing deals' in
which billions of dollars of finance capital was raised by syndicates of major U.S.
investment banks, data of Eccles and Crane (1988). We show that the model
performs well with respect to empirical validity. The modeling framework can be
applied and extended to other corporate network settings, and provides measures
appropriate for theoretical analyses of markets and corporate relations concep-
tualized as embedded within social fields.

Analysts as diverse as Coleman and Bourdieu put forward virtually identical
definitions of 'social capital' as denoting the resources for social attainment that
individuals acquire through networks of mutual acquaintance, obligation, and
information channeling. Bourdieu defines social capital as the sum of the resources,
actual or virtual, that accrue to an individual or a group by virtue of possessing a
Dimensions of Corporate Social Capital- 119

durable network of more or less institutionalized relationships of mutual

acquaintance and recognition (Bourdieu and Wacquant 1992: 119).
Coleman (1990: 302) too emphasizes that, unlike physical capital and human
capital, social capital 'inheres in the structure of relations between persons and
among persons.' Relevant here is the considerable body of cross-national research on
the impact of social networks and the transmission of job information on income and
occupational attainment (reviewed in Coleman 1990: 302; Lin 1990: 250-52; Burt
1992: 11-13), productivity (Bulder, Leeuw, and Flap 1996), and the organi-zational
side of job searches (Marsden and Campbell 1990). Studies such as these illuminate
concrete mechanisms by which individuals are linked to larger structures through
organizations and labor markets (see the review in Breiger 1995).
Much of the recent excitement generated by research on corporate networks
results from a focus on how the structure of their ties both affects and results from
the interests, resources, and positions of firms in the network-i.e., their social
capital. Galaskiewicz (1985) studies business philanthropy as relations among
corporations creating 'a grants economy.' Baker (1990) examines the interface
between corporations and investment banks with relation to power-dependence
concepts. Leifer (1990) seeks to explain authority and market relations among
organizations as embedded within the multiplex relations among key actors. White
(1992) emphasizes the mutual relations of corporate identity and intercorporate
structures and processes of control. Focusing on profit and a typology of markets,
Burt (1992: 82-114) shows how firms in production markets can use gaps in social
structure-'structural holes'-to their advantage in negotiating transactions with
suppliers and customers. Podolny (1993, 1994) theorizes and studies how status
orders arise from market and network relations among firms. Haunschild (1994)
investigates the effects of interorganizational relationships on the decision of how
much to pay when acquiring another company. Han (1994) shows how the interplay
between inclusion and exclusion among firms in a market yields a status dimension
affecting networks of imitation leading to isomorphism in the selection of audit
services. Many chapters in this volume also deal with the issue in a variety of


The new work on corporate social capital, such as the studies cited above, has
resulted in an explosion of new thinking and new knowledge about social networks.
Nonetheless, little work has been done on the analytical problem of which aspects,
precisely, of a corporate network might be identified as manifesting the concept.
Where in a specific configuration of network ties is the corporate social capital
located? Is network capital a unitary phenomenon or are there various ways to
conceptualize it? If multiple meanings exist, how might each of them be measured?
These are the questions concerning which we seek to contribute clarification and
To illustrate the scope of our contribution, including its limitations, consider the
network of co-management relations among the major investment banks in the u.s.
in the 1984-86 period. Eccles and Crane (1988) show that, in order to do their job
effectively, investment banks create a complex network of ties to other banks, in the
120 - Corporate Social Capital and Liability

form of syndicates containing 'lead manager banks' and 'co-manager banks' with a
separate syndicate organized around each specific 'deal' that is successfully put
together. Within the syndicates. lead managers and co-managers do the bulk of the
distribution. and work most closely together in the underwriting. In the aggregate
these 'deals' generated billions of dollars of financial capital in the mid-1980s in the
The data in our Table 1 are taken from Eccles and Crane's appendix (1988: 230-
31). Rows and columns list the major investment banks in the identical order. Rows
index each bank in its role as 'lead manager' of a 'deal' in the capital market. that is.
as a bank that works with a group of co-managers to form a syndicate to underwrite
a security issue. A lead manager normally 'runs the books' (manages the
underwriting and determines distribution allocation) and is usually the investment
bank that originated the 'deal' (Eccles and Crane 1988: 237). Columns index the
same banks in their role of 'co-manager': banks that work with the lead manager and
often a group of other co-managers in the syndicate. These are the top investment
banks in the country.! Entries off the diagonal in the Table are frequency counts of
joint participation in deals. For example. during the study period Salomon Brothers
served as the lead bank in 161 deals in which First Boston was a co-manager (see
[row 1. column 2] of Table I). whereas First Boston served as lead manager in 118
deals in which Salomon Brothers was a co-manager in a syndicate ([row 2. column
1]). Entries on the diagonal report the number of times that each bank served as lead
manager without any other top bank serving as a co-manager (either because it was
sole lead manager or because no co-managers were among the nineteen top firms).
For example. during the study period Salomon Brothers led 609 deals without
participation of other banks listed in the Table (see [row 1. column 1] of Table l)?


Table 1 is a network of ties among corporations. These network ties are at once
collaborative and adversarial. The ties are adversarial in that a fixed number of
securities must be allocated among the members of a syndicate. and also due to the
struggle among banks for recognition and position. The ties are collaborative in that
the offering must be distributed to desirable investors in a timely manner (Eccles
and Crane 1988: 93-94). This world of 'doing deals' is not a Hobbesian war of all
against all. which necessitates external control. but rather an instance of what we
might term the Simmelian problem of order. Georg Simmel. a theorist of sociology's
classical period. identified within certain forms of conflict a 'fight of all for all'
entailing the internal social control of an intrinsically ordered interweaving of
relations based on 'the possibilities of gaining favor and connection' (1955: 62). The
Hobbesian problem of order. as it is described critically by Markovsky and Chaffee
(1995: 255; Macy and Flache 1995) portrays the structure as seeking to extract from
its components something that they would not otherwise provide. We believe that
recent work on solidarity. with a new focus on reachability and relative unity of a
structure (Markovsky and Lawler 1994; Markovsky and Chaffee 1995) and on
models of macro-structure (Breiger and Roberts 1997). enables contemporary
Table 1. Data of Eccles and Crane (1988, pp. 230-31) on ties among investment banks, 1984-1986

Co-Managers: I 2 3 4 5 6 7 8 9 10 Il 12 13 14 IS 16 17 18 19
Lead Managers:
1 Salomon Brothers 609 161 123 23 86 243 34 73 36 19 23 20 17 IS 7 4 4 7 18
2 First Boston 1I8 331 159 15 100 1lI 20 122 35 19 6 Il 27 2 7 2 4 2 3
3 Goldman, Sachs 98 64 441 Il 46 57 10 49 30 4 5 10 1 18 4 IS I 3 7
4 Drexel 9 12 7 699 17 21 7 1 17 5 5 12 23 IO 0 5 8 2 2
5 Shearson Lehman 78 99 75 8 309 59 17 28 IS 24 25 33 8 21 5 10 8 2 2
6 Merrill Lynch 123 98 47 16 63 249 15 32 24 27 17 Il 20 13 2 14 9 5 7
7 Paine Webber 28 18 6 8 30 19 420 2 10 8 Il 6 4 2 1 I 0 3 0
8 Morgan Stanley 93 65 57 32 21 56 6 92 28 12 16 IS 18 8 4 7 9 9 3 I::'
9 Kidder, Peabody 22 Il 22 16 13 31 3 7 322 7 17 19 2 4 3 I 7 2 0
10 Pro-Bache 2 7 2 2 5 3 I 0 3 274 4 4 5 5 0 0 4 4 0
II E.F. Hutton 4 2 4 5 2 I I 0 I 3 227 6 5 II 1 I 2 1 2 :;,
12 Smith Barney 9 2 3 3 8 4 7 4 4 I 9 120 6 9 0 3 2 0 0 o'"
13 Bear, Stearns 2 8 2 4 4 6 5 3 1 1 I 1 131 1 I 1 I I 1 (j
14 Dean Witter 3 9 6 3 8 7 3 0 6 5 12 9 5 86 2 0 4 1 5
15 Dillon, Read 3 9 4 0 3 2 0 I 2 0 2 2 I 2 86 I 2 0 0
16 Alex. Brown 3 2 2 4 3 3 I 2 5 0 2 1 0 0 0 80 1 3 3 ~
17 DU 1 1 0 1 2 0 2 2 0 1 7 0 0 0 0 1 83 0 0 til
18 L.F. Rothschild 2 0 0 0 6 I 1 0 I 0 1 2 2 4 0 3 2 71 0
19 Lazard Freres 1 3 5 0 4 7 1 0 3 3 1 0 1 0 1 0 1 1 14 ~
122 - Corporate Social Capital and Liability

researchers to address the alternative, Simmelian, problem of order: that 'the

structural membership of the individual in his group always means some mixture of
enforced limitation and personal freedom' (Simmel 1959: 47; for further exposition
see Breiger 1990). We seek formal models and quantitative measures that allow us
to tease apart the subtle intermixture of various kinds of organizing principles
(hierarchy, reciprocity, and so forth) that constitute the intercorporate network.
The technical infrastructure for our modeling project is already in place. Within
the class of multiplicative and related models for contingency tables of frequency
data, we focus on quasi-symmetry and its special cases and generalizations.
Maximum likelihood techniques for estimating parameters are well known and
widely available (see Sobel, Hout, Duncan 1985; Goodman 1984; Clogg and
Shihadeh 1994). Thus freed from the necessity of creating new models, we
concentrate instead on the call of Sobel, Hout, and Duncan (1985: 371) for
researchers to 'attach new meanings to already existing models of this type and to
generate new models for the square table.' Related work of ours includes Breiger and
Roberts (1998), Breiger and Ennis (1997), and Han and Breiger (1996); see note 6


To establish notation, we will use fij to refer to an observed count in Table 1. For
example, fl2 is the count in Table 1 at the intersection of row 1 and column 2; as we
noted earlier, this entry reports that, during the period studied by Eccles and Crane,
the first-listed bank, Salomon Brothers, was the lead manager in 161 deals for which
the second-listed bank, First Boston, served as a co-manager. The set of fij defines a
quantitative network of ties among these investment banks. We seek to model this
network so as to bring out aspects of social capital. We will use Fij to refer to the
expected count that we derive for cell (i,j) on the basis of a model. We assume the
table is square and of size gxg, with the same entities (in our case, a set of
investment banks) indexed by rows and columns.
We consider several models, all of which satisfy or are special cases of quasi-
symmetry, a model which we will explicate. Other kinds of models might also be
considered for these data, but we have found quasi-symmetry models to be
particularly helpful in studying corporate social capital (cf. Friedkin 1998). These
models are distinctive in that they decompose a network of counts into separable
aspects comprising status, volume, and proximity, as we will demonstrate. The
existing literature on corporate relations is rather vague on whether these aspects are
conceptually separate, and their meanings have often been blurred. In fact, they
might be strongly related to one another; however, this should be an empirical
question. We seek explicit definitions of these concepts as features of networks, thus
enabling us to jointly model and measure the concepts in relational terms.
Following the path-breaking work of Sobel, Hout, and Duncan (1985), we para-
meterize the quasi-symmetry model as follows:

Dimensions of Corporate Social Capital- 123

Terms on the right-hand side are parameters to be estimated. Several

stipulations are imposed. An <X is estimated for each of the g actors indexed in the
table, and the product of the <Xj is 1. It is understood that ~i = ~j if i =j. The o's are
symmetric, Oij = Oji. Maximum-likelihood procedures for estimation of the
parameters and expected cell frequencies under this log-linear model of quasi-
symmetry are well-known; interested readers are referred to Bishop et al. (1975),
Goodman and Clogg (1984), Agresti (1990), Clogg and Shihadeh (1994).
It will be useful to consider the pair of cells Fij and Fji . For example, Figure 1
indicates expected frequencies among three pairs of banks. 3 We see for example
that, according to the model, Goldman Sachs is the lead manager in an estimated
45.618 deals in which Shearson Lehman is a co-manager. In the opposite direction,
Shearson Lehman is lead manager in a larger number of deals, estimated at 59.245,
in which Goldman Sachs is co-manager.
The ratio between the two, then, indicates and measures the extent to which the
pairing between the two actors is asymmetric. In pairing Shearson Lehman (i) and
Goldman, Sachs (J), for example, it is the former who is more likely to be the lead
manager, as shown in the first panel of Figure 1 (59.245 versus 45.618). The ratio
between the two, R ij , then, is 1.299, showing that Shears on Lehman (i) is 1.299
times more likely to be the dominant partner (i.e., lead manager) vis-a-vis Goldman,
Sachs (j). In other words, Rij is a measure of the dominance of i over j.
We may form an entire g x g matrix, R, of such ratios, with
ij=T. m

Although the equation above is definitional of dominance, we now turn to a

result. The matrix of R is a function of one dimension: the <X parameters in our
model. As may be seen by substitution of equation 1 into this definition of R:
F. a.
R.. =-2..=_) (3)
ji ai
Each actor in the network thus has a unique value of alpha, and the asymmetry
in any pairing between two actors can be described in terms of the ratio between the
two alphas. With respect to the tie in network R from Shearson Lehman to
Goldman, Sachs, for example, the ratio of the alphas (shown in panel b of Figure 1)
is .667/.513, which equals 1.299, the value of Rij given previously in the text. The
vector of alphas forms a linear ordering and in this sense captures the dominance
hierarchy among all the actors in the network. 4
Furthermore, if we define a column vector a to have 1/ <Xj as its j-th element, we
can trace the relationship between our alpha and a conventional measure of
centrality in the social networks literature (Bonacich 1972, 1987; Wasserman and
Faust 1994).
The centrality of an actor is very often operationalized as the sum of an actor's
social connections, weighted by the centrality of the others to whom the focal actor
is tied. A natural implementation of this concept of centrality is that centrality is an
eigenvector of the matrix representation of the social network (see also the
applications of eigenvector centrality measures in studies of corporate interlocks
124 - Corporate Social Capital and Liability


45.618 ) \ 20.665

9.245 5.96~

Goldman, ~ Smith
( sa:hS ~ Ba~ney \

~ 3.860 ~

Panel a: Expected frequencies (Fij) from Model 3 in Table 2.

Shearson Lehman

Goldman, Sachs _ _ _ _.~ Smith Barney
(.667) 2.668 (1.778)

Panel b: Ties between firms (Rij = Fij I Fji). Values in parentheses are Clj.

Shearson Lehman

Goldman, Sachs Smith Barney

Panel c: Oij; see equation 5.

Figure 1. Modeling the network of investment banks: a three bank example

Dimensions of Corporate Social Capital- 125

such as those referenced in Mizruchi et al. 1986).

With respect to the network R defined above, we have

Ra=A.a (4)

with A. = g, the number of actors in the network. This equation establishes that the ex;
parameters of the quasi-symmetry model define an eigenvector of matrix Rand
thus, in our modeling context, represent the dimension of networks that is typically
captured by the family of centrality measures including prestige, status, and
This interpretation of the ex; parameters illustrates the most fundamental feature
of our modeling approach: We formulate models for the network that produce
expected cell counts for the number of ties between any two actors on the basis of
parameters (a., ~, 8) which themselves are measures of network capital.
The models and measures are duals to each other, as roles and positions are
duals to the structure. The models themselves (such as quasi-symmetry) are well-
known. Our distinctive contribution is to develop their relevance to the analysis of
network data on counts, such as the co-manager ties of Table 1. From a statistical
point of view, the fit of the models to observed networks may be assessed by means
of standard maximum-likelihood chi-square procedures, making them feasible to
apply to square tables of frequency data in many different substantive contexts. 6
In models of quasi-symmetry for network data, ~i is the average volume of ties
sent and received by actor i, controlling for the other parameters in the model. In the
present context, this parameter indexes a bank's total involvement in deals (whether
as a lead- or as a co-manager, without distinguishing between these two roles but
focusing only on the extensiveness of its ties). In this sense, ~i indexes the
(relational) volume of an actor.
The third set of parameters, the 8ij' measures how strongly or closely i and j are
related to each other, net of a. and ~. The raw expected count, F ij , is by postulation a
function of all three parameters and thus is not a proximity measure. The geometric
mean of Fij and Fji could be a proximity measure but it also suffers from a built-in
dependence on all three parameters. However, if we norm this product appropriately,
then the following is a measure of proximity between the two actors that is net of the
asymmetric status (a.) and relational volume (~) effects, as may be seen by
substitution of equation 1 (see also Sobel et al. 1985: 364):

(~X~)=5, (5)

Although it is usual to interpret the 8 ij parameters with reference to the social

process of reciprocity (e.g., Sobel et al. 1985), in the context of our corporate data
we prefer the more direct interpretation given by the equation above: 8 is a measure
of the social proximity of two actors. The relations themselves may not be
symmetric, whereas 0 is an average of their intensities, the degree to which two
investment banks are likely to encounter each other in a syndicate formed to put
together a deal. In Figure I, panel c reports the estimated 8 parameters (computed as
126 - Corporate Social Capital and Liability

in the equation above) for three illustrative banks. It is seen that Goldman Sachs and
Shearson Lehman are much closer to one another (on the basis of the average net
intensity of their relations) than either bank is to Smith Barney. The 0 parameters
give us a network of proximity coefficients. The oij might well be related to
differences in status (<Xj I Clj) and in volume (/3i I /3j), as they indeed appear to be in
the illustrative example of Figure 1. Such relations among the parameters can be
investigated on the basis of a model that decomposes the expected cell counts into
just these three components.
A variety of models simpler than quasi-symmetry may be postulated; see Table
2. The simplest model we consider consists solely of the /3 parameters for volume;
that is to say, the model imposes that all <Xj = 1 and all Oij = 1. This model is taken as
the baseline model for analysis of square tables by Hope (1982), who terms it the
'halfway' model, and it is discussed by Goodman (1985) and by Hout, Duncan, and
Sobel (1987: 152), who note that the model imposes both independence and
marginal homogeneity.
The conventional model of statistical independence for rows and columns of a
square contingency table may be obtained from the 'halfway' model by adding to the
halfway model estimates of the <Xj parameters measuring dissimilarity among the
average counts in column j and in row j; see Table 2. In the independence model no
pairwise interactions are allowed (all oij 1). An alternative generalization of the
'halfway' model is to allow symmetric pairwise interactions (Oij) but to preserve
marginal homogeneity (<Xj = 1); this alternative defines the usual model of (full)
symmetry. Putting together, so to speak, the model of independence and the model
of full symmetry yields the model of quasi-symmetry that we have been discussing
in this chapter (see Table 2). In other words, we relax the marginal homogeneity
condition, specified in the full symmetry model (<Xj = I), while allowing the pairwise
interactions, prevented in the independence model (all Oij = I), with some
The remaining models of Table 2 provide more parsimonious representation of
the symmetric interaction parameters (the oij), portraying these pairwise terms as one
or more dimensions of interaction, rather than requiring one parameter for each pair
of actors. Notice for example that the 'homogeneous RC(I) model' requires
estimation of only (g-l) parameters more than the model of independence (the
difference in degrees of freedom is 305 - 287 = 18; see Model 3 in Table 2) in order
to represent all the pairwise proximities in terms of a single dimension (Jl). The
'RC(2)' model uses two dimensions in preference to estimating a parameter for each
pair of actors. All the models in Table 2 are well known and are discussed in detail
by Goodman (1984, 1985) and others (Sobel et al. 1985; Hout et al. 1987; Agresti
1990). As shown by the parameter specification in Table 2, all these models are
models of quasi-symmetry.
The fit of any of these models to data on frequency counts may be assessed in
the usual way by means of a comparison of the chi-square and the degrees of
freedom left by the model (see columns labeled G2 and df in Table 2). Or, one may
measure improvement in fit relative to a baseline model. The last column in the

Table 2. Models of quasi-symmetry *

Parameter Specification Clj Pi I)ij df G/ bie 1-{Gk2/Gl}


1 Halfway Pi 323 1624.4 -1099 0.0%

2 Independence Clj Pi 305 1248.0 -1324 23.2%

3 Homogeneous RC(I) Clj Pi exp{-~¢{u; -ll Y} j 287 779.1 -1641 52.0% ~.

4 Homogeneous RC(2) Clj Pi ex p{-.!. r.¢k{ut; -Ilkjr} 270 644.6 -1632 60.3% o......
2 k=l

5 Full Symmetry Pi I)ij 171 777.3 -664 52.1%

6 Full Quasi-Symmetry Clj Pi I)ij 153 330.9 -959 79.6% CIl
* As implemented in this paper, all models fit the diagonal cells exactly. Models are ordered by their degrees of freedom. a

128 - Corporate Social Capital and Liability

Table provides such a measure relative to Modell, the 'halfway' model.

Alternatively, because the number of deals indexed in Table 1 is large, and to avoid
overfitting (that is, the inclusion of terms in a model for which we have little or no
substantive interpretation) we may apply the Bayesian information criterion (BIC) of
Raftery (1986; see also Raftery 1995); values of this criterion are given in Table 2
for each of our models. In applying BIC one chooses the model with the lowest BIC
In fact, all the models of Table 2 are relatively more or less parsimonious efforts
to represent quasi-symmetry. For example, the oij values estimated from the one-
dimensional model of Table 2, RC(l), correlate .82 with the Oij parameters estimated
for the full quasi-symmetry model, and the single dimension estimated for the one-
dimensional model in Table 2 correlates .98 with the first dimension of the two-
dimensional model, RC(2).
According to the BIC criterion the best balance of parsimony and substance is
attained by the one-dimensional model, RC(I), with a single dimension
characterizing scores for both the row and (identically) the column categories of our
data (see Model 3 in Table 2). This one-dimensional model differs from (full) quasi-
symmetry only and precisely by modeling the pair-wise Oij parameters by means of a
single dimension of proximity, as specified by the l1i parameters in the column of
Table 2 labeled 'Oij'. The illustrative example of Figure 1 above is based on expected
counts and estimated parameters from this RC(I) model.


In this section, usil1g the model discussed earlier, we examine the social structure of
investment banking industry by analyzing the data on the ties between investment
banks. Our preferred model (Model 3 of Table 2) fits the data with three dimensions:
status (a), volume (~), and proximity (11). The estimated parameters are reported in
Table 3 for each of the nineteen investment banks in the dataset.
The model decomposes the relational structure among the investment banks into
three components. The first one, a, captures the disparity between playing the role
of the lead manager and the role of the co-manager. In other words, it measures the
propensity of an actor to be on one side of this asymmetric relationship versus the
other. In particular, the reciprocal of a measures the propensity to be the lead
manager in a deal rather than a co-manager, and indicates each actor's position in
the linear ordering of the dominance hierarchy among all the actors in the network.
The estimated a's closely reproduce the well-known 'bracket' structure in the
industry, the strong and elaborate status hierarchy among the investment banks,
typically shown in the 'tombstone' advertisements (Hayes 1979; Eccles and Crane
1988; Podolny 1993). All of the six special (or bulge) bracket firms-First Boston
(#2); Goldman, Sachs (#3); Morgan Stanley (#8); Salomon Brothers (#1); Merrill
Lynch (#6); and Shearson Lehman Brothers (#5)-are found at the top of the list of
estimated a's, with Paine Webber (#7) being an exception in the ordering. Drexel
(#4) and Dillon, Read (#15), follow closely after these six.
Dimensions of Corporate Social Capital- 129

Table 3. Estimated parameters for Model 3 of Table 2

ID Investment Bank Special Parameters
No. Name Bracket In(1/a) In(13) In(~)
I Salomon Brothers Yes 0.9099 3.6608 1.4346
2 First Boston Yes 0.8301 3.3161 1.7183
3 Goldman. Sachs Yes 0.4056 3.2473 1.4390
4 Drexel No 0.3369 3.4433 -0.5509
5 Shearson Lehman Yes 0.6670 3.2002 0.6637
6 Merri11 Lynch Yes 0.4041 2.9608 1.0601
7 Paine Webber No 0.5593 3.2998 0.1743
8 Morgan Stanley Yes 0.8607 2.6913 1.1855
9 Kidder. Peabody No 0.2063 2.9904 -0.0587
10 Pru-Bache Securities No -0.6952 2.4590 -0.4910
11 E.F. Hutton No -0.9994 2.2128 -1.5178
12 Smith Barney No -0.5757 2.1059 -1.0161
13 Bear. Stearns No -0.8767 1.9993 -0.4697
14 Dean Witter No -0.2102 2.1221 -1.2014
15 DiUon. Read No 0.3217 2.3880 0.3272
16 Alex. Brown No -0.3679 2.0071 -0.3828
17 DU No -1.2533 1.5828 -1.4986
18 L. F. Rothschild No -0.4190 1.9218 -0.9267
19 Lazard Freres No -0.1043 1.2674 0.1109

2.v 2.u

1.0' , i
; 1,

~5 i
j 4 a j
~5 i ~

~ 0.0' Yl 0.0 ~9
V:'" ~4 ~4
,... ~d.6 111 ~6
If If
3-1.0 ~o -1 .0 ~o

::: ~3 ~1 ~1
11 _2.n
-2.0 · 1.0 0.0 1.0 2.0 -2.0 -1.0 0.0 1.0 2.0

In(Beta). Standardized In(I1). Standardized

Figure 2a. Status (lIa) by Volume (13)· Figure 2b. Status (lIa) by Proximity (p).

• ID numbers in the Figure are keyed to the name of the investment banks in Tables I and 3.
Parameters are standardized by taking Z-scores.
130 - Corporate Social Capital and Liability

The second parameter, p, taps into the volume effect, the extensiveness of a
firm's involvement in deals either as the lead manager or the co-manager. Salomon
Brothers (#1) is on top, followed by Drexel (#4). These two parameters, a and p, are
highly correlated with each other. Eccles and Crane (1988) observed that a firm's
hierarchical position is based partly on its volume, and the volume of securities it
gets to underwrite and sell is based in turn on its hierarchical position.
Consider, however, the joint distribution of the two. The scatterplot in Figure 2a
shows a pattern of deviation from the expected association between the two sets of
parameters, lIa and p. Although most of the firms are along or near the regression
line describing the expected linear relationship between status and volume, there are
two groups of firms that fall far outside the expected range.

1. On the one hand, there is a group of firms-Morgan Stanley (#8);

Dillon, Read (#15); and Lazard Freres (#19)-known to be Estab-
lishment firms that enjoy high status relative to their volume (Podolny
1994; cf. Stuart in this volume). Morgan Stanley is known to be 'the
bluest of the blue-chip investment bankers' (Hayes 1971: 139; Eccles
and Crane 1988); Dillon, Read used to belong to the special bracket
(Hayes 1979); and Lazard Freres represents the old Wall Street
(Stewart 1991). This provides an example of how the inertia of the
status hierarchy (a) mitigates the raw influence of the market as
indexed by overall volume (P).

2. On the other hand, another group suffers low status despite high
volume. Drexel (#4) is the most prominent case among the latter group
of outliers, which also includes Prudential-Bache (#10) and E. F.
Hutton (#11). Drexel's reputation as an aggressive newcomer and its
strong association with Junk bonds' provide partial explanations for
this discrepancy,? with Eccles and Crane (1988: 115-16) noting in
some detail that 'during our project the firm was described to us ... in
the most unflattering ways' (see also Stewart 1991). This is an example
of how the hierarchy, defined and legitimized by the participants
themselves, acts as a conservative mobility barrier to firms trying to
shift their position.

Volume is important, and it is highly correlated with status, yet the two are
separate dimensions, and one does not necessarily translate into the other. The
contrast between Lazard Freres (#19) and Drexel (#4) clearly illustrates this point
(also see the more recent case of DU-Donaldson, Lufkin and Jenrette, #17-in
Doherty 1997). The model we propose is precisely suited to such a setting, for it
allows the two conceptually independent components to be separated out from each
other. Although the two are rather highly correlated in this case, for instance, the
underlying structural dimensions they tap are distinct. As shown in Table 4, it is p
that best captures the overall volume effect, as in total market share and number of
issues. By contrast, lIa does better at explaining the dimension that is associated
Dimensions of Corporate Social Capital- 131

Table 4. Correlations between I/a and ~ and other select variables *

Variable InOla) In(S)

Total Market Share, 1984-1986 .689 .857
Number of Corporate Securities Issued, 1986 .775 .843
Market Share in Investment Grade Bonds .687 .684
Market Share in Non-Investment Grade Bonds .274 .478
Status Score (Podolny 1993) .621 .535
* The sources for the other variables are Eccles and Crane (1988, Tables 5.4
and A.6) and Podolny (1993).

with prestige or status, an excellent example being the status scores computed by
Podolny (1993, 1994) applying a widely used measure of centrality to the placement
pattern of investment banks in the tombstone advertisements. Also revealing is the
way they correlate with the market shares of two different financial products,
investment grade bonds and non-investment grade bonds (see note 7). With the
former both a and ~ are correlated to the same degree, while for the latter, a high
risk product (Podolny 1994), the correlation with a is not statistically significant.
The third dimension, J.l, is about how close or distant the firms are from each
other net of a and ~. The more deals a pair of firms do together, the closer they are
to each other vis-a-vis the others. The vector of J.l'S thus forms a one-dimensional
space along which each firm can be located, and the proximity between any pair of
firms can be obtained from the distance between the two. The smaller the difference
between J.li and J.lj' the closer firm i and firm j are, and vice versa. For example,
Salomon Brothers (#1; J.li = 1.4346) and Merrill Lynch (#6; J.lj = 1.0601) are
relatively close to each other (J.l; - J.l:i = 0.3745), and they frequently serve as
partners. s In contrast, Morgan Stanley (#8; J.li = 1.1855) and Dean Witter (#14; J.lj =
-1.2024) are relatively far apart (J.l; - J.lj =2.3869), and they rarely do deals together.
Plotting relational proximity (J.l) against status (l/a) reveals a very important
social dynamic that occurs among the investment banks, that of status homophily.
The firms that are close to each other on J.l are likely to be near to each other on a as
well. The smaller (J.l; - J.lj), the smaller (<Xi - CX;). Or, to put it otherwise, the
investment banks that put together the deals tend to be status equals, which is the
central finding in Podolny (1993) based on somewhat different data for these same
This result also precisely matches what Eccles and Crane observed. The top six
firms, those in the special bracket, are clustered to the right in Figure 2b. They are
close to one another as a result of the security issues they do together. Yet they can
be broken down into two groups. The first one consists of Goldman, Sachs (#3),
Salomon Brothers (#1), and First Boston (#2). These three rely most heavily on
other special bracket firms as co-managers. The second group-Shearson Lehman
(#5), Merrill Lynch (#6), and Morgan Stanley (#8)-<10 so less often, bridging
instead to banks further down the status ordering. The relative location of the two
groups on J.l, i.e., the second group being closer to the rest of the banks to the left,
bears out Eccles and Crane's account. In particular, these authors report that 'a
partial explanation of the difference between the two groups is that a larger share of
132 - Corporate Social Capital and Liability

the deals led by the second group, particularly Merrill Lynch and Shearson Lehman,
were security issues in which regional firms were used as co-managers to get more
retail distribution' (Eccles and Crane 1988: 94).9


An emerging consensus on the importance of corporate social capital spans a broad
spectrum of theorists and researchers, as aptly illustrated in this volume. Although
the core ideas seem to be shared by many and have been shown to be operative in a
variety of settings, little work has been done to parse out analytically the mUltiple
and conceptually distinct aspects of social capital. In examining a network of 'doing
deals' constructed among the major U.S. investment banks (Eccles and Crane 1988)
from the analytic perspective of log-linear models, we rely on a set of well-
established techniques. Yet at the same time we heed the call to 'attach new
meanings' (Sobel et al. 1985: 371) to the models, and we expand their interpretive
horizon by means of our novel application.
We formulate models for the networks that produce counts for the expected
number of ties between each ordered pair of actors on the basis of a set of
parameters which themselves are measures of network capital. More specifically, the
model we prefer decomposes a network into separable dimensions comprising
status, volume, and proximity. We show that the model and its associated estimated
parameters, which situate the actors within a social space in relation to one another,
performs well with respect to empirical validity.
'Something about the structure of the player's network and the location of the
player's contacts in the social structure of the arena,' Burt argues, 'create a
competitive advantage in getting higher rates of return on investment' (1992: 57).
That 'something' is social capital, upon which economic constraints and
opportunities that confront a producer are very much contingent (Podolny 1993).
The approach proposed in this chapter makes it possible to formulate this intuitive
notion in a more explicit and structural manner. Providing measures appropriate to
analyze the intercorporate relations, competitive or otherwise, and a way to map the
configuration of the market as a whole, the model can easily be applied and
extended to other corporate network settings, such as joint ventures, strategic
alliances, or R&D collaboration (e.g., Freeman, Pennings and Lee, Smith-Doerr et
aI., and Stuart in this volume). The model thus should allow better informed
strategic deliberations in a variety of management settings (e.g., Porter 1980).
Furthermore, the model and its parameters offer a way to portray a social field
in general, beyond this specific framework. Consider, for instance, different
configurations of actors in <X-~ space, as in Figure 2a. For the investment banking
industry, there is a high correlation between a firm's volume of underwriting
business and its place in the industry'S hierarchy. The two, however, are
conceptually distinct. The model we propose is precisely suited to such a setting, for
it allows each dimension to be separated out from the other. It should be possible,
nonetheless, to find a case completely opposite to ours, one in which the two are
correlated negatively-as in luxury goods. Or, one might be able to find a setting in
which the mid-sized firms are the most prestigious (cf. White 1988, 1992). In <X-Il
space, shown in Figure 2b, we showed precisely a pattern of status homophily. And
Dimensions of Corporate Social Capital- 133

yet, other configurations are plausible as well. There are settings that exhibit social
proximity between status unequals, as for example between managers and
secretaries (Kanter 1977). Curvilinear relationships are also possible, as for example
among U.S. Supreme Court justices, where we have argued that those justices 'in
between' the coalitions of liberal and conservative ideologies are highest in the status
hierarchy (Han and Breiger 1996). Such an inverse-U shape is itself in contrast to
the U-shape for plotting status against social proximity that is often postulated in the
cases of the middleman minority (Bonacich 1973).
In sum, the theoretical and research work that is necessary to further develop the
concept of corporate social capital requires structural and systemlltic measures on
networks. In this chapter we have introduced relationally based measures which
appear to be most promising for future work.

For comments on an earlier draft we are grateful to the editors and to Robert Faulkner, Noah Friedkin,
Joseph Galaskiewicz, Philippa Pattison, and John M. Roberts, Jr.

1. That these nineteen investment banks are the major players seems to be a robust finding. Eccles and
Crane (1988: 228) report that, 'since nineteen is not a round number, we attempted to chose a twentieth
but could not. All the candidates were firms strong in narrow product categories, purely regional firms, or
varied in the strength of their performance through the three-year period' of 1984-86.
2. Of these, 488 were deals in which Salomon Brothers was the sole lead manager, and 121 others
involved deals with banks other than those listed in Table I.
3. The expected frequenCies in Figure 1 are derived from Model 3 in Table 2 of this chapter. This
model is a special case of quasi-symmetry, to be discussed below.
4. This discussion of R;Jis somewhat analogous to related formulations, including those of R. D. Luce
(who developed implications of the choice axiom for the scaling of preferences) and of S. E. Fienberg and
K. Lamtz (who showed that the Luce model implies a linear preference model in the logit scale, and who
formulated aspects of Luce's model as well as other models for paired comparison experiments in terms
of the quasi-symmetry model). See Agresti (1990: 370-74) and the brief review that appears in Breiger
and Roberts (1998).
5. Equation 4 may be obtained by substituting equation I into the definition of R;J.
6. For example, Breiger and Roberts (1998) and Han and Breiger (1996) examine networks of joining
in one another's opinions on the part of members of the U.S. Supreme Court. Breiger and Ennis (1997)
examine as a social network a cultural field of relations among writers in KOln, Germany, data of Anheier
et al. (1995).
7. A 'junk bond' or noninvestment-grade bond is a certificate of debt promising a high rate of return on
investment but carrying a high risk; these securities are often used to finance corporate takeovers.
8. Eccles and Crane (1988: 96) note that 'the tie between Merrill Lynch and Salomon Brothers was
particularly strong.'
9. See Doherty (1997) for a detailed account of DU' s (#17) relative location vis-a-vis the others and
its recent movement in the parameter space.
Organizational Standing as
Corporate Social Capital

Patrick Doreian

Organizations in social service delivery networks interact in order to provide many
services to a wide variety of client populations. In the course of these interactions,
staff and directors of these agencies form assessments of the utility of working with
other agencies. These assessments, as social network information, can be used to
operationalize a measure of network generated corporate social capital.
Organizations well regarded by other organizations have higher social capital.
Further, organizations well regarded by well regarded other organizations have
higher social capital. Input-output methods are used to generate measures of
standing. These assessments are also disaggregated by sector in a way that permits a
comparison of the relative contributions to social capital by sector. Data from the
SSDURC project are used to illustrate these methods.

Social service organizations are distributed across a variety of sectors. Scott and
Meyer (1991: 117) define a sector as '1) a collection of organizations operating in
the same domain, as identified by the similarity of their services, products or
functions, 2) together with those organizations that critically influence the
performance of the focal organization.' This seems too inclusive and it is useful to
modify the first characteristic to focus exclusively on the primary functions served
by organizations in their sectors. Two organizations in the same sector need not have
the same services and products: a residential facility providing community living for
the mentally ill will differ from a multifunction agency and a psychiatric ward in a
hospital will differ from both.
Organizational Standing as Corporate Social Capital - 135

The sectors considered here are defined in simple functional terms. Units in the
education sector educate students (primarily in schools); units in the health sector
provide services to promote good health and to treat ill-health; units in the judicial
sector process perpetrators and victims; mental health agencies provide services to
deal with people with mental illnesses andlor those who are developmentally
disabled while units in the poverty sector attempt to reduce poverty and lessen the
adverse consequences of being in poverty.
Scott and Meyer (1991: 120) argue that 'the concept of a societal sector suggests
the presence of organizational systems that are, to some degree, functionally
differentiated.' The five sectors listed above can be viewed as differentiated systems.
Yet there are many overlaps and, with them, potential conflicts. If social service
agencies, having differing charters, mandates, philosophies and technologies, are
distributed across a variety sectors, there is limited consensus concerning the core
technologies used by these agencies. The presence of 'multi-problem clients' makes
this even more acute. This is particularly the case when there are poor clients with
health andlor mental health problems and who break the law or are victims of others
who break the law. Distinct agencies will have claims, with differing degrees of
legitimacy across sectors, concerning their organizational domains. Further, the
conditions are created where there can be both duplication of services and clients
who 'fall through the cracks.'
Social service organizations have to interact under ambiguous conditions in
order to provide services and coordinate their activities. It is clear that organizations
working together with, or fighting over, specific clients have impacts on each other.
Or, if a sequence of services is provided for clients in different agencies, there have
to be clear referral pathways linking the organizations. The second defining feature
of sectors, as proposed by Scott and Meyer, has limited utility. On one hand, if
agencies are linked in networks, they cannot help but have impacts on each other
and the condition is satisfied trivially. On the other hand, even if they belong to
different sectors they do have impacts on each other and the second feature is
contradicted. In the following, Scott and Meyer's second criterion for defining a
sector is discarded here.
Regardless of these distinctions, organizations still have to work with each
other. This is not a straightforward process and takes place in a web of inter-
organizational relations (IORs). In part, these relations are generated by funding
streams and, in part, they are shaped by local interactions between organizations in
some geographical location.


Social service organizations are embedded in many lOR networks. It is a truism that
these networks provide access to needed resources for some organizations and that
they place constraints on the actions of other organizations in the network. Network
ties are created, changed, maintained or dissolved. In short, the lOR network (ION)
changes, and with this change, the opportunities and constraints shift also.
Organizations negotiate their way through their changing environments. As these
agencies interact, histories of interactions are generated across the network. Some
working arrangements are successful while others fail. Organizations build histories
136 - Corporate Social Capital and Liability

of which ties work for them and which ties do not work. As personnel in these
agencies come to form assessments of other organizations, these other organizations
get reputations. Shrum and Wuthnow (1988) argue that the reputational status, in
part, is driven by past performance.! Such performances-and the shared histories
they create-form foundations for the reputations that agencies acquire. Agencies
acquiring positive reputations acquire also social capital while those acquiring less
positive, or even negative, reputations lose social capital or build up social liability.
Given sector memberships and uses of different treatment procedures, the worth
of different technologies becomes ambiguous. There is limited agreement on what
constitutes an 'effective' treatment modality and assessing the merits of technologies,
together with the organizations employing them, becomes subjective. Knowing the
reputation of another organization makes this evaluation simpler. Consider an
organization and suppose, for whatever reason(s), there is a need to interact with one
or more other organizations. Some criteria are needed to judge if future joint action
with another unit is worth pursuing.
One is suggested by institutional arguments where procedures and forms come
to be accepted as 'legitimate' in some sense (Meyer and Rowan 1991). Part of this
stems from which agencies and programs are funded. Getting funded is a profound
legitimating force, assuming the funder is reputable. These funds have three primary
sources: 1) extra-local governmental (Federal, State and County level) funding
sources; 2) legitimate charities, and 3) agencies purchasing services from one
In the mobilization of relations for the delivery of services, agencies build social
capital through the quality of the relations that are established and used. Substituting
'agency' for 'person,' Coleman's (1990) insight that social capital is generated
through the 'structure of relations between agencies and among agencies' is
pertinent. Bourdieu and Wacquant define social capital as 'the sum of the resources,
actual or virtual, that accrue to an individual or a group by virtue of possessing a
durable network of more or less institutionalized relations .....(emphasis added)'
(Bourdieu and Wacquant 1992: 119). Han and Breiger (this volume) also use this
idea as a point of departure for establishing measures of social capital. Burt (1992:
8) views social capital of an actor as the relations the actor has with (relevant)
others. Lin (1982) regards an actor's relationships as resources for instrumental
action. Coupling these ideas, an organization's network generated social capital
comes from the number of ties with other organizations and the evaluated quality of
those ties (where the evaluations are made by the relational partners). Further, the
network generated social capital will be higher if the favorable evaluations come
from highly regarded organizations. The proposal here is to use the reputational
standing of an organization as an indicator of its (network generated) social capital.


The foundations for doing this in a social network context can be traced back to
Hubbell (1965) who used input-output models to generate a measure of standing for
the nodes in a network. 2 Doreian (1985, 1987) applied these ideas to actors in small
groups and journals in citation networks. For this application, a more important
Organizational Standing as Corporate Social Capital - 137

development is the method provided by Salancic (1986) to disaggregate such

standing measures, a procedure of particular use in studying IONs.
Let C be a (gxg) matrix containing the evaluations of agencies (by directors
and/or staft) in a square matrix form. In C, cij is the evaluation of agency j by
(representatives) of i. Let Tbe a (gxl) vector containing all of the evaluation activity
where T=C'u+Cu and u is a unit (gxl) vector. Let diag(1) be a diagonal matrix
where T is in the main diagonal. With v a (gxg) matrix of 1's and D=diag(1)v, a
matrix of weights, W, is created by the elementwise division of C by D. Next, let Sj
be the standing of the jth agency, then with ei an exogenous status measure:


which in matrix form is

s=Ws+e (2)

leading to


where I is a (gxg) identity matrix. This measure is akin to the measure of status of
Podolny and Castellucci (this volume) and, as noted by Han and Breiger (this
volume), to eigenvector based measures of centrality.
The operation in equation (3) is used to generate a set of measures of standing
for each organization in the network. However, these organizations also belong to
sectors and it is useful to disaggregate these measures into contributions to standing,
as social capital, from sectors. In the context of a journal citation network, Salancik
(1986) has provided a method for doing this. Let M be a matrix with a column for
each sector and mik is defined as:

mir 1 if organization i is in sector k

o otherwise
The intrinsic value of an organization in the ION can be viewed as stemming from
the sector to which the organization belongs. Let ek be this contribution for sector k.
and let E=diag(ek) be a diagonal matrix dimensioned by the number of sectors.
Equation (3) is then replaced by:


and if each sector has the same intrinsic value, E is the identity matrix. The
measures of standing generated, through the use of equations (3) and (4)
operationalize the idea of network based social capital.
138 - Corporate Social Capital and Liability


The data used here come from the Social Service Delivery Under Resource
Constraints (SSDURC) study. For details see Doreian and Woodard (1992),
Woodard and Doreian (1994) and Woodard (1995). An expanding selection
procedure was used to identify all of the relevant agencies in three Pennsylvanian
rural counties. We use the Fall County network from the first wave of the study. An
expanding selection procedure, using both director and staff respondents, was used
to identify all organizations in the network (Doreian and Woodard 1992). The data
used here come from the director responses to the following question 'To what
extent do you feel your relationship with this organization is productive?' with
response characteristics: 1, no extent; 2, little extent; 3, some extent; 4, considerable
extent and 5, great extent.
A network with sixty five (65) agencies in a social service delivery network is
considered here. These agencies are located in a variety of sectors. The education
sector is represented by the Intermediate Unit (lUI) providing special educational
services and five schools (AGSCH, BSCH, CNSCH, LHSCH, USCH). From the
health sector there are two hospitals (BVILL, UHOSP), a provider of human
services (GALL), Easter Seals (ES), Family Planning (FP), the local office of the
State Department of Health (PH) and the Association for the blind (BLIND).
The judicial sector has 20 agencies in the network. The core agency is Children
and Youth Services (CYS). There are four local police departments (CPOLC,
GPOLC, MPOLC, UPOLC) and the State Police (SPOLC). Two agencies, the
Women's Resource Center (WRC) and the Family Abuse Council (FAC), provide
services, usually for women and children, to deal with domestic and marital
problems. Present also are many units of the court system-the court office
(COURT), Domestic Relations (DR), the District Attorney Office (DA), JUDGEs,
Legal Aid (LAID), magistrates (MAGRT), the Mental Health Review Officer
(MHRO), the Public Defender's Office (PDEF) and the Victim Witness Program
(VICTW). Included also is the Juvenile Probation Office (JPO), the Drug and
Alcohol Office (DnA) and a residential treatment home (ARTH).
The mental health sector is dominated by the Mental HealthlMental Retardation
Office (MHlMR) and the Community Mental Health Center (CMHC). The CMHC
is the largest mental health agency, with a large array of programs. The MHlMR
Office does not provide direct services but is the largest conduit of mental health
funds into the local network of mental health organizations. There a second (and
geographically peripheral on the county boundary) community mental health center
(DHSI) and a psychiatric hospital (HPSYC). The Base Service Unit (BSU) provides
intake and case management services for the MHIMR. Two units providing
community living arrangements for the developmentally disabled (CLRI and CLR2)
are present. CMHC runs two partial hospitalization programs: STEP is for 8-13 year
olds and PHASE is for teenagers. SRRMI provides social and residential
rehabilitation for mentally ill patients. THOME provides domestic counseling, foster
placement and adolescent pregnancy foster placement services. The Fall County
Association for the Developmentally Disabled (CDC) provides a variety of services
and the Child Development Care Center (ECDC) provides daycare and preschool
services. 3 Finally, the Child and Adolescent Social Service Program (CASSP) was
Organizational Standing as Corporate Social Capital - 139

established to create coordination and referral services for agencies in different

Fourteen units are included from the poverty/social welfare sector. The
dominant agencies are the County Assistance Office (CAO) and the Community
Action Agency (COACT). Catholic Charities (CATH) and Community Ministries
(CMIN) are religion-based service agencies for the poor. The Housing Authority
(HAUTH) provides the poor with information and resources to obtain subsidized
housing they otherwise could not afford and the Food Bank (FOODB) provides food
for the poor. The Red Cross (RC) and the Salvation Army (SARM) provide
resources for people in poverty. The Women, Infants and Child (WIC) program
provides nutrition resources and screening to identify people who are
developmentally disabled. Social Security (SS) runs financial support programs and
provides Supplemental Security Income. Head Start (HEADS) provides educational
opportunities for the poor. Both the Private Industry Council (PIC) and the Jobs
Training Partnership Act (JTPA) provide job training, remedial counseling and
drop-out prevention programs. A transportation service unit (TRANS) provides
transportation for the poor.



o~ ______~====~ _________
Figure 1. Distribution of overall organizational standing

4 -r- MHMR



e h
m p
Figure 2. Side-by-side boxplot of standing by sector
140 - Corporate Social Capital and Liability

Table 1. Measures of standing for all agencies in network

Rank Agency Sector Standing Rank Agency Sector Standing
1 CYC j 5.010 34 UWAY f 0.501
2 CMHC m 4.040 35 HAUTH P 0.489
3 DnA j 3.896 36 CMIN p 0.477
4 MHMR m 3.651 37 PIC P 0.475
5 lUI e 2.771 38 PDEF j 0.462
6 CAO p 2.510 39 VICTW j 0.455
7 COACT p 2.376 40 CLRI m 0.453
8 WRC j 2.343 41 BLIND h 0.424
9 JPO j 2.214 42 JUDGE j 0.420
10 COURT j 2.189 43 CASSP m 0.395
11 FAC j 2.059 44 FP h 0 .381
12 SPOLC j 1.739 45 ARTH j 0.380
13 CDC m 1.253 46 PHASE m 0 .339
14 HPSYC m 1.253 47 DREL j 0.306
15 DA j 1.190 48 RC P 0.272
16 MAGRT j 1.142 49 THOME m 0.268
17 UHOSP h 1.078 50 WIC P 0.250
18 CNSCH e 0 .936 51 CATH P 0.242
19 CCOM f 0.832 52 BSU m 0.227
20 PH h 0.825 53 JSERV P 0.227
21 GALL h 0.780 54 MPOLC j 0.213
22 BVILL h 0.776 55 FOODB P 0.200
23 BSCH e 0 .762 56 UPOLC j 0.191
24 LAID j 0.734 57 ECOC m 0.182
25 YMCA 0 0.724 58 CLR2 m 0.166
26 USCH e 0.717 59 DHSI m 0 .160
27 ES h 0.716 60 GPOLC j 0.152
28 AGSCH e 0.670 61 MHRO j 0. 139
29 SRRMI m 0.650 62 CPOLC j 0.130
30 HEADS P 0.644 63 TRANS P 0.117
31 SARM P 0.569 64 STEP m 0 .116
32 SS P 0.540 65 CHCTR 0 0 .047
33 LHSCH e 0 .519


As argued above, the measure of standing can be used to operationalize corporate
social capital. These measures are given in Table 1 where the organizations are
ranked in terms of their stocks of social capital. These measures are displayed in
Figure 1 in the form of a boxplot. Figure 2 contains a side-by-side boxplot of these
measures by sector.
Organizational Standing as Corporate Social Capital- 141

Table 2. Contributions to standing in the education sector

Percentage Contribution by Sector
Agency Education Health Judicial Mental Health Poverty
lUI 17.2 11.7 8.4 37.0 15.9
AGSCH 2.5 17.0 7.5 Sl.8 19.2
BSCH 3.4 16.0 12.8 48.2 16.7
CONSH 3.0 13.8 32.5 25.9 15.9
LHSCH 12.6 4.0 17.1 38.5 24.1
USCH 4.7 3.3 40.0 28.0 20.3

There are eight agencies whose social capital makes them stand out as high
outliers in Figure 1. The overall dominant agency is CYS, a large multi-purpose
agency, which, despite the high variability of social capital among judicial sector
agencies, remains a high outlier in that sector. The Drug and Alcohol Office (DnA),
another agency from the judicial sector, is prominent also in Figure 1. The Rape
Crisis Center (WRC) is the third prominent agency from the judicial sector, one that
was particularly active in Fall County with a high profile director who sat on many
boards. Neither DnA nor WRC stands out when attention is confined social capital
measures of agencies in the judicial sector.
Both the CMHC and the MHlMR are prominent overall-consistent with their
domination of the mental health sector. Similarly, the County Assistance Office
(CAO) dominates the poverty-social welfare sector (with one of the largest budgets
state wide due to the high incidence of poverty in Fall County). The Community
Action Agency (COACT) is large also with many programs. The Intermediate Unit
(lUI) is prominent in the education sector. However, as argued below, its
prominence has less to do with dominating the education sector and more to do with
its linking role between sectors. There are no such prominent agencies from the
health sector. The two primary funding agencies (CCOM and UWAY) are
unremarkable with regard to social capital generated through the provision of
services. 4 Having CCOM well above UW AY reflects it being the larger funding
agency. From the view of service provision, it is surprising that the Fall County
Health Center is mentioned as it is the facility within which many agencies are
located. Having it rank last seems veridical as it provides no direct services.

Disaggregation of Corporate Social Capital by Sectors

The social capital measures are more interesting when they are disaggregated into
contributions coming from sectors. These are shown in Tables 2 through 6 and were
generated by the use of equation (4). To aid the interpretation of these contributions
the following labels are used. A contribution (from a sector) is labeled primary when
it contributes 50% or more of the total social capital of an organization. A
contribution is secondary if it is between 25% and 50% and tertiary if it is above
10% and below 25%. The primary contributions to social capital are bolded and the
secondary contributions are emphasized in the following tables. Attention is
confined to contributions from the education, health, judicial, mental health and
poverty/social welfare sectors. 5
142 - Corporate Social Capital and Liability

Table 3. Contributions to standing in the health sector

Percentage Contribution by Sector
Agency Education Health Judicial Mental Health Poverty
BLIND 24.3 5.5 5.5 37.7 21.9
BVILL 8.9 24.8 22.1 26.8 12.4
ES 10.2 8.0 6.8 32.7 36.0
FP 1.8 21.9 8.1 42.4 19.4
GALL 2.5 17.2 7.0 34.2 22.9
PH 10.5 29.5 5.9 13.4 24.4
UHOSP 4.3 12.7 18.5 35.8 24.7

The mental health sector, as a secondary contributor (37%), is the largest

contributor to the social capital of the lUI . The education, health and poverty sectors
are tertiary contributors. See Table 2. Mental health agencies were seen as useful by
the Intermediate Unit for providing their services. When the school districts are
considered, the mental health sector is a primary contributor (52%) to one district
(AGSCH) and a secondary contributor to all of the rest with contributions ranging
from 26% to 48%. This makes a lot of sense. Given the nature of this ION, the
schools are linked to other agencies through the treatment of 'unusual' students
(highly gifted, developmentally challenged and disruptive).6 The gifted and
developmentally disabled children are referred often to the Intermediate Unit 7 or to
mental health agencies . Hence the high contribution from the mental health sector
as agencies in that sector saw lUI as effective. Often, these children come from poor
households and poverty agencies are mobilized. The poverty sector is a tertiary
contributor to the social capital of all agencies in the education sector. While there
are referrals to judicial sector agencies, not all of these are seen as effective.
However, the judicial sector is a secondary contributor to the two largest schools
(USeH at 40% and CONSH at 32%). The location of the largest schools is the
location also of the largest police departments. The judicial sector is only a tertiary
contributor to another two schools. As judicial agencies process children and youth
as either victims or perpetrators they are seen as more useful when the disruptive
students are removed from the school system. That the education system does not
contribute much to the social capital of schools is not surprising. Their primary
interactions are with the lUI and agencies in other sectors and not with each other.
As shown in Table 3, no sector is a primary contributor to the social capital of
health sector agencies. The mental health sector is a secondary contributor for all but
one of the health agencies with contributions ranging from 27% (to BVILL) to 42%
(FP). It is a tertiary contributor to the remaining health unit (PH). The poverty sector
is a secondary contributor to one health unit ES at 36% reflecting the perception of
ES as an agency whose' members believed in charitable sevice. In this network, ES
worked with poverty ageffcies and was seen as having value in the provision of
subsidized service. This sector is a tertiary contributor for all of the remaining health
sector units. The health sector is a secondary contributor to the social capital of PH
and a tertiary contributor to another four units in its sector. The education sector is a
high tertiary contributor (25%) to BLIND which is consistent with the low
Organizational Standing as Corporate Social Capital - 143

Table 4. Contributions to standing in the judicial sector

Percentage Contribution by Sector
Agency Education Health Judicial Mental Health Poverty
ARTH 6.3 3.1 46.2 11.8 29.4
COURT 6.3 1.3 69.6 10.0 11.0
CPOLC 1.7 0.8 85.2 3.2 8.1
CYS 6.5 8.4 34.5 22.9 23.1
DnA 11.4 6.4 26.5 22.6 24.8
DREL 1.6 3.8 58.4 6.1 26.4
DA 2.4 2.0 82.8 5.8 5.6
FAC 2.0 2.6 55.6 11.4 20.4
GPOLC 0.0 0.0 100.0 0.0 0.0
JPO 14.5 2.6 52.6 16.7 11.4
JUDGE 5.8 3.5 57.4 17.2 13.6
LAID 3.2 6.5 53.9 5.8 28.4
MAGRT 15.9 1.8 62.0 10.1 8.4
MHRO 4.3 0.9 79.2 68 7.6
MPOLC 0.0 0.0 100.0 0.0 0.0
PDEF 4.0 1.0 84.0 5.1 4.9
SPOLC 12.2 1.7 67.1 11.8 5.8
UPOLC 7.7 3.6 63.2 12.4 10.3
Vlcrw 4.4 3.4 70.2 10.2 9.6
WRC 2.4 12.0 32.8 20.8 23.4

contribution of the education sector to the social capital of this agency. The relations
of education sector agencies to BLIND are educational services for the blind and
relations to the health sector are less consequential for BLIND. With the exception
of the hospitals (BVILL and UHOSP), the judicial sector does not contribute much
to the social capital of agencies in the health sector-a veridical result.
The primary feature of the social capital of agencies in the judicial sector is that
they are generated primarily in that sector. There are sixteen judicial sector units for
which the judicial sector is a primary contributor. See Table 4. These contributions
range from 53% to 100% (with the latter figure holding for two local police
departments). Many of these units can be viewed as being in the court system of the
sector: Public Defender Office (84%), District Attorney (83%), Legal Aid (53%),
Mental Health Review Officer (79%), the Court (70%), Magistrates (62%),
Domestic Relations (58%), Judges (57%), Legal Aid (56%) and the Juvenile
Probation Office (53%) all have most of their social capital generated within the
judicial sector. For the remaining four units of the sector, it is a secondary
contributor to their social capital with the contributions ranging from 26% to 46%.
This is an inward looking subsystem within the overall social service system. When
children and youth enter as perpetrators they tend to remain and judicial agencies are
useful for each other within the subsystem boundaries with most judicial agencies
lacking strong ties elsewhere.
144 - Corporate Social Capital and Liability

Table 5. Contributions to standing in the mental health sector

Percentage Contribution by Sector
Agency Education Health Judicial Mental Health Poverty
BSU 0.5 1.4 2.7 85.2 5.5
CASSP 16.1 4.8 12.1 54.7 8.6
CDC 7.6 10.6 3.9 49.4 24.2
CLRI 1.7 4.4 8.6 52.5 17.5
CLR2 2.5 6.4 12.4 17.1 29.6
CMHC 8.8 6.8 15.5 30.1 32.9
DHSI 6.6 4.9 55.5 19.0 9.6
ECOC 0.5 \.3 12.0 2.5 92.5
HPSYC 5.9 16.0 37.2 24.7 11.4
MHMR 5.2 12.9 31.8 18.6 20.5
PHASE 0.9 0.7 34.2 57.2 4.1
SRRMI 2.0 3.1 12.7 56.4 6.2
STEP O. 0.3 0.5 97.2 1.0
THOME 0.9 1.5 7.7 13.6 9.6

The poverty sector is a secondary contributor to the social capital of four

judicial agencies-ARTH (29%), DnA (25%), with CYS and the Rape Crisis Center
(WRC) at 23%. These are exactly the four agencies for which the judicial is 'only' a
secondary contributor to their social capital with contributions at 46% (ARTH), 34%
(CYS ), 33% (WRC) and 26% (DnA). All four of these agencies have significant
ties outside the judicial sector. The mental health sector is a tertiary contributor for
CYS, WRC and DnA. These are exactly the three high social capital agencies in the
judicial sector and while the judicial system overall is an inward looking system,
these three agencies have social capital generated from three sectors. The poverty
sector is a secondary source of social capital for Legal Aid (28%) and Domestic
Relations (26%) both units that provide legal services for the poor. The mental
health sector is a tertiary contributor for nine judicial agencies. With the education
sector being a tertiary contributor to only four judicial agencies and the health sector
being a tertiary contributor to only one unit in the judicial sector, it is clear that these
sectors are not a source of social capital for units in the judicial sector.
Just as there are agencies in the judicial for which social capital is generated
largely within their sector, there are corresponding (but fewer) agencies in the
mental health sector as shown in Table 5. They are STEP (97%), the BSU (85%),
PHASE (57%), SRRMI (54%), CASSP (55%) and CLRI (52%). The first three are
internal to the CMHC. Both SRRMI (mental illness) and CLRI (developmentally
disabled) provide residential treatment and are internal to the mental health sector. It
is not surprising, therefore, that most of their social capital is generated within the
sector. CASSP was a unit set up (with funds through the mental health sector) to
promote coordination of services between sectors. Having most of its social capital
generated within the mental health sector, it is not surprising that it had little success
in establishing between sector coordination. The mental health sector is a secondary
Organizational Standing as Corporate Social Capital - 145

Table 6. Contributions to standing in the poverty sector

Percentage Contribution by Sector
Agency Education Health Judicial Mental Health Poverty
CAO 2.3 9.5 26.0 14.1 36.4
CATH 0.5 0.9 7.5 6.9 38.4
CMIN 1.5 2.6 29.8 11.4 37.4
COACT 4.3 14.0 15.6 12.2 39.2
FOODB 1.8 6.7 7.0 6.0 72.0
HAUTH 3.0 20.3 7.2 5.4 60.8
HEADS 2.0 30.4 1.7 19.6 25.8
JSERV 1.6 5.2 13.8 11.0 62.5
PIC 2.8 4.9 10.4 15.8 61.0
RC 2.1 26.0 3.2 4.6 59.4
SARM 1.7 10.8 9.5 10.7 57.8
SS 2.3 6.4 17.9 39.8 28.7
TRANS 3.7 33.3 7.6 7.6 38.7
WIC 0.9 47.6 5.1 8.1 34.5

source of social capital for the CMHC (30%) and CDC (49%). It is a tertiary source
of social capital for another five agencies. This includes MHMR for which only 19%
of its capital is generated within the sector. However, the judicial sector is a
secondary source of 32% of its social capital and the poverty sector is a source of
another 33%. As with CYS in the judicial sector, MHMR has a significant amount
of its social capital generated in multiple sectors. The CMHC has 33% of its social
capital generated in the poverty sector. The only mental health agency for which
social capital is not generated within the mental health sector is ECDC. Virtually all
of its social capital (92%) was generated by its ties to the poverty sector. As noted
above, a strong case can be made for viewing this agency as part of the poverty
Six of the agencies in the poverty sector have most of their social capital
generated in that sector. See Table 6. The Food Bank at (72%) heads this list
followed, in order, by JSERV (62%), HAUTH (61%), PIC (61%), RC (59%) and
SARM (58%). The poverty sector is a secondary contributor to the social capital of
the remaining agencies in the sector. This suggests, rightly, that this is another
inward looking sector. However, only the education sector is not a contributor to the
social capital of poverty sector agencies. The health sector is a secondary contributor
to the social capital of four poverty agencies-WIC (48%), TRANS (33%), HEADS
(30%) and RC (26%) underscoring the interdependence of poverty and health
problems. This is particularly the case for the nutritional program (WIC) for women,
infants and children. TRANS provides transportation services for health provision
services for the poor. The health sector is a tertiary contributor for another three
poverty agencies. Mental health agencies are a secondary contributor for the social
capital of two agencies and a tertiary contributor to another seven, with
contributions ranging from 40% (SS) to 11%. Finally, the judicial sector is a
146 - Corporate Social Capital and Liability

secondary contributor to two poverty agencies (CMIN with 30% and CAO at 26%)
and a tertiary contributor for another four.

With organizations occupying niches located in inter-organizational networks, they
generate ties in order to coordinate their actions and deliver social services. The
services are located in different sectors and are directed at a variety of client pools.
As organizations have scarce resources, issues of how to allocate them arise. While
it is clear that some agencies cannot and do not work together, the range of
collaborative efforts is quite remarkable. Over time, collaborative (and, on occasion,
adversarial) relations are explored.8 Organizations come to acquire reputations.
(Indeed, we came to have a working hypothesis that agencies giving us trouble also
gave trouble to other agencies in these networks. It was confirmed.) The evaluations
they form of each other can be used to provide a measure of network generated
corporate social capital.
While this measure is useful and a veridical image of standing as social capital,
there are limitations to the materials presented here. One is the use of a unit vector e
for the overall measure of standing and E for the disaggregated measure. With
mUltiple waves for these networks, it is possible to use standing, s(t-l) at a prior year
as e for the estimation of social capital s(t) in the following year. In a similar
fashion, E(t-l) can be used as E to generate E(t). While it is not surprising that the
dominant agencies of sectors have high social capital, a next step will be to seek
predictors of high social capital other than size.9 Clearly, funding from extra-local
sources and the nature of the working relations will be relevant. A secondary gain
from using these measures of social capital is measurement of the relative
contributions to social capital from different sectors. The overall dominant agencies
have their social capital generated from multiple sectors. Examining the relative
contributions by sector permits images of where cracks form in the overall system.
Of particular importance is the idea that the judicial and poverty sectors are the most
inward looking sectors. It is clear that collaboration must involve agencies from
multiple sectors and, given the nature of these counties, even if the focus is on health
and mental health issues, both judicial and poverty agencies will be involved. This
will not be easy and a reasonable speCUlation is that these service delivery systems
will work best when the social capital of enough key agencies is generated from
multiple sectors. Exploring the generation of social capital in multiple sector
networks remains another important next task.

This work was supported by NIMH Research Award #ROI-MH44-1948. Comments by the editors and
Katherine L. Woodard on an earlier draft are much appreciated.

I. They argue that another two factors-organizational structure and network location-operate. In
their framework, all three factors can be used to 'explain' the variation reputational standing. As my focus
here is measuring standing, as social capital, such explanations are not the primary focus . Also, if I am
using the network structure, within which organizations are located, to define standing, network location
cannot be a non-redundant predictor of standing. Of course, this is my problem and not theirs.
Organizational Standing as Corporate Social Capital- 147

2. Viewing the structure of an economy as a social network, this lineage can be traced back further to
Leontief (1951)
3. As this agency is primarily funded by the County Assistance Office, a strong case can be made for
including it in the poverty sector. This receives support when the contributions to social capital by sector
is examined.
4. Again, there is an emphasis on the provision of services which tends to make those agencies
providing resources to fade into the background
5. While the funding agencies (CCOM and UWAY) and the other agencies (CHCfR and YMCA)
were used in the estimation of social capital, they are not used in these comparisons. Thus the row sum
for lUI in Table 2 is 90.2% leaving slightly less than 10% coming from the 'non-sectoral' agencies.
6. These categories are not all mutually exclusive
7. This accounts for the 17% as the highest contribution to the social capital from the education sector
to an agency in that sector.
8. Han and Breiger's (this volume) empirical example considers ties that are always both cooperative
and adversarial.
9. Podolny and Castellucci (this volume) propose a status measure that seems similar to the one
proposed here. Also, Han and Breiger (this volume) are persuasive in their decomposition of social
capital into (related) components.
Customer Service Dyads:
Diagnosing Empirical Buyer-Seller

Interactions along Gaming Profiles
in a Dyadic Parametric Space

Dawn Iacobucci

This chapter proposes a methodological approach to studying buyer-seller
interactions to understand corporate social capital in consumer services. The method
combines the theoretically strong game theory tradition with newer dyadic modeling
that should provide fruitful means of characterizing buyer-seller relations.

Managers and corporate leaders acknowledge that their satisfied customers are an
essential asset to their firms. Those customers can be other firms, in the roles of
supplier, distributor, partner, etc., or the customers can be the end-users, the
consumers. While many of the chapters in this book (in sections 3 and 4) focus on
the corporate social capital that resides in inter-firm relations, this particular chapter
focuses on the social capital inherent in the relationship of a firm with its consumers.
Within the class of purchases consumers can make, this chapter further focuses
on services industries (e.g., hotels, airlines, health maintenance, legal advice) rather
than manufactured goods (e.g., shampoo, toothpaste, blue jeans, etc.). One of the
primary distinctions between services and goods is that the former is comprised of a
greater interpersonal interaction between the buyer and seller (Iacobucci 1998): for
services, the customer engages in a transaction with the service provider, a frontline
representative of the service firm. By comparison, for goods, a customer selects the
desired packages from shelves, and deals rather minimally with a retail
representative, and not at all with a manufacturer representative. Thus, services
would seem to offer a better environment (than goods) in which to study corporate
Customer Service Dyads - 149

social capital at the consumer level (e.g., also see the chapters by Ferlie and Doreian,
this volume, who study networks of relationships in the health services setting).
Indeed, Reichheld and Sasser (1990) describe the importance of consumer
loyalty, drawing the analogy between the manufacturer's goal of 'zero defects' and a
service provider's goal of 'zero defections.' Heskett et al. (1994) take the
implications of loyalty further, demonstrating the long-term profitability of satisfied
customers; their life-time customer value. The long-term phenomenon of
relationship management (Baker and Faulkner 1991) is enhanced by positive
employee-customer relations. Strong positive interactions add value and contribute
to customer satisfaction (Crosby and Stephens 1987; Koelemeijer 1995; Price,
Arnould, and Tierney 1995) and future sales opportunities (Crosby, Evans, and
Cowles 1990). Thus it is clear that the consumer quite directly provides corporate
social capital, and that the process by which this occurs is at least in part social or
interpersonal, involving the interactions between the consumer and the provider.
Schlesinger and Heskett (1991) emphasize the interdependent nature of a firm's
employees and its customers. They illustrate the importance of a firm keeping its
employees happy and competent (i.e., well-paid and well-trained) so that they may
in turn keep the customers happy. For example, Marriott is known for relatively
satisfied employees (high retention rates) and customers (high room occupancies),
and the argument is that each drives the other. Given that customer and employee
satisfaction are positively correlated, the converse is also true; i.e., customer
dissatisfaction and poor customer service can also be attributed in part to
dysfunctional customer-employee interactions. The service provider's inability to
anticipate customer needs or recover from service failure (i.e., resolve the customer
complaint to the customer's satisfaction) is frequently seen as the source of customer
disgruntlement (Bitner et al. 1990; Brown and Swartz 1989; Solomon et al. 1985;
Surprenant and Solomon 1987). There are, of course, also other classes of
explanation for poor customer service, e.g., systemic industry problems, poor
organizational cultures, etc., but the dyadic exchange is of primary importance to the
present investigation.
In addition, we know that services purchases are typically characterized as being
more intangible, and accordingly the quality of the purchase is more difficult for the
customer to assess. Hence, customers rely upon cues to quality, including the
interpersonal nature of the service provider. Customers place credence in those
providers whom they trust, so the relational aspects that exist in the customer-
provider dyad are critical. Positive relationships, conveying trust, thus bring the
service provider the corporate social capital of drawing and keeping customers.
Researchers are coming to recognize that repeat purchases do not necessarily
imply loyalty. However, if the service organization provides quality and value and
satisfaction, perhaps with additional perks such as customization and friendly front-
line staffs, customers will be more inclined to return to these providers, given that
there are always some search and start-up costs in switching to alternatives. These
provisions, e.g., quality and value, are not monumental requests on the part of the
consumer-which is not to say that many service organizations provide such. And
certainly there are customers who are over-demanding, constituting social liability,
and firms are also coming to recognize that they do not necessarily wish to support
150 - Corporate Social Capital and Liability

the efforts of building relationships with all consumers, rather presumably those
segments projected to be more profitable.
We might also offer the caution that while firm-to-firm and consumer-to-
consumer relations are those between matched entities, there is a peculiar
asymmetry inherent to a firm-consumer relation (Iacobucci and Ostrom 1996), and
such imbalances in power tend not to favor (Hibbard and Iacobucci 1997) long-term
relational development. A dissatisfied customer may vow to never return to a
provider, but there are masses ready to take his or her place, thus usually voting with
one's purchases going elsewhere rarely feels like a satisfactory solution. On the
other hand, many firms are now talking about the relationships they wish to develop
with customers, e.g., through interactive marketing, data-base and direct marketing,
etc., yet inquiries to customers suggest that their frequent user accounts do not make
relationships. Clearly the relationship with a customer is precious, and firms that
manage customer interactions can enhance these ties, strengthening the social
bonding capitals with their target segments.
In sum, provider-customer relationships do not necessarily yield social capital.
Overdemanding customers create social liability, and service providers are better off
without them. In addition, providers can not take the maintenance of profitable
customer-relationships for granted and are starting to actively intensify these ties.
If we have established thus far that consumers contribute to corporate social
capital, e.g., in the form of their relational purchasing behavior, and that services
sectors may be an ideal business environment in which to study buyer-seller
interactions, we might also query as to strategies for studying these transactions. As
in studies of inter-firm relations, most researchers survey one side of the dyadic
relationship, primarily for ease of data collection. Critics of one-sided perspectives
would say the approach is suboptimal for studying the entire relational phenomenon,
and that dyadic methods are inherently superior: both points of view in a dyad must
be integrated to understand the individual actors in the relationship, as well as the
gestalt effect of the relationship itself. For example, Swartz and Brown (1989)
studied the interaction between medical professionals and their clients as dyadic:
they obtained both parties' perspectives on perceived quality and service encounter
expectations. Researchers studying household purchase decisions must also integrate
sometimes conflicting dyadic perspectives (e.g., Corfman and Lehmann 1987;
Davis, Hoch and Ragsdale 1986). Perhaps not surprisingly, Menon et al. (1995)
found that accuracy of proxy-reports improved with greater partner communication.
This chapter will provide a dyadic analysis of buyer-seller interaction. Two
methods will be used. First, we will draw from elementary game theory, a highly
evolved theoretical framework that is fundamentally dyadic in nature. Second, we
will explore the utility of dyadic interaction models that have been created within
the social network paradigm to model data that would arise in a customer-provider
engagement. We begin with a brief description of game theory. This introduction
may be skipped for the reader familiar with the area.


We should begin by acknowledging that the area of game theory comprises a huge
Customer Service Dyads - 151

Table 1. Diagram of outcomes for the prisoners' dilemma game

Suspect B:
Do Not Confess C Confess 0
Suspect A Do Not Confess C 3,3 1,4
Confess 0 4,1 2,2
Note: The first number in each pair corresponds to the outcome for suspect
A, the second number corresponds to the outcome for suspect B. Higher
numbers represent better outcomes.

literature, and developments are far more complex than we need for the current
research purposes. Many good references exist for additional information. Recent,
lucid introductions include Zagare (1984) and Van Lange et al. (1992).
This class of modeling refers to 'games' because two or more parties engage in
interactions, each player has goals to achieve, and they play by given rules-
together these properties characterize gaming. The simplest games require only two
players. Certainly researchers have extended games to more than two players, as in
studies of social dilemmas and coalition formation, but we can focus on the two-
party game, given that it most closely resembles the simple customer-employee
dyad. Even with only two players, Rapoport and Guyer (1966)' presented a
taxonomy of 78 classes of games, the most familiar of which is the 'prisoners'
dilemma game.'
The intuitive motivations of the players in that game follow (Dawes 1980: 182):
two prisoners have jointly committed a felony and have been apprehended by a District
Attorney who cannot prove their guilt. The D.A. holds them incommunicado and offers
each the chance to confess. If one confesses and the other doesn't, the one who confesses
will go free while the other will receive a maximum sentence. If both confess they will
both receive a moderate sentence, while if neither confesses both will receive a minimum
These outcomes are diagrammed in Table 1. One suspect's choices (i.e., to confess
or not) form the rows of a 2x2 matrix, the other's comprise the columns. The Table
contains the valuation of the outcomes for each party. For example, note that if A
confesses, and B does not, A receives the best outcome (a '4') and B the worst a ('I');
and vice versa if B confesses and A does not. If both confess, the outcomes for both
are worse than if both resist confessing.
As this game illustrates, each player chooses how to behave. For simplicity,
games may be restricted to decisions between two choices. Depending on the
research context, the choices take on different guises, but one choice can generally
be referred to as 'cooperative' (commonly labeled 'C') and the other as 'competitive'
(labeled 'D' for defect from cooperation). In this game, the prisoners who choose to
not confess are said to be cooperating with their partners.
Both parties obtain their outcomes as a function of the joint dyadic choices, thus
the players are interdependent because the actions taken by each affects the
outcomes of the other. Each player is expected to maximize his or her outcomes, but
note that in this game, there is a conflict of interest between maximizing one's own
outcomes versus maximizing the joint dyadic outcome; i.e., individualistic versus
collectivistic rationality. For both A and B, confessing yields better outcomes for
152 - Corporate Social Capital and Liability

themselves than not confessing, when considered independently of what the other
chooses to do. Thus, confession is said to be a dominant strategy for the individuals
involved. Unfortunately, if both players select their dominant strategy, a nonoptimal
collective outcome results, because both players could have done better as
individuals under a different choice (i.e., had they not confessed). Individual
rationality prescribes noncooperation, whereas collective rationality prescribes
cooperation (Van Lange et al. 1992).


In our context, cooperating may be defined as the customer and provider interacting
pleasantly and helpfully, whereas defection may represent a break-down in smooth
interpersonal transactions. More extensively, cooperation may be the provisions of
positive elements in the marketing exchange, e.g., providing good service at
reasonable prices perhaps with some customization on the part of the provider, and
repeat purchases, positive word-of-mouth, and creative input on the part of the
customer. Defection may be more broadly defined as well, e.g., providers who over-
charge, act impudent and uncaring, are unresponsive to customer requests, or
tempermental customers who take much of the providers' time and effort without
repeat purchasing, and so forth.
Given the superiority of the collective outcomes for mutual cooperation, and the
implications for social interactions and welfare more generally, much research has
been devoted to determining how cooperative behavior might be enhanced (Pruitt
and Kimmel 1977). Three strategies seem especially effective. First, the payoff
structure might be modified; i.e., the incentive to cooperate can be increased or the
incentive to not cooperate can be decreased. Second, if the parties are allowed to
communicate, they can promise to cooperate with each other (Misumi 1989). Third,
for games with multiple trials, each player can adopt the so-called 'tit-for-tat'
retaliation strategy whereby a player chooses 'C' or 'D' at time t to mimic the other's
choice at time t-I (Axelrod 1980). Thus, a cooperative choice is answered by
subsequent cooperation, and a competitive choice is reciprocated with competition.
This responsive strategy is thought to encourage cooperation because neither party is
taken advantage of, and neither party makes a first strike; accordingly, each player
shapes the other's behavior. The retaliation strategy requires that the dyadic
interactions iterate over multiple trials.
Table 2 contains several patterns of dyadic interactions that will illustrate
different gaming structures. The leftmost column depicts the 10 trials or time
periods over which the pairs of actors interact with each other, learning how best to
respond to the dyadic partner to optimize individual or collective outcomes. The
next three pairs of columns of numbers depict three patterns of dyadic interactions
that may be described as exemplar, in that they may not be likely in real data, but
they depict particular dyadic structures clearly, at the extreme. The first of these
patterns is the 'mutual cooperation' pattern in which both parties, actor i and partner
j, make the cooperative choice, defined in these data as a '0.' In contrast, the second
pattern illustrates 'mutual competition' in that both parties choose to continually
strike the other competitively, defined by the 'l's. The third dyadic pattern is also a
Customer Service Dyads - 153

Table 2. Dyadic interactions in gaming data

trial 3 exemplar strategies ~timal data interactions

mutual mutual chump retaliate random rand-comp
cooperation competition
ij ij ij ij ij ij

00 11 01 00 00 10
2 00 11 01 00 00 10
3 00 11 01 01 10 10
4 00 11 01 11 00 01
5 00 11 01 10 01 10
6 00 11 01 01 00 10
7 00 11 01 10 00 10
8 00 11 01 01 10 10
9 00 11 01 11 00 00
10 00 11 01 10 00 10

clear, but probabilistically unlikely extreme, that we may tenn, 'chump,' due to the
fact that one actor, i continues to choose to cooperate even though the other actor, j,
always responds with competition;j always takes advantage of i.
The fourth dyadic pattern is labeled 'optimal,' because it depicts the
aforementioned tit-for-tat retaliation strategy. For purposes of illustration, the
choices by j were chosen randomly (literally by the flip of a fair coin), and whatever
choice j selected at time t, actor i chose at time t+ 1.
The last two data patterns are termed 'data interactions.' The data in the pattern
designated 'random' were selected by subsequent coin tosses, resulting in a 50-50
chance for the 0-1 datum for each of the actor i and partner j at each of the ten trials.
The data in the pattern designated 'rand-comp' were selected similarly, except that
for actor i, there was a 25% chance of obtaining a '0' and a 75% chance of obtaining
a '1 '; i.e., it was a random pattern that favored the competitive choice.
The proposal in this chapter is as follows. We will model each of the three
exemplar patterns, and the optimal retaliation pattern, and thereby obtain their
parametric description via dyadic modeling. The two data interaction patterns will
then also be analyzed using the same modeling, and the resulting parameters will be
used to classify the data streams as resembling more or less one of the previous four
theoretical patterns. While the first four dyadic patterns are indeed exemplars and
therefore highly unlikely in real data, we can use them to anchor the parametric
space that can be used to characterize dyadic interactions in real data. As the
parameters describing the real dyadic interactions tend toward the ideal bounds of
the exemplar space, it will become spatially clear what sorts of structures underly
the real dyadic interactions. We turn now to the dyadic models.
154 - Corporate Social Capital and Liability


The dyadic models that will be useful in this chapter have been described elsewhere
in detail, so we are brief here (Iacobucci and Hopkins 1992; Iacobucci and
Wasserman 1988; Iacobucci and Wasserman 1987). Typically, the dyadic model:


is fit via the hierarchical log linear model:

U +Ul(i) +U2(j) +U3(k) +llt(1) +UI2(ij) +UI3(ik) +U24(jI) +U23(jk) +UI4(il) +U34(k1) (2)

to the four-dimensional y-array defined as Yijkl=1 for actor i sending relational ties of
strength k (i.e., 0 or I) to partner j and receiving at strength I (also 0 or 1 for these
data)? The a, ~, and p parameters are usually of particular interest to network
researchers. The first two parameters reflect actor-level behavior called
'expansiveness' and 'popularity,' tendencies for actors to send and receive relational
ties at strengths k and I, respectively. The p parameter is dyadic in nature, reflecting
mutuality or the extent to which relational ties are reciprocated. The a parameters
would represent an actor's tendency to act cooperatively, ~ would represent the
likelihood that a party typically elicits cooperation, and p would represent
tendencies for mutually cooperative interactions. We say more shortly about these
parameters in terms of which ones would expected to be statistically significant in
the presence of different gaming strategies.
For the strategies described by the data in Table 2, however, a modification of
the dyadic model (1) may be of greater interest. In real data, we would presumably
have replicate dyads, or even groups of dyads that differed on theoretically
interesting properties (e.g., games occuring with and without communication
between parties, say). For the purposes of our illustration, Table 2 contains only a
single dyad, for each of six patterns. Furthermore, the data in Table 2 occur over ten
trials, so model (1) must be modified to allow for longitudinal structural effects. In
particular, we may begin by examining adjacent time points, a time lag of two; i.e.,
collapse over the ten trials to obtain pairs of data for times t and t+ 1. We may
aggregate Table 2 over the individuals comprising the dyad and focus on the
relational structure (note that in the very act of aggregating, the model examines
relational structure with no regard to the particular identity of the actors, but rather
the typical interaction stream, e.g., for a class or segment of customers):

V kl ,II,k2,12 = :Ei:Ej Y i,j,kl,II,k2,12. (3)

We may now fit the model:

In P{YkI ,II ,k2,12=Ykl,l1,k2,12}= 91c1+911+9k2+912+Pkl,l1+Pk2,12

+'¥kI,12+fu,l1 +$kl,k2+$II,12 (4)
Customer Service Dyads - 155

via the log linear terms:

u +Ul(kl) +U2(11) +U3(k2) +U4(12) +UI2(kl.lI) +U34(k2.12)

+UI4(kl.12) +U23(k2.II) +UI3(kI.k2) +U24(1I.12)' (5)

The model (4) includes the main effects for the relational ties, the S's, which
essentially reflect the volume of 'l's (vs. 'O's). The p's are the reciprocal effects, as
per model (2), but now depicted for time 1 (or t) and time 2 (or 1+1).
The terms new to model (5) are the y's, which express a multivariate exchange:
what one actor does at time one (kl) determines with greater than random likelihood
what they receive from their partner at time two (12). Similarly, what the other actor
does at time one (11) determines what their partner does at time two (k2).
The cj>'s are also new terms. These parameters reflect multiplexity, the constancy
of behavior on the part of the actors. The term reflects an actor's autocorrelative
behavior; what an actor does at one point in time (e.g., kl) is related to what they do
at the following point in time (e.g., k2)'


Cooperation and Competition. We can now reexamine the six dyadic interactive
patterns in Table 2 with an eye toward these model parameters. The data patterns for
mutual cooperation and mutual competition appear similar in structure, in that both
actors' streams of relations to their partners are constant, either O's in the case of
cooperation or 1's in the case of competition. Accordingly, the parametric structure
for these two cases should also appear similar in the reciprocal structures. Both
patterns depict mutuality, albeit (1,I),s for competition and (0,0) for cooperation, so
we may test the proposition that for the cooperative and competitive interactions, the
reciprocity parameters should be statistically significant:
PI: both Pkl.1I and Pk2.12;tO.
On the other hand, the pattern for competition clearly depicts mutual strikes (1,1)
and that for cooperation is its mirror image (0,0), so the baseline volume parameters
should appear in the opposite direction (i.e., positive for competition and negative
for cooperation, given the data coding):
P1 : Skit SII, Sk2, and SI2 will be negative for cooperation, and positive for

Chump. The chump pattern should exhibit high degrees of autocorrelation because
both parties continue to do what they had previously done without having been
affected by their partner's past behavior:
P 3 : cj>kI.k2 and cj>1I.I2 will be statistically significant and positive.
In addition, the reciprocity parameters should be statistically significant, but
negative in direction to express the asymmetry of these dyadic interactions-all
dyads are of the class (0,1):
P 4: Pkl.1I and Pk2.12:;t0 and are negative.
156 - Corporate Social Capital and Liability

• Retaliation

• Competitive • Competitive
• Cooperative • Cooperative

p p
• Random
• Rand-Ccmp
• • Random
• Retaliation
.Chu np

Figure 1. Gaming strategies in 3-dimensional parametric space

Retaliation. The retaliation dyadic exchange should not result in statistically

significant reciprocity parameters, given thatj's behavior was determined randomly,
without consideration for i's behavior:
Ps: Pkl,lI and PU.12 = 0; i.e., not statistically significant.
And yet i's behavior at time t is perfectly predicted by knowingj' s behavior at time
t+ 1. Thus, we should expect to see statistically significant multivariate exchange
P,: {'Yk1.12} ={fu.lI} ;t O.
Random Interactive Behaviors. The final two dyadic interactions were the randomly
generated pattern, and that which was a cross between random and competitive
action patterns. For the wholly random data, we would expect all parameters to be
statistically non-significant, unless the modeling approach were hypersensitive to
erratic structures, or unless the particular random stream were odd. For the random-
competitive data, we would expect to see parameters between the near-zero random
parameters and those describing the competitive interactions. Notice too, that the
random-competitive data look fairly similar to a mirror image of the data for the
'chump' stream.
P,: All parameters should be near-zero for the random data.
P s: Parameters for the random-competition should moderately resemble those
for the competitive data.
Let us examine the results of the dyadic modeling and evaluate these propositions.

Figure 1 contains the results of the dyadic models applied to the data in Table 2,
presented in two plots. It represents a p-dimensional space, where p=3 is the number
of parameters being used simultaneously to create diagnostic profiles to characterize
interaction data. The first plot in Figure 1 represents the estimates Pkl ,I2 and 'YkI ,12,
and in the second, Pld.l2 is plotted against chl .u (1 and 2 represent times t and HI)?
The values plotted in Figure 1 represent the parameter estimates, though plots of the
fit statistics (i.e., essentially representing statistical significance values) confirm the
predictions regarding statistical significance. (Points far from the origin are
Customer Service Dyads - 157

statistically significant; e.g., competition and cooperation strategies did not yield
statistically significant yor $ values.)
Let us first focus on reciprocity parameter, Pld.lI, comprising the first dimension
in these plots. These parameters were predicted to be statistically significant for the
cooperative, competitive, and chump patterns (PI and P4), and indeed these are all
large. The cooperative and competitive estimates are positive indicating symmetry
(O,O's and 1,1 's), and the chump reciprocity is negative indicating the asymmetry of
the (O,I),s. The retaliation dyadic interaction was expected to exhibit a statistically
insignificant reciprocal pattern (Ps), and indeed, it is near zero.
One prediction was made about the multivariate exchange parameter, the 'YkI,12'S
and one about the mUltiplex autocorrelation, $t1.U' P6 predicted that for the
retaliation data, 'Yk1.12 should be statistically significant and positive, and indeed it is.
P3 predicted that for the chump data, $11.12 would be statistically significant and
positive, and it is.
For the random data patterns, P7 had predicted that all parameters would be
near-zero and indeed none of them are statistically significant-the point
representing the purely random data maintains a position near the origin in both
plots. The prediction for the random-competition data, Pg, stated that parameter
estimates should appear somewhere between the random and competitive data
patterns, and that they may resemble the opposite of the chump parameter results.
The latter characteristic appears to have dominated; evidently the asymmetry of the
random-competition dyadic interactions drove the model parameters more than the
symmetric-appearing competition element. Finally, the main effect parameters, 9k i>
911 , 9 u , and 912 are not included in the three-dimensional plot in Figure 1, but P2 had
stated that these effects should be negative for cooperation, and positive for
competition, and indeed they were.
In terms of a brief summary, we might offer a profile of the gaming strategies in
terms of the main parameters investigated: Pk\,\i> 'Yk1 .12, and $11.12. That is, rather than
examining the plots for distances between empirical profiles and the exemplar
gaming profiles, we might alternatively find just as diagnostic an exercise by which
we compare the parameters' profiles more discretely. For example, cooperation, as
per the plots in Figure 1, might be characterized qualitatively along the three-
dimensional parameter vector as: high, low, low. Competition would be
characterized similarly, except that a fourth parameter, the 9's would be required to
differentiate these patterns. The remaining profiles are also included in Table 3.

Table 3. Profile of Kaming strategies

Ptl,l1 "(kI,12 ~1.t2 Btl
Cooperation high (+) low low
Competition high (+) low low +
Chump high (-) high (-) high (+)
Retaliation low high (+) high (-)
Random low low low
Rand-Camp high (-) medium(-) low
158 - Corporate Social Capital and Liability

Table 3 also shows how the data pattern labeled 'random competition' fits none
of the exemplar row-profiles perfectly, presumably because it is more errorful, as
would be real data. Nevertheless, it is fairly straightforward to identify the closest
candidate case as the 'chump' profile for these random competition data.

This methodological approach, the combination of gaming strategies and dyadic
modeling, could be used somewhat like a discriminant analysis in that any number
of dyads may be modeled, and their resulting p-dimensional vector of parameter
estimates used to plot the dyad in space. The distance between the empirical dyads
and the exemplar gaming strategies may be interpretable like ideal point
characterizations, in that dyads lying nearer the retaliation extreme would
presumably be predicted to be more optimal in outcomes than dyads lying nearer
any of the other points. Dyads lying nearer cooperation than, say, chump or
randomness could be predicted to yield more social capital, extended longevity,
even greater profitability depending on the research context, etc. The pure data
patterns have been simulated and serve as functional anchors to the p-dimensional
space, and are therefore useful in describing the presumably more errorful data
It should certainly be noted that this demonstration is only a beginning; both
game theory and dyadic modeling are further developed than what is suggested by
the fundamental tools used in this chapter. However, this chapter is an initial
combination of a theoretically strong gaming tradition with a newer sophisticated
dyadic modeling heritage that should provide fruitful means of studying buyer-seller
From the models we presented, it turns out that some buyer-seller interaction
patterns lead to more intensive and long-lived cooperation than others. With the use
of empirical data, we can describe (and perhaps predict) which patterns are likely to
bring sufficient social capital to make the interaction worthwhile to both parties.
Finally, we might also suggest that while we presented this approach as ideal to
study buyer-seller interactions in a consumer services framework, the approach is
purely methodological and could certainly be applied to inter-firm relations as well.
The requisite data need only be dyadic and temporal, though surely these qualities in
data will be more easily obtained on persons rather than firms.

1. Cited in Zagare 1984.
2. If multiple dyads were to contain common buyers or sellers, the dyads would not be wholly
independent units, but researchers have demonstrated fair robustness of this class of models to such
circumstances (e.g., Frank and Strauss 1986; Strauss and Ikeda 1990).
3. For these simple data, the results for the reciprocity parameters at times 1 and 2 were similar in sign
and magnitude, so we use only the first set for these plots. This finding is not a general result, and the
plots could be extended to four-dimensions, or any p parameters the researcher wishes to model and
examine simultaneously, e.g., including higher order terms as say, 0%1.11.>2.

• Structure at the Individual Level

social capital in jobs and careers
The Sidekick Effect:
Mentoring Relationships and the

Development of Social Capital

Monica Higgins
Nitin Nohria

This chapter examines the benefits and pitfalls of mentoring relationships with
respect to a protege's ability to develop social capital, measured here as ties across
multinational subsidiary boundaries that might produce access to information and
resources. The results indicate that early mentoring relationships are negatively
related to a protege's stock of social capital and that later mentoring relationships
are positively related to a protege's stock of social capital. We call for a contingency
approach to studying how mentoring relationships affect career outcomes and
discuss implications for future research.


Mentoring is not a new concept. The word 'mentor' may be traced back to Homer's
Odyssey, in which a guardian named Mentor took on the role of adviser and teacher
to Odysseus' son Telemachus. Building on this ancient tradition, research on
mentoring has flourished of late. According to the International Center for
Mentoring in Vancouver, British Columbia, academic research on mentoring
includes more than 500 published articles, 225 conference papers, 150 doctoral
dissertations, 65 books and 150 mentoring program descriptions (Caruso 1992).
Most of this literature concludes that the effects of mentoring relationships are
quite positive. Mentoring has been shown to enhance career development (Kram
1985; Phillips-Jones 1982), career progress (Zey 1984), rates of promotion and total
compensation (Whitely, Dougherty, and Dreher 1991), and career satisfaction
(Fagenson 1989; Riley and Wrench 1985; Roche 1979). Given these generally
positive findings, much of the recent research has focused on two important
162 - Corporate Social Capital and Liability

questions. First, what are the major factors affecting the amount of mentoring a
protege actually receives and/or initiates? And, second, what types of help do these
mentoring relationships actually provide, or, put differently, what functions do
mentoring relationships serve?
In addressing the first question, scholars have uncovered important debates
around individual difference factors such as proteges' gender (Scandura and Ragins
1993), socioeconomic status (Ragins and Scandura 1994), and personality
characteristics (Turban and Dougherty 1994). In addressing the second question,
scholars have demonstrated that mentors provide career and psychosocial help and
other functions, such as protection or role-modeling (Turban and Dougherty 1994;
Scandura 1992). These questions and answers have brought us much closer to
understanding the antecedents and consequences of mentoring relationships. Yet we
still lack a good understanding of how these relationships actually work. As Turban
and Dougherty (1994: 699) put it, 'we now need to more closely examine how
mentoring influences career success.' For example, how do career mentoring
functions such as providing exposure and visibility enable proteges to obtain
We suggest and adopt two means of better understanding how mentoring
relationships have an impact on career progress. First, we suggest that, in order to
understand what makes mentoring relationships work, it is also important to
consider the conditions under which mentoring relationships might not work. To do
this, we consider both early and late mentoring experiences and generate hypotheses
regarding both when mentoring should work and when it should not work with
respect to individual career variables. Second, we suggest that, in order to develop
theory in this area, it is important to examine specific mechanisms through which
mentoring might facilitate career progress. To do this, we focus our research on one
career factor that has been empirically linked to upward mobility-the development
of beneficial social ties (social capital), or ties with others in an organization. PuIling
these two approaches together, then, we examine the benefits and the pitfalls of
career mentoring with respect to a protege's ability to develop social capital.
Researchers have already cautioned that there may be liabilities associated with
having a mentor (also see Brass and Labianca, this volume). Scholars have warned
that having a mentor creates the possibility for overdependence (Fagenson 1988;
Zey 1984), smothering (Kram 1985), anxiety that can turn to dependence (Clawson
1980), and clinging (Clutterbuck 1985). Still, most of the empirical work on
mentoring has focused on positive career outcomes. And, when negative outcomes
have been explored, the research has tended to focus on detrimental effects to the
mentor-protege dyad itself and not to relationships that extend beyond the mentoring
alliance (for an exception, see Fagenson 1994).
The career variable we study here, social capital, measures an individual's
potentially beneficial relationships beyond a particular dyad. We define social
capital here as the number of ties an individual has across multinational subsidiary
boundaries. Research on networking and on multinational organizations has
demonstrated that the volume of social relationships is related to upward progress in
individuals' jobs, higher paying positions, and career performance (Burt 1992; Lin
1982; Nohria and Ghoshal 1997). The volume of a protege's cross-subsidiary ties
The Sidekick Effect - 163

thus reflects social capital by providing the protege with access to information and
resources which are beneficial to his or her individual performance. Although
scholars of the mentoring literature have not directly tested the effect of mentoring
on this career variable, many have suggested that mentoring relationships may
enhance career progress because these relationships enhance the protege's ability to
develop ties with others. Specifically, researchers have espoused that career
mentoring enables a protege to overcome barriers of access to elite groups, to bypass
hierarchies (Fagenson 1994), and to have a 'special form of entry into important
social networks' (Dreher and Ash 1990: 540).
In this article, we empirically test the link between mentoring and social capital
by testing hypotheses regarding when mentoring should and should not work to
increase a protege' s stock of social capital. We do not take the blanket view that
mentoring is either generally beneficial or harmful, but rather argue that mentoring
may have its own time and place in helping an individual develop social capital.


Early Career Mentoring
In general, scholars concur that mentoring relationships provide at least two major
types of assistance: career help and psychosocial help. Career help includes
sponsorship, coaching, protection, exposure, and challenge, while psychosocial
functions include role modeling, counseling, acceptance, confirmation, and
friendship. During the 'initiation' stage of a mentoring relationship, usually during
the period from six months to a year on the job, mentors are said to perform career
helping functions for the protege, such as providing exposure, visibility, and high-
profile work assignments (Hill and Kamprath 1991; Krarn 1985). At the beginning
of the second (or 'cultivation') stage, a period of two to five years, both the mentor's
career development functions and psychosocial functions emerge-with the career
development functions emerging first. Thus, at the initial stages, the mentor is said
to provide the protege with a maximum level of 'challenging work, coaching,
exposure-and-visibility, protection, and/or sponsorship' (Krarn 1985: 53).
What are the short term benefits for proteges receiving early career help from
their mentors? Research and practical experience provide us with at least three
possible answers. First, career mentoring signals the value of the protege to the
organization and, hence, facilitates the newcomer's socialization process. Second,
career mentoring provides the protege with quick access to important information
and resources by lending the mentor' s social capital to the protege. Third, career
mentoring not only brings the protege 'to the racetrack' by providing access, but
actually places him or her 'on the starting blocks' by providing concrete
opportunities to display valuable skills and talents. We review each of these benefits
in turn, paying particular attention to how these factors affect the protege's stock of
social capital over the course of his or her career in an organization.

Mentors as Signaling Agents

Having a mentor early in a protege's organizational career I can signal to others that
the protege is a valuable player in the organization and, hence, can smooth his or her
socialization into the organization (Van Maanen and Schein 1979). Because
164 - Corporate Social Capital and Liability

organizational members can't fully assess the qualities of newcomers, they rely upon
information that reflects the protege's prior work experience, such as a resume or
recommendation, and they rely upon available signaling devices, such as who hired
the individual or who the individual's mentor is. Both actions taken on behalf of the
protege by the mentor and the protege's mere association with the mentor can
facilitate this socialization process. Specifically, providing exposure and visibility
by introducing the protege to high status others in the organization grants the
protege 'reflected power,' (Fagenson 1994; Kanter 1977), allowing the protege to
reap positive short term rewards in two ways. First, the protege's reputation may get
a boost as he or she rides on the mentor's coattails of success (Whitely, Dougherty,
and Dreher 1991). Second, people's first impressions of the protege may be
enhanced since others may assume that the mentor has, to some degree, selected this
person above others to be his or her protege.
However, beyond these immediate rewards, there are also liabilities associated
with developing one's reputation in an organization through an early association
with a mentor. Kram (1985: 26) describes such liabilities as 'the risk that others will
question the extent to which the younger individual can thrive on his or her own
without a particular senior's support.' The protege's reputation can become
inextricably linked to that of the mentor such that his or her competence, apart from
that touted by and associated with the mentor, is called into question. Therefore,
while short-term benefits may exist, the protege may, over the long run, be viewed
as the mentor's sidekick rather than as a worthy individual contributor with whom to
develop an independent relationship. Consequently, we expected that having a
mentor early in a protege's organizational career would be negatively associated
with the amount of social capital the protege was able to develop later on.
Mentors as Access-Providers
Having a mentor who provides career help early can also assist a protege by giving
him or her quick access to important information and resources. In effect, the mentor
lends the protege his or her social capital (Burt 1997) by introducing the protege to
others in the organization. This enables the protege to bypass traditional hierarchies
(Fagenson 1994) and barriers to information and enables the protege to tap more
directly needed resources. Therefore, not only do career functions such as exposure
and visibility grant the protege reflected power in the organization, they grant the
protege 'legitimacy' as well (Burt 1997)-Iegitimacy that is needed to get things
done or to get others to get things done for the protege.
In the short run, borrowing a mentor's social capital provides the protege with
the resources he or she needs to get specific tasks accomplished. It may also increase
the protege's returns of human capital (Burt 1997), since it positions the protege in
spots in which his or her knowledge can increase. However, there are long term
costs associated with borrowing someone else's social capital. Relying upon a
mentor for access to social networks may, over time, reduce one's own motivation
to develop instrumental relationships with others in the organization. As Fagenson
(1994: 56) warns, mentoring functions may 'prompt [proteges] to be less attentive to
and concerned about relationships external to the mentor-protege union ...
[suggesting] that proteges will have less favorable peer, superior, and department
The Sidekick Effect - 165

relationships than nonproteges.' Thus, we again expect that having a mentor early
will be negatively associated with a protege's social capital over the longer run.
Mentors as Opportunity-Providers
Having an early mentoring relationship can also benefit a protege in the short term
by giving a protege opportunities to prove him- or herself to others in the
organization. Providing high visibility work assignments is beneficial beyond
merely introducing a protege to important others; it gives a protege the chance to
demonstrate unique skills and competencies to those who might not yet know him or
her. The risk of providing such opportunities early is that, despite the mentor's best
intentions, the protege might not be ready for such assignments. Noe (1988a: 475)
has implored scholars to develop a 'readiness for mentoring' scale to ensure that
proteges who are selected for mentoring programs in organizations actually benefit
from their mentoring experiences.
We agree with Noe that advising is not necessarily advisable. There are indeed
long term risks to providing a protege with high-exposure work opportunities before
he or she has built sufficient individual-level competence and confidence at work.
From an outside-in perspective, organizational members might view this early
exposure as 'jumping the gun' or providing the protege with too much, too soon.
Jumping the gun could result in early career failure or, at a minimum, could result in
the perceived risk of failure, such that the protege has neither the capability nor the
referent power (French and Raven 1968) needed to develop substantial social
capital. From an inside-out perspective, providing career opportunities early, when
the protege has not yet developed a sense of self-efficacy or confidence in his or her
ability to perform (Bandura 1977) might backfire in building alliances with others.
The social support research suggests that people with high self-efficacy are better
able to obtain effective support from their established social networks than people
with low self-efficacy (Wills 1991). Without such confidence and competence, we
expected that, over time, proteges would be unable to maintain the social capital
they had borrowed from their mentors and/or that they would be unable to develop
substantial social capital on their own:
HI: Having a mentor early in a protege's career in an organization will be
negatively associated with his or her long term stock of social capital. 2

Later Career Mentoring

The aforementioned liabilities associated with receiving early career help from a
mentor do not apply to the protege who has spent more than a couple of years in an
organization-i.e., once the socialization process is over. By then, he or she should
have established some sort of reputation and identity in the organization so that there
should be substantially less risk that the mentor would smother the protege and,
hence, prevent him or her from developing ties with others. And, with his or her own
professional identity established, there should be less risk that others would view the
protege as his or her mentor' s sidekick. Rather, we expected that the mentor's role
as signaling agent would shift from socializing the protege to confirming the
protege's value to the organization. Now, more than before, the mentor's affiliation
with the protege should signal to others that the protege is a valuable organizational
member-that the protege has actually demonstrated potential. Therefore, the
166 - Corporate Social Capital and Liability

mentor should accentuate, rather than overshadow, the protege's accomplishments.

Hence, we expected later career mentoring to be positively associated with a
protege's stock of social capital.
Regarding access to important resources and information, earlier concerns
should subside once again with respect to later mentoring relationships. After
several years in the organization, the protege is no longer an organizational member
seeking legitimacy who needs to borrow another person's social capital in order to
gain organizational legitimacy. The protege should have his or her own ties with
others in the organization and, hence, should be attentive and motivated to develop
ties with several organizational members. Moreover, at this later stage in the
protege's career in the organization, the mentor may serve as a linchpin to facilitate
the connection of clusters of subnetworks of individuals in the organization. In this
way, the mentor's introductions should build upon or enhance the protege's own
stock of social capital. Therefore, we again expected a positive relationship between
having a mentor later in a protege's organizational career and having a large number
of intraorganizational ties.
Finally, we do not believe that later mentoring relationships that provide high
visibility work opportunities run the same sort of risk of jeopardizing the protege's
ability to develop social capital as is the case for the new protege. At this point, there
should be less risk that the protege would fall on his or her face by jumping the gun,
since he or she had already joined the race, so to speak. Although the stakes may
still be high for individuals later in their organizational careers, we presume that the
protege has amassed sufficient self-efficacy to benefit from potential helping
situations and to view them as challenges rather than as threats. Therefore, we
expected that the opportunities provided by a mentor to a protege later in his or her
organizational career would facilitate the development of social capital and, hence,
would be positively associated with a protege's stock of social capital:
H2: Having a mentor later in a protege's career will be positively associated with
his or her long term stock of social capital.

Respondents and Setting
Data were collected from 323 managers in the consumer electronics businesses of
three major multinational corporations (MNCs), Philips of Holland, Matsushita of
Japan, and Nippon Electronics Corporation (NEC) of Japan. Managers were either
department heads or general managers of subsidiaries in 24 countries in which all
three organizations had national subsidiaries. The subsidiaries in our sample were
selected in consultation with corporate managers of the MNCs and constituted, in
their view, a representative sample of all their subsidiaries. Questionnaires were
mailed to all departmental managers in each of these subsidiaries. The response rate
to the survey was 87% in Philips, 93% in Matsushita, and 75% in NEC, representing
an overall response rate of 83%. In no subsidiary did we get a response rate of less
than 60%.
The population from which our sample was drawn was a well-defined group of
upper middle-level managers in all the firms. Respondents had worked for an
average of 13.6 years in their current organization and had careers spanning more
The Sidekick Effect - 167

than 20 years across different organizations. Although we did not collect data on
their ages, we estimate that the median age of the managers in our sample was 40+
years. All the managers were men, reflecting the gender bias of the population
surveyed. There were no female senior subsidiary managers in the particular MNCs
we surveyed.
Because of the uneven distribution of responses across countries (there were
several small countries in which there were only a few subsidiary managers at the
level we had selected), we narrowed the sample to include respondents from only
the top ten countries (on the basis of the total number of respondents in all three
firms) for this analysis. After cases were eliminated because of missing data, the
final number of managers included in our analysis was 177. The composition of our
final sample was not statistically significantly3 different than the population
surveyed (based on t-tests comparing the means of all the variables included in our

Social Capital. An individual's social capital is generally determined by the
breadth or range of his or her social networks as well as the position that the
individual occupies in them. People are presumed to have greater social capital if
they have greater range (i.e., they have contacts that span the different clusters
within the network) and they occupy a more central position or are positioned to
take advantage of the structural holes in the network (Burt 1992; Nohria 1992b).
In this study, we operationalized social capital as the range or the number of
national subsidiaries of the firm in which the respondent had contacts. Range was
calculated as the number of different subsidiaries in which the respondent reported
having a contact with whom he communicated, regardless of the frequency.
Given that each national subsidiary represents a relatively separate network,
these contacts can be viewed as 'bridging' ties as they allow the individual to tap into
otherwise disconnected social clusters (Granovetter 1973). Indeed, Burt (1992) has
suggested that the number of bridging ties possessed by an individual may be one of
the best measures of social capital. Our focus on ties across subsidiaries was also
influenced by previous research on multinationals that shows that such ties are
crucial conduits for information and resource flows and have a positive impact on an
individual's performance (Nohria and Ghoshal 1997). Research by Brass (1984) and
Blau and Alba (1982) provides additional support to the notion that ties across
organizational subunits are important to individual career outcomes.
We recognize that-although it is an important measure of an individual's social
capital-range (as we have operationalized it here) is, nevertheless, a limited
measure of social capital. It does not fully capture the strength or weakness of a
person's network position. Ideally, we would have liked to measure the individual's
centrality as well as autonomy in various networks. However, that would have
required data on the ties among all the actors in the multinationals we studied-a
data collection effort that was impractical because of the very large number of
individuals who would have to be surveyed.
Mentoring Relationships. To measure whether respondents had early and later
mentoring relationships, we asked them if they had developed a close personal
168 - Corporate Social Capital and Liability

relationship with any senior manager in the company during their first two years of
employment with the organization and also if they had such an informal advisor or
sponsor at present (again, after working an average of 13.6 years at the company).
These questions were based upon Kram's (1985) work on mentoring relationships
and are consistent with more recent definitions of mentors as high-ranking,
influential members of an organization who are supportive of their proteges' career
development (Ragins and McFarlin 1990). These questions allowed us to examine
the existence of an informal mentor at two distinct points in time to see whether, and
when, such mentors affect an individual's formation of social capital.

We included several control variables in our analysis in order to account for factors,
other than the individuals' mentoring relationships, that might influence their range
of intersubsidiary contacts. A brief description and rationale for including these
controls follows.
Firm. We included dummy variables to control for the possibility of firm-
specific variations across the three firms. One might expect that being a part of a
firm that placed greater emphasis on career development might have an impact on
the effectiveness of early and/or later mentoring relationships, as well as a protege's
stock of social capital. Since we lacked information regarding the relative emphasis
placed on career development for these three firms, we had no expectations that one
firm would be more important with respect to a protege's ability to develop social
capital than another.
Subsidiary. Within any multinational system, some subsidiaries may be more
important and central than others in the internal resource-flow network (Bartlett and
Ghoshal 1989; Gupta and Govindrajan 1991). In such subsidiaries, managers may
have more opportunities to form intersubsidiary contacts. To control for this source
of internal variation within any MNC, we included a variable that measured the
relative importance of the respondent's subsidiary in the multinational system. This
measure was based on ratings of each subsidiary' s relative strategic importance and
resource capabilities provided by five headquarters-level respondents in each
multinational. The final score was the average rating of the subsidiary on an additive
scale comprising both dimensions. The scale ranged from subsidiaries that scored 1
(low importance) to 5 (very important). Given that our final sample was composed
of responses from the top ten countries across these multinationals, we ended up
with subsidiaries that were all quite important in each multinational, as reflected by
the high mean and low variance of the subsidiary measure for our sample of
Function. Because our respondents were either functional department heads or
general managers in their respective subsidiaries, we controlled for their job
responsibility. Several dummies were included to control for the possibility that
being a general manager as opposed to a functional head of marketing-or
manufacturing, research and development, finance, purchasing, legal affairs, or
administration--could have an influence on an individual's social capital. We
expected general managers to have a broader range of contacts than the heads of
functions by virtue of their more integrative organizational role.
The Sidekick Effect - 169

Career History. We also controlled for variations in the career histories of our
respondents that might influence their social capital. We included variables that
measured their tenure in the firm, the number of years they had worked in other
organizations, the number of other functions they had performed within the
organization, the number of other subsidiaries in which they had worked, the speed
of their promotion to their current position, and whether they were expatriates. We
expected each of these measures to have a positive influence, if any, on an
individual's range of intersubsidiary contacts.
Networking Opportunities. We controlled for differences in social capital that
might arise as a result of activities in which the respondent participated that could
provide him with unique opportunities to network or make contacts. In keeping with
work done by Ghoshal, Korine, and Szulanski (1994), we controlled for the time
(days per year) the respondent spent attending meetings and conferences;
participating in committees, teams, or task forces; attending training programs; and
whether or not the manager had received any formal training upon joining the
company. We expected all of these networking opportunities to have a positive
influence, if any, on the range of contacts developed by the managers in our study.
Perceived Constraints. Finally, we tried to control for the extent to which the
respondents perceived they had autonomy in their organizations and the extent to
which they perceived that they were not restricted by formal organizational rules and
procedures. Both measures were based on scales of perceived autonomy and
informality proposed by Ghoshal and Nohria (1989). We expected that managers
who perceived their jobs as more autonomous and/or informal would be more likely
to develop more extensive social networks than those who felt more constrained.

We used multiple linear regression (MLR) initiaIly to explore the effects of the
various independent variables on range. The output of this first regression suggested
that there are two clear groups of respondents: those with many ties, or high range,
and those with some ties, or low range. We defined 'many ties' as having 11 to 17
ties with other subsidiaries, while those 'some ties' as having from 1 to 10 ties with
other subsidiaries. No individual responded that he had no ties at all with other
subsidiaries. Given the results, it was appropriate to recode the subjects into these
two groups and to perform subsequent analyses based on this new dichotomous
dependent variable. 3
Following correlational analyses, two different levels of analyses were
conducted. First, using logistic regression, we tested the initial hypotheses: having a
mentor early hurts one's development of social capital, while having a mentor later
helps a manager develop a broad range of intersubsidiary contacts. Second, using
analyses of variance, we examined more closely the differences among individuals
falling into one of four categories: 1) those who had a mentor neither early nor late;
2) those who had a mentor late but not early; 3) those who had a mentor both early
and late; and 4) those who had a mentor early but not late.
170 - Corporate Social Capital and Liability

Table 1. Means, standard deviations, and correlations a

Variables Mean s.d. 2 3 4 5 6 7 8

1. Mentor early .46 .50
2. Mentor late .65 .48 .18*
3. Employee of .60 .49 -.03 .20**
4. Employee of Philips .17 .38 -.02 -.14 -.55 **
5. Relative importance of 3.71 .76 -.13 .06 -.09 .10
6. General manager .14 .34 .17* -.12 -.21 ** .22** .05
7. Marketing function .22 .42 -.13 -.01 -.04 .01 .06 -.21 **
8. Manufacturing function .19 .39 -.00 .05 -.16* -.10 -.19* -.19* -.25**
9. Purchasing function .07 .25 .02 .01 -.01 -.00 -.03 -.11 -.14 -.13
10. Research & .06 .23 -.03 -.03 .10 -.11 -.07 -.10 -.13 -.12
development function
II. Finance function .15 .36 -.03 .07 .01 .03 .05 -.16* -.22** -.20**
12. LegaVother .10 .30 .09 -.00 .03 -.10 .09 -.13 -.17* -.16*
administrative functions
13. Tenure at current finn 13.58 7.69 -.09 .09 .18* .21** -.04 -.03 -.07 .18*
14. Years at other 7.10 8.58 .06 -.17* -.17* -.02 .12 .08 .03 -.12
15. Number of other 2.20 1.98 .18* .01 -.06 .05 .02 .22** -.05 .08
functions worked in
16. Number of other .98 1.56 .02 -.00 -.03 .13 .05 .03 .03 .13
subsidiaries worked in
17. Speed of promotion .23 .12 .11 -.11 -.11 -.11 -.11 .12 -.05 -.12
18. Expatriate status .41 .49 -.24** .23** .12 .05 .06 -.10 -.03 .13
19. Time spent in cross- 13.69 15.04 .09 .04 .03 -.02 .25** .13 -.22** -.10
subsidiary meetings
20. Time spent in cross- 7.99 12.01 .07 .12 .00 -.15* .17* .07 .11 -.01
subsidiary teams
21. Time spent in cross- 4.39 10.22 .15* -.04 -.04 .22** -.06 -.01 .03 -.01
subsidiary training
22. Received initial training .47 .50 -.12 .32** .41** -.22** .01 -.11 -.12 .16*
23. Perception of 3.58 .74 -.09 .14 .21 ** -.10 .07 -.18* -.05 .04
24. Perception of 2.29 .57 -.19** -.13 -.22** -.17* .09 -.04 .12 -.12
25. Social capitalb 9.95 6.91 -.23** .15* -.14 .05 .18* -.07 .14 -.17*
*p<.05, ** p<.OI, aN = 177, Social capital is defined as the range or the number of national
subsidiaries of the firm in which therespondent has contacts.
The Sidekick Effect - 171

9 lD 11 12 13 14 15 16 17 18 19 20 21 22 23 24


-.11 -.lD
-.09 -.08 -.14

-.04 -.02 .07 -.14

-.01 -.11 -.09 .12 -.55**

-.04 -.05 -.24** .08 .18* -.00

-.05 -.06 -.03 -.06 .18* -.08 .49**

.18* .06 -.05 .04 -.42** -.34** -.16* -.07

.04 -.06 .11 -.20** .46** -.46** -.07 .17* -.11
-.lD -.01 -.08 -.12 .09 -.11 .08 -.04 .02 .15

-.02 .02 -.11 -.01 .00 -.02 .06 .02 .00 .06 .21 **

.12 .14 -.09 -.03 .04 -.07 .02 -.01 .08-.15* .05 .08

.06 .01 -.04 -.08 .33 ** -.36 ** .02 .13 -.lD .51 * .07 .10 -.01
.06 .13 .lD -.01 .24** -.20" -.17* -.14 -.12 .20** .lD .03 .17* .25**

.05 .02 -.lD .11 -.16* .lD -.02 -.01 .07 -.02 -lD .03 -.04 -.06 .03

.04 -.00 .12 -.02 -.06 .04 -.13 -.00 -.03 .18* .04 -.lD -.16* -.04 .03.18*
172 - Corporate Social Capital and Liability

Correlational Results
Table 1 presents the means, standard deviations, and correlations of the variables.
Intercorrelations between the control variables were of low to moderate strength.
The correlations between the control variables and the two mentoring variables were
all low, ranging from 0.00 to 0.24, indicating no problems of multicollinearity for
our regression analyses.

Logistic Regression Results

The results from the logistic regression are shown in Table 2. The logistic model fits
the data nicely. Overall, 72.32% of the respondents were correctly classified by our
model. The goodness-of-fit statistic (chi-square of 171.00 with 151 degrees of
freedom) further confirms that the model fits the observed data. We observed a
statistical significance level of p=0.13, indicating that our model does not differ
statistically significantly from a 'perfect' model in which the observed probabilities
exactly match those predicted by the model. Finally, similar to the overall F-test for
multiple linear regression, the logistic model chi-square (chi-square = 50.7, 25
degrees of freedom, p=O.OO 17) allows us to reject the null hypothesis that all of the
coefficients in our model are 0, except the constant. In sum, our model fits the
observed data quite well.
Our results show that the only variables that have a statistically significant effect
(at the p<0.05 level) on the respondent's range of contacts are having a mentor late,
having a mentor early, and the amount of time respondents spent in teams. These
results confirm hypotheses 1 and 2. Specifically, having a mentor early decreases the
odds-whereas having a mentor later increases the odds-of an individual being in
the high-range group. Counter to our expectations, the time a manager spent in
meetings actually had a negative impact on their range of contacts. With respect to
the other controls, the only other variable that had any statistically significant
influence was expatriate status, which had a positive impact on range if we relaxed
the confidence level (p<O.1 0).
The values shown in the last column of Table 2 allow us to gauge the magnitude
of these effects. They indicate the amount by which the odds are increased that a
respondent would fall in the category of those with a broad range of contacts, given
a unit increase in the explanatory variable. Our results indicate that if a respondent
had a mentor early in his career (the 'mentor early' variable changes from 0 to 1), the
odds of the respondent's being in the high-range group falls statistically significantly
(by a factor of .41). In short, an individual with an early mentor is only 40% as
likely as someone who does not have an early mentor to fall into the high-range
group. In contrast, we find that if a respondent has a mentor later in his career (the
'mentor late' variable changes from 0 to 1), the odds of the respondent being in the
high-range group increase dramatically (by a factor of 3.81). These odds ratios
confirm our main hypotheses.
In addition, expatriate status had a sizable effect on range. If the respondent was
an expatriate, he was much more likely (by a factor of 2.67) to be in the high-range
group. This result makes sense because expatriates are likely to have a cosmopolitan
The Sidekick Effect - 173

Table 2. Results of logistic regression analysis for social capitala,b

Regression Standard Significance Odds
Coefficients Errors Levels Ratios
Mentor early -0.90 0.41 0.03 0.41
Mentor late 1.34 0.46 0.00 3.81
Employee of Matsushita -0.61 0.53 0.25 0.54
Employee of Philips 0.06 0.73 0.94 1.06
Relative importance of subsidiary 0.30 0.26 0.25 1.35
General management 0.52 0.95 0.59 1.68
Marketing function 1.13 0.93 0.22 0.31
Manufacturing function 0.10 0.93 0.91 1.11
Purchasing function 1.16 1.11 0.30 3.18
Research & development function 1.19 1.16 0.30 3.29
Finance function 1.02 0.97 0.30 2.77
LegaVother administrative functions 0.74 0.99 0.45 2.09
Tenure at current fmn -0.01 0.05 0.92 1.00
Years at other employers 0.03 O.OS 0.48 1.03
Number of other functions worked in -0.09 0.12 0.44 0.91
Number of other subsidiaries worked in 0.04 0.16 0.79 1.04
Speed of promotion 0.53 3.13 0.86 1.71
Expatriate status 0.98 0.52 0.06 2.67
Time spent in cross-subsidiary meetings 0.01 0.01 0.45 1.01
Time spent in cross-subsidiary teams -0.03 0.02 O.OS 0.97
Time spent in cross-subsidiary training -0.03 0.02 0.27 0.97
Received initial training -0.48 O.SI 0.3S 0.62
Perception of organizational autonomy 0.01 0.28 0.98 1.01
Perception of organizational informality 0.S4 0.38 O.1S 1.72
a Social capital is defined as the range or the number of national subsidiaries of the fmn in which
the respondent has contacts. For this analysis, a manager may have 'many ties' (contacts in 11-17
subsidiaries) or 'some ties' (contacts in 1-10 subsidiaries).
b Goodness of fit:
Chi-Square df Significance
Model Chi-Square 50.71 2S 0.00
Goodness of Fit 171.00 lSI 0.13

orientation (Ritti)5 as well as contacts in their home country. Research has also
shown that expatriates often serve as gatekeepers to the outside world in many
subsidiaries, which also partially explains the results we observe (Edstrom and
Galbraith 1977).
Finally, although participation in teams was a statistically significant variable in
the overall model, the size of the effect is negligible; the odds of being in the high-
range group versus the low-range group remain largely unchanged as participation
174 - Corporate Social Capital and Liability

Analyses of Variance Results

Having confirmed our initial hypotheses, we performed a second level of analysis to
explore further differences among respondents who fell into one of four groups: 1)
those who had a mentor neither early nor late; 2) those who had a mentor late but
not early; 3) those who had a mentor both early and late; and 4) those who had a
mentor early but not late. In this way, we examined if having a mentor both early
and late or neither early nor late had an equivalent effect on an individual's ability to
develop social capital. In addition, we strove to understand individual differences
between respondents that may have influenced the mentoring relationships they
We used one-way analyses of variance with a dichotomous dependent variable
to assess differences between the four groups. In accordance with our regression
analysis, the results (see Table 3) show that respondents in each of these groups
differed statistically significantly in their range of intersubsidiary contacts. More
specifically, the range of contacts for respondents having a mentor late but not early
(Group 2) was significantly greater than each of the other groups (i.e., having a
mentor early but not late (Group 4), having a mentor both early and late (Group 3),
and not having a mentor at all (Group I)). Interestingly, although not statistically
significantly so, having no mentors at all was better than having an early mentor, in
line with recent research that has suggested that those who take personal
responsibility for building their social capital do better than those who rely on others
to build contacts (Gabbay 1995, 1997). Also, having a mentor only early was clearly
the worst scenario because those who had mentors both early and late (Group 3)
appeared to have had somewhat higher range than those who had only an early
mentor. All of these results are consistent with our hypotheses: having a mentor
early creates a sidekick effect, whereas having a mentor later in one's organizational
career helps a manager develop social capital.
Beyond differences in their range of contacts, there were a few other noteworthy
differences among these groups. For instance, expatriates were more likely to have
formed a mentoring relationship later in their career than earlier. We speculate that
one reason might be that these individuals moved to a subsidiary late in their
careers-thus becoming expatriates-and, at that stage of their careers, sought out
mentors to help them perform better in the context of the broader multinational
Another difference to note is that initial training can be a substitute for initial
mentoring. Those who received some formal training on joining the organization
were most likely not to have an early mentor and most likely to have one later on
(Group 2). In sharp contrast, 90% of those reporting having a mentor early but not
late (Group 4) did not receive initial training. This finding has interesting
implications for the design of career development programs. Rather than focusing on
forming early mentoring relations, as many career development programs are
increasingly attempting to do (Caruso 1992), firms may do better by providing
initial training as a way of socializing new employees into an organization and
letting individuals develop mentoring relationships on their own later in their
careers. Indeed, research has suggested that some of the greatest benefits of the

§. ~ g- ; s. ~ Description of Variable Groups ;'

!l '0g'
5:_. ~E¥ -l9@ 8.e?..;' Different· ..S!:
~. (1) g n {:, §. !'"
n 101 (l.. C» • CI5
~ 8: 5' Mentor early No No Yes Yes ~
c:: "'I!:S ~
;~ s..... "s.[ Mentor late No Yes Yes No
::r _. '" !!! iii'
Ef ~ ~ ~. g Dependent Variable o
~ ~ >-j
g g- Social capitala 9.73 12.69 8.57 7.19 (1,2) (2,3) (2,4) §
j!i Control Variables
(f,I U (f,I
g. g a. '<~
.....::: !Z el ",- a Employee of Matsushita 0.41 0.76 0.58 0.57 (1,2) fg,
g 61 g- 8~ Employee of Philips 0.27 0.11 0.15 0.19
~ g. § .€I.. Relative importance of subsidiary 3.72 3.86 3.64 3.52 <
;. iii' <Ill a· ~ General management function 0.17 0.02 0.18 0.24 (2,3) §
_ ~ Sl g ~ Marketing function 0.29 0.25 0.18 0.10 ~
- ~
!.. "'" (1)
- if 0
3 j!i Manufacturing function 0.20 0.18 0.22 0.10 ~
-; p;' .. ~ ~ Purchasing function 0.02 0.09 0.05 0.14
g. g- ~ § ':-' Research & Development function 0.07 0.05 0.05 0.05
e; ii" .., ~ Finance function 0.10 0.20 0.13 0.14
~. g~ ~ Legal lother administrative functions 0.07 0.07 0.12 0.14
o g~.
_ 0 !!!
Tenure at current Firm 13.20 14.98 13.22 11.71
:- < i ~. Years spent at other employers 8.68 5.03 6.88 10.00
~ ~. ~ ~. Number of other functions worked in 2.27 1.60 2.80 1.95 (2,3)
:! if go ~ Number of other subsidiaries worked in 0.98 0.95 1.02 1.00
~ ~
.g C/.l
~~ iii'
~ ~ Speed of promotion to current position 0.23 0.22 0.23 0.29 '"
8' ~ ii' ci. Expatriate status 0.32 0.67 0.33 0.14 (1,2) (2,3) (2,4) 0..:

S- ~ ~ ~ Time spent in cross-subsidiary meetings 11.83 13.02 15.07 15.14 t::

~ 3 e- g Time spent in cross-subsidiary teams 5.54 8.45 9.58 7.05 ?I"
s· ~ ~ ;;;, Time spent in cross-subsidiary training 4.49 1.89 6.12 5.67 trl
! ;,; 8 ~ Received initial training 0.34 0.67 0.52 0.10 (1,2) (2,3) (3,4) ~
o § S· ~ Perception of organizational autonomy 3.42 3.81 3.51 3.50 (1,2)
J1; ~ ::r Perception of organizational informality 2.46 2.35 2.14 2.28 (1,3) -.l
!l ::r U.
176 - Corporate Social Capital and Liability

mentoring relationships are derived through informal means and pertain to informal
aspects of the company (Collin 1979). Alternatively, early training may serve as one
of many mechanisms through which individuals can begin to develop helping
relationships that, in the longer run, could blossom into mentoring relationships.
Future research is needed that investigates the relationship between formal
organizational initiatives, such as training programs, and the development of
valuable informal relationships, such as mentoring.
Finally, the results suggest that those who form no mentoring relationships
(Group 1) may be 'prisoners of their own perceptions.' These individuals scored
lowest on perceptions of both organizational autonomy and informality. Because
these people view the organization as being constrained, they may undervalue the
benefits of mentoring relationships that can serve as pathways to accessing the
informal side of the organization. This finding supports Turban and Dougherty's
(1994) view that we need to examine more closely how an individual's personality
and attitudes might affect the mentoring relationships he or she seeks and initiates
and the benefits derived from them.


Our objective in this chapter was to better understand how mentoring relationships
actually work in organizational settings to affect specific career outcomes. We
pursued this goal by examining both when mentoring relationships do work and
when mentoring relationships do not work with respect to one career variable that
scholars have linked to career progress-social capital, or beneficial ties with others
in an organization. Our results suggest that the timing of having a mentor affects a
protege's ability to develop social capital over the course of his or her organizational
career. Specifically, having a mentor early in one's organizational career can
produce a sidekick effect in which individuals are overshadowed by their sponsors
such that their development of independent social capital is difficult. However, this
sidekick effect disappears once the protege has developed a reputation as an
individual contributor within the organization. At this later time, a mentor may be
quite helpful in facilitating the formation of new ties since the protege has already
developed some form of professional identity and reputation and, therefore, is not
viewed as walking in anyone's shadow.
In addition, two other findings from this research were interesting and consistent
across our analyses: expatriate status had a positive effect, and integrating
mechanisms (such as participation in team work), had a negative effect upon the
development of social capital. First, it appears that being an expatriate increases the
likelihood that a protege will have many ties across multinational subsidiary
boundaries. This seems plausible, given that an individual who has lived outside of
his or her home country may have a better understanding of the cultures, history,
language, and customs of other nations. Overall, expatriates may develop a more
cosmopolitan orientation (Ritti)5 that later serves to facilitate the development and
maintenance of international ties.
Second, integrating mechanisms, such as the amount of time spent engaged in
team work or participation in training activities seems to have a negative impact on
the ability to develop ties with others across subsidiaries. Although the specific type
The Sidekick Effect - 177

of integrating mechanism varied somewhat across our analyses, this general finding
seems worthy of further investigation. Drawing on the work of Louis (1990), we feel
that these types of interactions may lead to densely clustered subnetworks which
impede one's ability or interest in fostering a broad range of connections with others
outside the subnetwork. As Louis' work suggests, those who develop strong 'buddy
relationships' are less apt to negotiate relationships beyond their local work
environments. Thus, peer group interaction can lead to fraternizing that impairs
newcomer acculturation into the organization. In contrast, 'isolates' are more likely
to gain a better understanding of their work environment simply because they have
been forced to rely on multiple and varied sources of information. Thus, while
interaction with others may provide social support, it does not necessarily lead to a
broader network of contacts across intraorganizational boundaries.
Our study has several limitations that would be important to address in future
research. First, and most importantly, our work is based on a limited amount of data
regarding the nature of the mentoring relationships we studied. All we asked our
respondents was whether they had a mentoring relationship or not. Factors not
known to us-such as the identity of the mentor, the mentor's structural position in
the firm, the process by which the mentor was chosen, and the nature and strength of
the mentoring relationship--should all be studied to see whether and how they
mediate our main findings. For example, we did not know whether the early mentor
was also the protege's direct boss, and so could not compare supervisory and
nonsupervisory mentoring relationships-a topic Scandura and Schriesheim (1994)
have suggested for future research. Indeed, as Burt's (1992) work showed,
hierarchical networks are most effective when they revolve around sponsors who are
distant from the individual's immediate work group--that is, when the sponsor is
not the manager's direct boss. In short, we expect that some mentors may be better
than others at preventing their proteges from being seen as sidekicks.
Second, we studied just one issue of importance to an individual's career-the
formation of social capital. Although social capital plays an important role in
shaping such career outcomes as advancement, mobility, and performance, the
amount of human capital (educational qualifications and skills) and physical capital
(money) possessed by an individual can play an equally important role. Thus, even
if we find that early mentoring relationships have a negative effect on the
development of an individual's social capital, we cannot conclude that such
relationships necessarily harm the individual's career because they may have a
countervailing positive effect on the development of the individual's human capital.
Third, although we attempted to control for several factors that might influence
the range of intersubsidiary contacts possessed by our managers, we recognize that
there are other factors for which we did not control. For instance, recent research
suggests that an individual's personality (on which we collected no data) can
influence the extent to which they seek out and make use of mentors (Turban and
Dougherty 1994). It has also been argued that gender and race can playa significant
role in mentoring relationships (Ibarra 1993c; Thomas 1993). We were not able to
assess gender effects in this study since all of our respondents were male; neither did
we have data on race and, hence, could not control for such effects.
178 - Corporate Social Capital and Liability

Finally, the group of managers we studied are only a subset of the population of
all managers. They all worked in subsidiaries of foreign multinationals and were all
male. Thus, our findings have limited generalizability. It would be useful to explore
these issues in other organizational settings, especially in U.S. firms, and with
managers of more diverse demographic characteristics.
Despite several limitations, this work has many important implications for those
conducting research on mentoring relationships and on helping relationships more
generally. First, by examining the impact of mentoring relationships on a protege's
stock of social capital, we began to unravel possible explanations of how mentoring
functions actually work to affect proteges' career outcomes. We proposed that there
are several roles that mentors can take in providing career help to proteges; mentors
may serve as signaling agents, providers of access, and providers of opportunities.
Although we expect that such roles may lead to near term benefits for the protege,
our findings suggest that there are also long term costs associated with having such
relationships. Specifically, having a mentor early in one's career can produce a
sidekick effect such that the development of ties with others in the organization
becomes difficult.
Second, our findings and the rationale that supports them challenge assumptions
in the mentoring literature. For instance, much mentoring work asserts that different
mentoring functions work together to influence career outcomes, and that the more
functions provided by a mentor, the more beneficial the mentoring relationship
(Kram 1985; Noe 1988a). The research presented here, however, points to a
contingency model in which we consider how different types of mentoring
assistance provided at different times during a protege's career can affect career
outcomes. This raises important questions regarding the appropriate sequence and
ordering of specific career mentoring functions over the span of a protege's career.
For example, is it beneficial for a mentor to provide access before the protege has
established confidence and competence in his or her ability to excel in an
organization? On a related note, this research calls for future work to examine
whether there is a trade-off between which sets of mentoring functions best serve the
protege early versus later in his or her organizational career. For example, is
psychosocial help early and career help later the best combination of help for a
protege who is trying to advance in an organization?
Third, this work suggests a host of opportunities for future research on both the
timing and types of interventions designed to help in general (Higgins 1997a). For
example, since our questions tapped whether or not a respondent had a mentor at
different points in time, regardless of whether this was the same person doing the
mentoring, our research opens up the possibility for future work on the optimal mix
of help-providers over the course of one's career. Stretching this notion of timing
even further, one could also consider how a protege's career is affected by the mix
of mentors he or she has across the stages of adult development (Levinson et al.
1978) or during the transitions between them (Daloz 1986; Kegan 1982). Thus,
rather than explore the timing of different types of help that are given by one help-
provider over the course of one mentoring relationship, researchers might also
consider the combination of different types of help-providers an individual has over
the course of his or her career and life.
The Sidekick Effect - 179

In short, we need to consider not just knowing how mentoring relationships

work, but also what effects the who and when of such relationships have on a
protege's ability to progress in his or her career (Higgins and Thomas 1997). This
approach would capture the inherently more contingent nature of careers and
helping relationships. Most importantly, it would allow our thinking to move beyond
the 'myth of the perfect mentor' (Hill and Kamprath 1991) to be more in line with
recent research that has suggested that career developmental relationships are not
monodyadic, hierarchical, and intraorganizational, but rather exist in constellations
(Kram 1985) or portfolios composed of both superiors and subordinates and can
exist both inside and outside the firm (Higgins 1997b; Thomas and Higgins 1996).
Indeed, the sidekick effect itself is likely to be affected under such a perspective, for
a person may be less likely to be overshadowed if he or she has several people from
whom to draw career support rather than just one prominent superior.
In sum, the sidekick effect highlights the potential negative implications of early
mentoring relationships with respect to one important career factor, the development
of social capital, within a literature that has characterized such relationships as
primarily positive. From a practical perspective, when initiating informal mentoring
relationships, this work suggests that it is important to consider the possibility of a
sidekick effect which could hinder a protege's ability to develop ties with others.
And, from a theoretical perspective, when trying to understand how mentoring
relationships actually work in organizational settings, this research suggests that it is
important to recognize the inherently contingent and complex nature of giving and
receiving help--in particular, to consider how the effectiveness of the type and
timing of mentoring relationships can affect individual career outcomes.

We would like to thank Chris Bartlett and Sumantra Ghoshal for providing us with the data used in this
study and Jack Gabarro, Richard Hackman, Linda Hill, Herminia Ibarra, David Thomas, and Douwe
Yntema for their helpful comments on this research.

1. We use the term 'organizational career,' rather than 'career,' because this research focuses on the
impact of mentoring on an individual's career in one organization, as opposed to an individual's career in
more than one organization.
2. 'Early' is defined as within the first two years of joining an organization. 'Later' or 'long term' is
defined as anytime thereafter.
3. Statements on statistical significance refer to the .05 level.
4. We reached this conclusion after examining the frequencies table of the dependent variable and after
reviewing the plot of residuals versus predicted values, which showed two parallel lines of scattered
points. Note that this procedure is not equivalent to simply dichotomizing the range variable using a
median split. Instead, the logistic regression analysis is conducted because the residuals violate the
random distribution assumption in ordinary least squares models.
5. As referenced by W. Humphrey (1987).
Social Capital in Internal

Staffing Practices

Peter V. Marsden
Elizabeth H. Gorman

This chapter examines the information sources that U.S. employers use in the course
of internal staffing, that is, when promoting or transferring employees. We focus on
the use of methods involving informal ties: referrals and direct approaches to
candidates for promotion or transfer. Such ties may produce 'social capital' by
providing employers with information about the qualifications and abilities of
personnel; at the same time, they provide employees with information about
opportunities for mobility within the workplace or firm. Data from a representative
sample of work establishments indicate that informal methods are widely used in
filling vacancies with internal candidates, often in combination with more
formalized procedures such as job posting and seniority systems. Differences in
internal recruitment procedures across types of employers and jobs suggest that they
are selected in light of both efficiency concerns and pressures for equity and
procedural rationality in the treatment of employees.

Organizational employment policies and practices playa central role in structuring
the inequality of rewards among individuals (Baron 1984). Careers consist of
sequences of positions, which may involve moves-promotions, lateral transfers, or
even demotions-within as well as between organizations. Promotion practices, in
particular, shape upward career mobility for individuals (Baker, Gibbs, and
Holmstrom 1994; Rosenbaum 1979a, 1979b; Stewman 1986; Stewman and Konda
1983) and influence earnings (Cappelli and Cascio 1991; KaUeberg and Lincoln
1988; Le Grand, Szulkin, and Tahlin 1994). One strand of research on
Social Capital in Internal Staffing Practices - 181

organizational promotion practices has examined structural, institutional, and market

correlates of the prevalence of job ladders and other structured mobility channels
within organizations (Baron, Davis-Blake, and Bielby 1986; Bridges and Villemez
1991; Kalleberg, Marsden, Knoke, and Spaeth 1996; Pfeffer and Cohen 1984).
Other studies have explored the shape of formal job ladders and their interplay with
informal career trajectories (DiPrete 1987; Miner and Estler 1985; Kanter 1983b).
Still others have examined the criteria applied when rewarding individuals with
promotions (Baker, Gibbs, and Holmstrom 1994; Bills 1988; Kanter 1977;
Nicholson 1993).
In light of the centrality of intraorganizational mobility to careers, it is
surprising that researchers have given little attention to the processes through which
it occurs, that is, the manner in which organizations identify and select candidates
for promotion. Both formal and informal methods exist for accomplishing these
tasks. Seniority systems and job posting are the principal formal procedures. They
identify and select from candidate pools using abstract, universalistic criteria. As
such, they can be viewed as part of a larger process of bureaucratization within the
employment relationship (Baron, Dobbin, and Jennings 1986; Baron, Jennings, and
Dobbin 1988; Bridges and Villemez 1991), along with other features of internal
labor markets. Indeed, the dearth of research on internal recruitment procedures may
reflect an assumption on the part of researchers that promotion systems necessarily
rely on formal selection procedures.
Yet even relatively bureaucratic organizations can and do make use of informal
channels when identifying and selecting candidates for promotion (Pinfield 1995).
Such methods include referrals as well as direct contacts between the selecting
official(s) and candidates for promotion and transfer; they limit the pool of
candidates to those with direct or indirect social ties to the selecting official(s).
Studies examining individual differences in mobility within organizations have
shown that particular configurations of social ties are associated with promotions
(Burt 1992; Podolny and Baron 1997), and it is therefore of interest to examine them
from the organization'S side.
In related research on methods of hiring from external markets, researchers have
demonstrated that organizations frequently rely on recruitment and selection
procedures that make use of candidates' social ties (Fernandez and Weinberg 1997;
Fevre 1989; Jenkins, Bryman, Ford, Keil, and Beardsworth 1983; Kalleberg et al.
1996: chapter 7; Marsden and Campbell 1990). A cognate body of work has
demonstrated that the form and content of social ties is linked to individual success
in finding jobs (e.g., Boxman, De Graaf, and Flap 1991; Flap and Boxman, this
volume; Granovetter 1995; Lin, Ensel, and Vaughn 1981; Marsden and Hurlbert
1988; Wegener 1991). This chapter focuses on similar phenomena in the
intraorganizational setting.
In the next section, we discuss the manner in which we view organizational
information channels as avenues for the operation of social capital. We then describe
the various methods that organizations use in disseminating information about
opportunities for promotion and transfer. Our theoretical account of variation in the
use of informal methods is based on efficiency and strategic concerns, together with
external pressures for equity in the allocation of opportunities. After introducing the
182 - Corporate Social Capital and Liability

data base for our analysis, the 1991 National Organizations Study (NOS), we present
findings, and conclude by discussing their implications for organizations and


We view recruitment procedures that draw on preexisting social ties as settings in
which the operation of social capital within organizations may be observed. Because
the concept of social capital has been used in diverse ways, however, as illustrated
by discussions in Foley and Edwards (1997), Greeley (1997), and Newton (1997),
analyses invoking it must be explicit about its denotation. Most or all variants on the
concept posit that social ties may serve as resources for individual or collective
action; we take this as the essential core of social capital, as distinct from related or
additional features such as norms of trust that may be associated with the presence
of social ties. The ties in question need not be channels deliberately developed to
facilitate purposive action; indeed, much is made of the unplanned or adventitious
use of social ties originating outside of a given context. Thus, Granovetter (1985)
contends that economic transactions are often 'embedded' in pre-existing social
relations, while Coleman (1988) stresses the 'appropriability' of social
organization-i.e., its use for purposes other than those for which it was
established-as a fundamental characteristic of social capital.
Different conceptualizations of social capital highlight its role as either a
property of collectivities and communities or an individual resource. Both
conceptual strands are evident in Coleman's (1988) influential early statement. As a
collective property, social capital inheres in the web of social relations within a
collectivity, and the set of obligations, information channels, and norms that those
relations support and enforce (Coleman 1988). Like clean air or municipal policing,
social capital at this level is a collective good, one that can facilitate both collective
and individual actions (Coleman 1988; Putnam 1993b, 1995a). As an individual
resource, social capital resides within the direct and indirect social ties giving the
social location of a focal individual. Like human capital, social capital at this level
provides individuals with differential competitive advantages (Burt 1992).
In this study, we emphasize the second conceptual strand; we conceive of
informal channels within organizations as vessels for social capital that enable
employers to gather information of superior quality about candidates for promotion
or transfer. Individuals, of course, benefit from being in social locations permitting
access to such networks. As well, the staffing decisions reached by making use of
the social capital contained within informal channels may result in collective
benefits such as strengthened organizational performance, but we do not stress this
here. We direct our attention to the nature and determinants of organizational
staffing procedures that draw on individual social capital and thereby enhance its
influence on individual mobility outcomes.


The principal contrast we draw in studying internal staffing methods parallels the
formaVinformal distinction used in existing literature on job matching in external
Social Capital in Internal Staffing Practices - 183

markets. We begin by discussing informal methods involving referrals and direct

approaches to employees. We then mention two formal methods, seniority and job
posting systems, more briefly.
Informal recruitment involves information transfers in social networks, and may
take several distinct forms. Within a work unit, supervisors usually monitor the
performance of employees on an ongoing basis. Using this information, they may
directly approach employees they deem qualified when higher-level vacancies occur
or new positions are created. Information about employees who might be promoted
or transferred may also be shared through word-of-mouth referrals from other
supervisors, other employees, or even people outside the firm such as clients or
customers. Such channels contain social capital because they can convey relatively
rich, qualitative information about the capabilities of potential candidates for a new
position; reciprocally, they may also provide candidates with 'realistic job previews.'
A prototypical case in which one would observe informal staffing is that of
'drive systems' involving direct, personal control (Edwards 1979; Jacoby 1985).
Such systems delegate substantial discretion to supervisors who may be capricious,
even despotic. Pinfield's (1995) description of staffing as an 'administered process'
in his case study of a forestry firm involves important elements of informal
recruitment, though it occurred in a context where formal procedures also were
present. He observes that managers preferred the administered appointment process
because it took less time, and permitted them to 'manage lines of progression' for
their subordinates. With administered appointment processes, managers could use
the promise of promotion as a means of motivating lower-level employees within
their units; they were also able to reduce their risks because informal channels
yielded trustworthy information about the personal reputations of candidates.
Of course, as drive systems and personal control remind us, by using informal
channels to disseminate and gather labor market information, organizations provide
an opening for not just supervisory discretion and good jUdgment, but also prejudice
and favoritism, to enter promotion and transfer decisions. There may, then, be
liabilities as well as benefits to the use of social ties. This highlights the omnipresent
distinction between form and content in the study of social networks; it is
problematic to infer content from form. The fact that personal ties may be used not
only to transmit subtle, local 'impacted' information, but to exclude and/or unduly
advantage some candidates at the expense of others, raises concerns about equity in
the treatment of employees and calls for limitations on the discretion of supervisors.
Formal methods of internal recruitment have arisen in response to these
concerns. Seniority-the allocation of promotion opportunities on the basis of length
of service-represents one approach to limiting managerial caprice and thereby
increasing universalism within the promotion process. Kochan, Katz and McKersie
(1994) include seniority among the central elements of the Job control' industrial
relations system that was widespread in the U.S. in mid-century, particularly in
unionized settings. Freeman and Medoff (1984) identify several ways in which the
seniority principle may enter into promotion decisions, ranging from seniority as the
dominant factor in promotions (irrespective of ability) to seniority as a basis for
setting priorities among candidates whose qualifications are otherwise equivalent.
184 - Corporate Social Capital and Liability

Job posting constitutes a distinct approach to formalizing procedures for

promotions and transfers within internal labor markets. A posting system publicizes
internal job opportunities through written notices on bulletin boards or employee
newsletters; see Pinfield (1995) for a generic description of such a system.
Announcements of vacancies include job descriptions and specifications of
qualifications; current employees are encouraged to put themselves forward as
candidates to fill vacancies. Posting systems introduce open competition into the
allocation of job opportunities to current employees.
Pinfield (1995) observes that posting systems provide a high degree of
procedural rationality and legitimacy to promotion and transfer decisions. Jacoby
(1985) notes that unions view posting as an alternative or complement to seniority as
a mode of developing job ladders and providing advancement opportunities for their
members. Additionally, many seeking policy levers with which to reduce sex
segregation in employment (e.g., Kanter 1977) advocate job posting as one approach
to increasing equality of opportunity within organizations by publicizing vacancies
beyond the reach of informal ties that are often confined within work units.


Our concern in this chapter lies in the factors that make it more or less likely that an
organization will use informal channels in internal staffing decisions. Some factors
of interest refer to the organization per se, while others pertain to differences in the
nature of work done in positions within organizations. We argue in this section that
these factors operate through several distinct intervening mechanisms. From a
rationalist standpoint, some factors are linked to the benefits to be gained by using
informal channels or the costs (including opportunity costs) of doing so. Others
instead reflect institutional pressures toward equity or procedural rationality in the
management of staffing, which may emanate both from sources internal to an
organization and from the environment in which it is situated.

Benefits: Information QUality

From the standpoint of rationality and efficiency, one would expect organizations to
adopt staffing methods that rely on interpersonal connections under circumstances in
which the benefits of such methods are greater. The principal benefit of informal
methods drawing on social capital is that they may provide richer, more extensive,
and more situated information concerning candidates than can formal, impersonal
Informal methods involving social ties should be especially likely to yield
higher-quality information when interpersonal skills and the ability to form networks
of contacts are central to effective performance. Such skills are important in jobs
that involve frequent communication with clients or customers; for example,
investment banks rely on their employees' networks of external ties to bring in new
business and maintain client relationships (Eccles and Crane 1988). We therefore
expect to find greater use of informal internal staffing methods in professional, sales,
and service occupations.
Social Capital in Internal Staffing Practices - 185

Interpersonal skills and social networks are also important in flexible

organizations with staffing strategies that adapt goals and objectives to the
capabilities of their personnel, rather than seeking people who fit the requirements
of standardized jobs. In such organizations, staffing can be seen as equivalent to
strategy formation (Snow and Snell 1993). They make less use of formal job
descriptions and rules to clarify and coordinate tasks, and instead expect the
contours of jobs to be altered to suit the talents and initiative of their incumbents. In
such situations, networks of internal ties serve as a coordination mechanism (Eccles
and Crane 1988) and convey clear normative expectations associated with each
employee's role in the organization (Podolny and Baron 1997). Given the need for
rather nuanced, local knowledge of someone's capabilities in such systems, we
anticipate that organizations without formal job descriptions will tend to make
greater use of informal staffing procedures.
Informal selection methods should also provide better-quality information when
the skills involved in the position to be filled are difficult to measure objectively. If
the responsibilities of the position to be filled are nonroutine and involve the
exercise of discretion, objective standards are difficult to apply. Subjective
evaluations obtained through social networks may provide the best available
information concerning a candidate's potential (Pfeffer 1977). This argument
suggests that organizations will rely more heavily on informal selection methods for
managerial, professional, and skilled-craft positions. For managers, this claim is
supported by a study of supervisory selection in plants within eight manufacturing
industries (Northrup et al. 1978) which found that 'most procedure is highly
Finally, informal selection methods should provide better-quality information in
circumstances where the costs of a selection error are relatively high. In such cases,
employers are more likely to be interested in obtaining the fine-grained information
concerning candidates' personal characteristics that only informal procedures can
provide. The costs of a selection error are higher when an employer invests in the
training of promoted employees, because such investments will be lost if the
subsequent performance of the employees is poor. Selection errors can also be costly
when promotions occur within multiple-level job ladders, because incumbents in
such positions are likely to have long tenures in the organization. Due process
guarantees may make it difficult to discharge them, and selection for a position on a
job ladder is likely to enhance an employee's eligibility for further promotions. We
therefore expect to find greater use of informal methods in occupations that receive
formal training and that have mUltiple levels (i.e., that form part of job ladders).

Administrative and Opportunity Costs

The intrinsic costs of administering informal procedures for identifying and
selecting promotion candidates are generally low. Because substantial informal
information is acquired passively in the course of undertaking other tasks, it can be
viewed as essentially costless. Even when invoked actively, the use of such methods
may involve nothing more than picking up the telephone or wandering down the hall
to discuss a candidate; of course, intensive pursuit of closely-held information
through informal ties may be costly, especially for confidential matters.
186 - Corporate Social Capital and Liability

One reason for relying on relatively low-cost informal methods is that

alternatives are too expensive. Small, single-site organizations may not find it cost-
effective to administer formal staffing procedures, for example. Small
establishments that are part of multi-site organizational systems may, however, be
required to use centrally prescribed formal procedures for making promotion and
transfer decisions, and hence be less apt to use informal methods of doing so.
Because the range of candidates that can be identified using informal ties is
limited to those who can be tapped through the interpersonal networks of selecting
officials, a principal cost (or social liability) of informal methods is the foregone
opportunity of missing talented candidates who are not identified through the social
ties used for an internal personnel decision. In small, single-site establishments, this
opportunity cost should be low, because selecting officials are apt to have direct or
indirect social ties to most or all employees. In larger establishments and
establishments within multi-site organizations, on the other hand, the pools of
candidates that can be defined on the basis of social ties will usually consist of much
smaller fractions of the potentially eligible workforce, creating a greater opportunity
cost. Accordingly, we expect to find less use of informal procedures in large or
multiple-site establishments.

Constraints: Equity and Rationality Pressures

As noted above in the discussion of seniority systems and job posting, constituencies
both internal and external to organizations exert pressures on them to allocate
rewards, including promotion and transfer opportunities, in an equitable and
procedurally rational manner. Informal methods of identifying and selecting
promotion candidates are more vulnerable to charges of prejudice and favoritism
than are formal procedures that rely on objective, universalistic criteria. Informal
procedures are also less consistent with norms of rationality and bureaucratization,
which many see as institutionalized myths or societal values that infuse
organizations (Meyer and Rowan 1991; Bridges and Villemez 1991; Baron, Dobbin,
and Jennings 1986).
External pressures emanate from unions and government regulators. Differences
in personnel practices between unionized and nonunion workplaces have been
stressed by a number ofresearchers (e.g., Jacoby 1985; Baron, Dobbin, and Jennings
1986; Dobbin et al. 1988). Cohen and Pfeffer (1986), for example, argue that unions
advocate formalized staffing procedures because such procedures can prevent
employers from penalizing employees with pro-union attitudes. Institutional
arguments hold that exposure to the public sphere places organizations under special
pressure to conform to evolving norms about legitimate employment practices
(Dobbin et al. 1988). Larger establishments and establishments that are part of
multi-site organizations are more visible to regulators, and consequently such
establishments should be more reluctant to make use of informal promotion
procedures. Public sector establishments, in particular, must demonstrate high levels
of fairness, objectivity, and openness in their employment practices; they are also
subject to civil service laws and regulations that mandate the use of certain formal
procedures (DiPrete 1989; Tolbert and Zucker 1983). Thus public sector
Social Capital in Internal Staffing Practices - 187

establishments should be especially likely to avoid the selection of personnel

through informal methods.
Internally, personnel departments are likely to favor formal methods of
identifying and selecting promotion candidates. One reason is that personnel
professionals are especially aware of the constraints and sensitivities of external
constituencies (Jacoby 1985); a second is that they enhance their intraorganizational
power through the possession of specialized knowledge about the conduct of
personnel actions (Pfeffer and Cohen 1984). We therefore expect that establishments
having personnel departments will make less extensive use of informal staffing


The data presented in this chapter are drawn from the National Organizations Study
(NOS), conducted during 1991. Telephone interviews were completed with
informants for a multiplicity sample (see Parcel, Kaufman and Jolly 1991)
representative of U.S. work establishments. l When contacting establishments, NOS
interviewers were instructed to speak with 'the head of the personnel department or
the person responsible for hiring.' Overall, the NOS attempted to contact informants
for 1,067 establishments; it successfully conducted interviews with 688 of them, a
completion rate of 64.5 percent. 2 For additional details about field procedures used
in the NOS, see Spaeth and O'Rourke (1994) or Kalleberg et al. (1996: chapter 2).
To take into account possible between-occupation, within-establishment
variation, the design of the NOS interview schedule specified that several question
sequences, including that having to do with internal staffing, were to be repeated for
up to three different occupations in each establishment. One of these was the job
title of the employees 'most directly involved' with the main product or service
provided by the establishment; below we refer to this as the 'core' occupation. A
second was the occupation given by the General Social Survey (GSS) respondent
who had given the name of the establishment. Finally, questions were also posed
about 'managers or other administrators.' This multiple-occupation design permits
the separation of establishment- and occupation-level influences on several NOS
outcome variables.


To measure the procedures used when filling a job internally, the NOS used this
question sequence:

When you fill this job with a person already in the organization, how often do you
1. Consult a seniority list?
2. Inform current employees by posting or circulating a vacancy notice?
3. Ask the person leaving the job to recommend other current employees?
4. Ask others at your workplace for recommendations?
5. Go directly to specific employees and encourage them to apply?
188 - Corporate Social Capital and Liability

Table 1. 'Frequent' use of internal staffing methods by NOS establishments

Percentages of Establishments
Referrals Any
Seniority Job from Referrals Direct Informal
Occupation Lists Posting Incumbent from Others Approach Method
'Core' employees
Unweighted 38.5% 67.8% 8.1% 24.1% 19.4% 34.8%
Weighted 26.9 42.2 12.7 25.0 24.1 38.1
(N) (387) (388) (382) (386) (386) (388)
'GSS' employees
Unweighted 34.4% 64.7% 8.3% 16.9% 16.4% 29.0%
Weighted 39.0 45.3 12.8 22.6 25.6 34.0
(N) (221) (221) (218) (219) (219) (221)
Managers and
Unweighted 18.4% 59.5% 12.4% 25.2% 28.4% 43.0%
Weighted 29.8 41.6 11.7 22.4 33.1 41.1
(N) (446) (449) (442) (445) (447) (449)
Note: Questions about internal staffing were asked only of informants who stated that current
employees were 'sometimes' promoted or transferred to fill vacancies in a given occupation.

Informants were asked to indicate whether they used each method 'frequently,'
'sometimes,' or 'never.' The sequence of questions was repeated for each of the three
occupations, if the informant stated, in response to a filter question, that the
establishment ever fills vacancies in that occupation with people it already employs.3
Table 1 presents the percentages of informants who answered 'frequently' to the
items in the sequence, separately for the three occupations studied in the NOS. It
shows that formal staffing procedures-seniority and job posting-are most widely
used. The un weighted percentages, which reflect the experience of the typical U.S.
employee (see note 1) show that between three-fifths and two-thirds of internal
vacancies are frequently advertised through posting of lists or circulation of vacancy
notices. The seniority principle is frequently used for almost 40 percent of vacancies
in core jobs, and even for 20 percent of managerial positions.
There is, nonetheless, appreciable use of informal methods in internal staffing.
Only about a tenth of positions are frequently filled by asking the previous occupant
of a position to recommend possible successors, but recruiting for nearly a quarter of
them often draws on other interpersonal referrals. Employers make frequent direct
approaches to candidates for promotion and transfer for about a fifth of core
positions, and for almost 30 percent of managerial jobs. The final column shows that
at least one of the three informal methods is frequently used for more than a third of
core jobs, and for 43 percent of manageriaVadministrative vacancies. Results not
displayed in Table 1 indicate that informal methods were 'never' used for less than
15 percent of jobs studied in the NOS.
The weighted results in Table 1 estimate the distribution of internal recruitment
methods across establishments (see note 1). Percentages for job posting are lower in
Social Capital in Internal Staffing Practices - 189

the weighted results, indicating that this procedure is less often used in the smaller
establishments which become more numerous when the data are weighted.
Conversely, weighted percentages generally exceed the unweighted ones for
seniority and the three informal methods, suggesting that these methods may be
more often used in smaller establishments.
Crosstabulations of the frequency with which establishments use formal and
informal methods of staffing (not shown) reveal negligible associations. The fact
that these associations are not negative indicates that formal and informal methods
do not constitute mutually exclusive approaches: informal channels are sometimes
used alone, and sometimes as a supplement to posting or seniority. Establishments
using one type of informal channel also tend to use others: there are substantial
positive associations linking all three pairs of informal methods in Table 1. 4


We have argued above that staffing methods are chosen in light of their anticipated
information benefits and both administrative and opportunity costs, as well as
internal and external pressures for equity and rationalization in the management of
internal personnel actions. We suggested that as a result of the operation of these
mechanisms, we should observe variation in the use of informal methods across
establishments according to size, affiliation with a larger, multi-site organization,
sectoral location, and the presence of personnel departments, specific job
descriptions, and unions. At the occupational level, our three mechanisms predict
variation in the use of informality across types of occupations, and according to the
presence of training and job ladders. This section presents our findings; descriptions
of our measurements of these explanatory variables appear in the appendix.

Bivariate Differences in Internal Staffing Methods

Table 2 provides a preliminary indication of the relationships between our
organizational and occupational covariates and the use of informal methods of
identifying and selecting candidates for promotion or transfer. The first column
reports percentages of establishment-occupation observations S of a given type for
which informants reported that at least one informal method was used 'frequently.'
The other columns give similar percentages for the three specific informal methods.
The results in Table 2 are broadly consistent with the theoretical ideas
developed above. As establishment size increases, informal methods are less often
used to fill occupational positions; nearly half of observations from establishments
size 10-49 use at least one informal method frequently, but only about 30% of those
from establishments with more than 500 employees do so. About 44% of
observations from independent establishments make frequent use of an informal
method, as compared to a third of those from within multi-site organizations. These
differences reflect the greater visibility of larger and multiple-establishment firms,
as well as size differences in the information benefits, opportunity costs, and
administrative efficiency of staffing methods.
190 - Corporate Social Capital and Liability

Table 2. Bivariate differences in internal staffing methods (all occupations)

Percentage Using Method Frequentlya

Any Incumbent Referrals Direct
Variable Informal Referrals from Others Contact Nb
1-9 38.9% 14.9% 22.7% 30.0% (87-90)
10-49 46.0 9.7 26.5 32.4 (247-250)
50-99 38.2 9.1 22.9 23.2 (142-144)
100-499 34.0 10.6 21.9 19.8 (255-259)
500+ 31.3 8.9 22.0 14.1 (302-307)
Multi-site organization
No 43 .6 9.5 20.9 27.1 (346-351)
Yes 33.8 10.2 27.3 20.4 (696-707)
Private 40.4 12.2 24.3 26.6 (615-622)
Public 32.2 6.9 21.3 17.5 (334-342)
Nonprofit 33.0 6.5 21.3 14.9 (93-94)
Union presence
None 41.0 11.0 25.1 20.6 (520-531)
Some 33.3 9.1 21.2 24.8 (518-523)
Personnel department
No 41.5 9.9 25.1 28.8 (523-530)
Yes 33 .2 10.5 21.3 16.5 (497-506)
Job descriptions
No 43.2 13.2 22.6 34.5 (136-139)
Yes 36.1 9.5 23.1 20.8 (906-920)
No 34.6 9.6 18.7 23.5 (239-243)
Yes 38.1 10.1 24.6 22.5 (791-803)
Multiple levels in
No 39.8 9.9 22.6 24.6 (322-327)
Yes 35.8 10.0 23.3 21.7 (720-731)
Managerial 42.3 12.0 25.0 28.0 (518-525)
Professionalffechnical 34.7 9.2 26.7 15.2 (141-147)
Sales/Service 43.0 8.4 25.2 26.2 (107)
Administrative Support 27.6 10.5 18.6 14.0 (86-89)
Craft 43.3 20.0 26.7 30.0 (30)
Unskilled 22.2 3.1 13.7 13.0 (160-162)
a Based on unweighted NOS data.
b Observations are occupational groups within establishments. Owing to missing data, Ns for the
different methods vary slightly.
Social Capital in Internal Staffing Practices - 191

Informal methods are less often used frequently when an organization has a
personnel department, when most positions have written job descriptions, when an
establishment operates under public or nonprofit auspices rather than in the private,
for-profit sector, or when unions are present. These findings substantiate the
arguments we offered above about staffing strategy and constraints as factors
shaping the selection of staffing practices.
Turning to occupational characteristics, Table 2 shows slight tendencies to make
more use of informal procedures when those selected to fill a vacancy are to receive
formal training. Contrary to our expectations, informal methods are slightly less
likely to be used in occupations with multiple levels, which offer the prospect of
future promotion. As we anticipated, though, the use of informal methods differs
appreciably across kinds of work. Internal staffing decisions for managers,
professional and technical occupations, sales and service occupations, and craft
occupations make use of informal methods substantially more often than those for
clerical (administrative support) and semi- or unskilled occupations. This pattern
suggests that organizations rely on informal methods when filling autonomous,
structurally unique positions in which performance is not readily metered.

Multivariate Analyses
We use ordinal logistic regression to examine the partial effects of organizational
and occupational variables on the frequency with which informal staffing methods
are used. This model is the appropriate choice for the study of ordered dependent
variables like those studied here (Long 1997; Winship and Mare 1984). A positive
coefficient indicates that an increase in the value of the explanatory variable is
associated with a rise in the odds of using a given internal staffing method at all
rather than 'never,' or 'frequently' rather than less often. 6 Table 3 presents the results
of these analyses. 7
Beginning with organizational factors associated with the use of any informal
method of internal staffing, we find statistically significant8 tendencies for
establishments linked to multisite firms, those having personnel departments, and
those in the public rather than the private sector to be less likely to use informal
processes in internal staffing. The odds of using an informal method in multisite
establishments, net of other factors, are 0.69 times as large as in independent
establishments; in public sector establishments, they are 0.56 times as large as in the
private sector. Of particular interest is the finding that the bivariate negative
association of establishment size with informal methods in Table 2 becomes
statistically negligible when we adjust for the other independent variables included.
Thus, the size differences in staffing methods seen in Table 2 are attributable to the
fact that size is associated with factors that have statistically significant coefficients
in Table 3, including the presence of personnel departments, multi site affiliation,
and occupational mix. Contrary to our expectations, the presence of a union does not
seem to reduce the use of informal staffing procedures, net of the other factors we
have controlled. Written job descriptions are negatively associated with the use of
informality in promotions and transfers, but this association is statistically
192 - Corporate Social Capital and Liability

Table 3. Organizational and occupational correlates of informal internal staffing: ordinal logit
coefficients a
Internal Staffing Method
Explanatory Any Informal Incumbent Referrals
Variables Method Referrals from Others Direct Contact
Estabishment size (log) -.002 (.047) -.014 (.048) -.007 (.046) -.002 (.047)
Multisite organization -.373** (.135) .069 (.138) -.190 (.133) -.511 ***
Public secto{ -.573*** (.149) -.520*** (.153) -.421** (.146) -.621***(.150)
Nonprofit sectorC -.240 (.228) -.039 (.227) -.202 (.225) -.154 (.224)
Union presence -.030 (.107) .083 (.109) .038 (.104) .134 (.108)
Personnel department -.345* (.172) -.082 (.176) -.406* (.168) -.549** (.175)
Job descriptions -.213 (.199) -.201 (.204) .030 (.194) -.413* (.200)
Formal training .500** (.158) .210 (.163) .476** (.154) .395* (.158)
Multiple levels -.260 (.142) .051 (.146) .217 (.140) -.138 (.141)
Managerial C 1.042*** (.188) 1.222*** (.206) .366* (.181) 1.015*** (.190)
Professionalffechnicalc .810*** (.241) 1.139*** (.257) .731** (.233) .584* (.238)
Sales/Servicec .850*** (.255) .529 (.273) .683** (.244) .835*** (.253)
Administrative SupportC .216 (.262) .946*** (.282) .052 (.258) .281 (.263)
CraftC .574 (.398) .914* (.412) .346 (.378) .486 (.403)
Threshold parametersb
First cutpoint -1.821 .822 -.370 -1.194
Second cutpoint .644 3.082 1.667 1.230

Log-likelihood -966.89 -918.59 -1046.91 -987.48

N 1014 998 1006 1010
Source: 1991 National Organizations Study. Standard errors given in parentheses.
* p<.05 ** p<.Ol ***p<.OOl
aBased on unweighted NOS data. Observations are occupational groups within establishments.
Owing to missing data, Ns for the different methods vary slightly.
bConstant term is constrained to zero.
cReference category for auspice dummy variables is the private for-profit sector; reference
category for occupation dummy variables is unskilled occupations.

The odds of using an informal staffing method grow by a factor of more than
1.6 when those in an occupation receive formal training, suggesting that employers
seek more detailed information when considering internal candidates for such jobs.
The presence of job ladders (as measured by multiple levels), on the other hand, has
no statistically significant association with informality in internal staffing.
Consistent with our hypotheses, informal procedures are used much more often for
some types of occupations than for others. The findings show that the use of any
informal channel is most common for managerial employees; however, such
Social Capital in Internal Staffing Practices - 193

methods are also more common for professionaVtechnical and sales/service

occupations than for semi- or unskilled ones. As anticipated, the coefficient
contrasting craft and unskilled occupations is positive, but it does not reach
conventional levels of statistical significance.
Several differences in the effects of certain organizational variables across
specific informal methods of identifying and selecting internal candidates are worthy
of note. Differences are more strongly patterned for direct approaches than for either
type of referral. Affiliation with a multi-site organization primarily reduces the
frequency with which direct approaches to candidates are used, as does the presence
of formal job descriptions. In contrast, location in the public sector has a statistically
significant negative coefficient for all three informal methods. The presence of a
personnel department is negatively associated with relying on referrals from
employees other than the position's current incumbent, and with making direct
approaches to candidates; but having a personnel office does not reduce the
frequency with which employees leaving a job are asked to recommend potential

This chapter examined the organizational side of promotion and transfer events, and
the extent to which they involved individual-level social capital, in the form of
interpersonal networks linking selecting officials to candidates. Using data from a
national study of U.S. employers, we found that informal methods of recruitment
and selection-direct approaches to candidates as well as referrals-play a
substantial part in internal staffing actions; they were 'frequently' involved in
promotions and transfers into more than a third of the occupations studied, and
'never' involved in only 15 percent of them. While formal procedures for allocation
of internal vacancies, particularly job posting, were even more widespread, this
nonetheless indicates that many U.S. employers find it advantageous to rely on
interpersonal as well as impersonal sources of information when making internal
staffing decisions.
Our multivariate analyses demonstrated that variation in the use of informal
methods for internal staffing is patterned across types of organizations and types of
work in a fashion to be anticipated, given the differential information benefits,
administrative expenses and opportunity costs we associate with such methods in
distinct situations. Particularly notable are the between-occupation differences,
which suggest that employers seek to activate the social capital embodied within
informal ties to obtain information about candidates for positions involving training,
autonomy, and interpersonal skills, in which performance cannot be readily
assessed. We found it interesting, however, that scale per se was not linked to
informality in staffing, given our covariates.
Our data also indicate that an employer's use of social capital to inform
promotion and transfer decisions is strongly constrained by forces both internal and
external to an organization. The presence of a personnel department substantially
reduces reliance on informal ties in internal labor markets; externally, location in the
public sector has a similar effect. These findings are consistent with institutional
theorizing (e.g., Dobbin et al. 1988) which stresses the maintenance of
194 - Corporate Social Capital and Liability

organizational legitimacy as an important priority for organizations. Here,

legitimacy is presumably enhanced through the use of staffing methods that assure
equity and procedural rationality. We found it somewhat curious, therefore, that the
presence of unions was not similarly associated with a decline in the use of informal
networks in internal staffing decisions.
In closing, we offer two further observations about the part played by social
networks in promotion and transfer processes. The first is that there is some reason
to think that networks are even more involved in internal staffing processes than our
data indicate. We have examined the procedures that establishments use to
disseminate knowledge about vacancies and assemble information about candidates
for intraorganizational mobility. Candidates need not necessarily hear about
opportunities via these channels; even formally distributed information quickly
enters and diffuses through workplace social networks. Suggestive evidence on this
point comes from the 1991 ass, which provided the sampling frame for the NOS.
ass respondents who had held more than one job with their current employers were
asked about how they heard about their present jobs. The most common answers
cited interpersonal contacts: more than half (51 %) of the respondents said that they
had been approached and asked to take their new jobs; nearly a third (31 %) learned
of their new jobs through a supervisor, while 15% referred to a coworker. Less than
a fifth heard about their jobs by virtue of job posting or a vacancy list (17%) or as a
result of unions or seniority (16%).9
Secondly, theoretical understandings of social capital stress the superior
capacity of interpersonal ties to convey certain types of information. Taking this as a
premise, we have invoked the usual 'norms of rationality' (Thompson 1967) to make
predictions about which employers will make use of particular staffing methods. We
enter the caveat, however, that informal methods may be used in ways that diverge
from received theory. Indeed, this may occur for formal methods as well: Reskin
and Hartmann (1986) review cases in which bidding restrictions or seniority
provisions undermined efforts to broaden gender equality in access to desirable jobs
through job posting practices. It seems especially likely, however, that informal
methods of staffing might be used for reasons other than their superior information
benefits. Though as social capital informal ties have the capacity to convey subtle
and nuanced information about the capabilities of candidates, as social liabilities
they simultaneously can convey other, less meritocratic information-about, say,
interpersonal loyalty rather than technical ability. By using interpersonal channels-
that tend to link socially similar persons-when diffusing information about
promotion and transfer opportunities and assembling information about candidates
for them, employers run the risk of excluding those lacking such social capital from
consideration. This at once raises equity concerns (from the individual's side) and
opportunity cost worries (from the employer's).
We think that organizations face a continuing dilemma about how to balance the
information benefits and risks associated with the use of individual social capital in
personnel actions. To the extent that selection into jobs in contemporary
organizations requires information that is accessible only through interpersonal ties,
employers must also provide safeguards against supervisory malfeasance in using
networks, guaranteeing that this takes place in an atmosphere of procedural
Social Capital in Internal Staffing Practices - 195

rationality. An intriguing direction for future research would be to determine what

differences exist between employers relying on pure 'network staffing'-that is,
those that use informal methods alone-and those using networks as an adjunct to
more impersonal methods.


This appendix describes the measurements of independent variables that appear in
Tables 2 and 3. Descriptive statistics reported are for the set of 1091 organization-
occupation observations examined in this chapter, for which internal vacancies were
ever filled using current employees. More details on measures in the NOS appear in
Kalleberg et al. (1996).
Size. Natural logarithm of the number of full-time employees in the
establishment (mean, 4.98; standard deviation, 1.98). Table 2 uses actual size.
Multi-Site Organization. Dummy variable identifying the 67% of observations
from establishments that are part of larger, multiple-establishment organizations.
Personnel Department. Dummy variable identifying the 52% of observations
from establishments that have a separate department or section responsible for
personnel and/or labor relations.
Job Descriptions. Dummy variable identifying the 87% of observations from
establishments in which there are written job descriptions for most jobs.
Union Presence. Scale combining four items indicative of the presence of
unions (mean, 1.50; standard deviation, 0.64).
Public Sector. Dummy variable identifying the 32% of observations from
establishments operated by federal, state, or local governments.
Nonprofit Sector. Dummy variable identifying the 9% of observations from
private, not-for-profit establishments.
Training. Dummy variable identifying the 76% of observations in which those
in an occupation had received formal training within the past two years.
Multiple Levels. Dummy variable identifying the 69% of observations in which
an occupation has more than one level.
Occupational Categories. Core and ass occupations were classified into three-
digit 1980 Census codes by the NOS. We subsequently grouped them into the six
broader classes used in this chapter. No specific occupational title was used by the
NOS when asking about 'managers and administrators,' so we classified all
observations for these occupations into the 'managerial' group. Overall, 50% of our
observations are managerial, 14% are professional, 10% are in sales or service
occupations, 8% are in administrative support occupations, 3% are in craft
occupations, and 15% are in semi- or unskilled occupations.

Data collection and writing were supported by National Science Foundation awards SES-89-11696 and
SBR-95-l17l5. For helpful comments, we are indebted to the editors, Roger Th.AJ. Leenders and Shaul
M. Gabbay.

I. A work. establishment refers to a specific geographic site or address. Some establishments are part of
larger, multi-site firms or organizations. The sample of establishments was drawn as part of a topical
module on 'Organizations and Work.' included in the 1991 General Social Survey (GSS; see Davis and
196 - Corporate Social Capital and Liability

Smith, 1996). In 1991, the GSS interviewed a random sample of 1,531 English-speaking U.S. adults. At
the end of the interview, each employed respondent was asked to give the name, address, and telephone
number of herlhis workplace; married respondents were asked to provide the same information about the
workplaces of their employed spouses. This generated a multiplicity sample in which work
establishments have known, but unequal, probabilities of inclusion; the probability that an establishment
is included in the NOS is proportional to its number of employees. Thus, there are more large
establishments in the NOS than would appear if workplaces were to be drawn at random from some
listing of establishments. The unweighted NOS sample describes work settings from the standpoint of a
typical U.S. employee, since it gives each GSS respondent equal weight. To instead describe the
population of U.S. work establishments, the data must be weighted inversely to workplace size. Most
figures presented here are for the unweighted sample.
2. Owing largely to the clustering entailed in the area probability design of the GSS, some
establishments were sampled more than once. The data reported in this chapter include only one record
for such duplicated cases. Including duplicates, there were 1,127 interview attempts and 727 completions.
3. It did not seem sensible to ask organizational informants about practices that they do not enact. The
fact that questions about internal staffing practices were posed only for occupations that were ever staffed
internally raises the prospect of sample selectivity. The selection criterion-internal filling of a position-is
central to definitions of internal labor markets, and we observe that many of the covariates in our
empirical analyses below (especially size, formalization, and affiliation with multi-site firms) are also
strongly linked to the presence of ILMs (Kalleberg et aI. 1996). Net of the effects of these common
covariates (especially establishment size) on selection into our sample and on staffing methods, we
believe that any association between a propensity to fill vacancies with internal candidates and the use of
staffing procedures is weak, and therefore that any residual selection bias is slight.
4. For incumbent and other references, Goodman and KruskaI's gamma is 0.62; for both types of
references and direct approaches it is 0.51.
5. That is, there may be up to three records for each establishment in this Table-<>ne for each
occupation (core, GSS, or managerial) for which the establishment's informant answered the sequence of
questions about internal staffing.
6. The coefficients can be understood in multiple ways. For an ordinal dependent variable with J
categories, ordinal logistic regression models the natural logarithm of the odds that an outcome will be in
category j or lower, rather than above j, as '1:J - pX, where '1:j is a 'threshold parameter' or 'cutpoint' and X is
a set of independent variables. The log-odds of being in a category above j, rather than j or lower, are
therefore -'1:J - pX. The proportional change in the odds of being in a category above j, rather than in
category j or a lower one, that is associated with a one-unit increase in a given explanatory variable Xi is
the same at any value of j and is equal to expC/3;). Thus, for example, the coefficient of -.573 for public
auspices and the use of any informal method indicates that for public-sector establishments, the odds of
using any informal method 'frequently' rather than 'sometimes' or 'never' are exp(-.573)=O.56 times as
large as they are for private, for-profit establishments. The same proportional factor applies to the odds of
using at least one informal method 'frequently' or 'sometimes' rather than 'never.'
7. Because of the clustering of observations on occupations within establishments, we estimated
multilevel models including between-establishment random effects on the likelihood of using a given
staffing method. The results of these analyses indicated that, net of our covariates, only small and
statistically insignificant components of variance are associated with organizational differences.
Accordingly, we elected to present findings using a simpler and more easily interpretable single-level
8. Statements on statistical significance refer to the .05 level.
9. The percentages reported here total more than 100% because respondents were permitted to mention
more than one source of information when responding to the question. Case bases (Ns) for the
percentages range from 238 to 251.
Getting a Job as a Manager

Ed Boxman

In his seminal study, Granovetter (1995) demonstrated how the job-attainment process
is embedded within social networks. The ensuing research effort and theoretical
discussion left two points unclear. To what extent do people with higher societal
position use informal channels to find a job? Do they receive positive returns in terms
of income? A replication of Granovetter's analysis in a large sample of Dutch
managers at larger companies (n=1402) in 1986-1987 shows that Dutch managers
generally rely on their social contacts to find a job, and they do so more frequently at
higher executive levels. Moreover, using informal job-finding methods leaves them
with higher earnings. Granovetter' s weak-ties argument has been refuted: although
they are the most widely used, finding a job through weak ties does not produce a
higher income level. It is not true that managers rely more on informal contacts later in
their career.
Our own more general hypotheses on social capital have been confirmed.
Managers with more social capital (association memberships and external work
contacts) find a job more frequently through some informal channel. Moreover, they
earn a higher income independent of their human capital. Burt's hypothesis that social
ties enlarge the returns of human capital has been refuted. Human and social capital do
interact in that social contacts help workers to earn more income at any level of human
capital, but the returns of human capital decrease at higher volumes of social relations.

Many members of the workforce in Western societies have found a job through some
kind of informal contact. In the 1980s roughly a third of the employed people in the
198 - Corporate Social Capital and Liability

Netherlands did so, and in the 1990s as many as half did so (Moerbeek, Flap, and
Ultee 1997). Comparable figures for former West Germany and the U.S. in the 1980s
are 42 percent and 59 percent respectively (De Graaf and Flap 1988). According to a
recent study of the French labor market, 32 percent of all the employees used an
informal channel to find their job (Forse 1997), and around 1990 this was also true of
nearly half the employees in Spain (Requena 1991). These facts are at odds with the
official universal ideologies in these countries.
In contrast to the official ideology, popular opinion holds that if you want to get a
job, it helps to mobilize your networks. Networking also gets you a better job.
Furthermore, social relations are important, especially at the top of the societal ladder,
since the people there have business relations who bring them financial and technical
information that can be useful in their current job and also labor market information
that can be useful, inter alia, in learning about job-openings. Because of the often
influential positions of these contacts, they can be instrumental in helping those at the
top of the ladder get similar jobs. However, general social surveys in industrial
countries such as the U.S., Germany, France, Spain, and the Netherlands have
demonstrated again and again that it was precisely the people who are lower on the
ladder who found their jobs through informal channels. In addition, this research has
shown that the use of social contacts rarely brings a better job than the use of more
formal channels or applying directly.
A number of empirical studies of specific occupational groups do, however,
indicate that people with better jobs make extensive use of informal contacts. In his
seminal study Getting a Job, Granovetter (1995) noted that over 60 percent of the
people in higher positions (he studied a group of persons with technical, professional,
or managerial occupation in Newton, Massachusetts, a suburb of Boston) found their
jobs via someone they knew. That was also how they found the better jobs.
Preisendorfer and Voss (1988) described how almost three-quarters (74 percent) of
200 employees at universities and vocational colleges in the former West Germany
found their jobs through some informal channel, although networking did not help
them find a better job-for example a steady full-time job.
Findings like these suggest a U-shaped association between a person's social
status and the use he or she makes of informal contacts in finding a job. This type of
U-shaped association probably remains hidden in random sample surveys of the
general population because those surveys usually contain only a small number of
people in higher positions. Findings like these also make us wonder about the
circumstances under which it is profitable to mobilize contacts to get ajob.
A small research literature has recently emerged on the social networks of
managers, their determinants, and the returns of their networks while doing their job as
a manager. There is a special interest in the occupational group of managers, since
their actions are supposedly more decisive for the fate of enterprises and of the
economy at large. Managers are in the business of entreprendre, bringing people and
other production factors together (Burt 1992: 274}-that is, they are in the business of
networking. One can assume that if they are good at this job, their social networks will
capture not only individual social capital but also ftrm-Ievel social-capital-and that
the firm will benefit-for example, in terms of better survival chances, larger sales
volume, or greater profit. Moreover, these managers will probably also benefit as
Getting a Job as a Manager - 199

individuals and receive a promotion, a higher salary, or other compensation because of

the advantages they bring to their frrm.
This research concentrates more on the specific characteristics of managers'
networks and less on the use and possible social capital returns of these networks,
since data on actual use and outcomes for individual managers-and to an even greater
extent on frrms-are hard to come by. Our contribution adds to this research by
concentrating on how managers and would-be managers use their own social network
in the job-finding process and it deals with the returns of their networks during their
careers as managers.
Case studies with anecdotal material (Mintzberg 1973; Kotter 1982) have
demonstrated that managers usually have large work-related networks. Managers talk
with numerous people at their frrms to keep informed about upcoming business
opportunities as well as potential threats. It is also clear that managers execute their
agendas by mobilizing their social networks.
Other, more quantitative studies allow for more solid conclusions. l Ibarra (1992,
1997) has demonstrated that women often experience a sex-specific division within
their network at work: for friendship and emotional support they turn to their female
colleagues, and for work-related advice and other instrumental help they calIon their
male colleagues. This probably puts them at a disadvantage if vacancies arise at a
higher level, since they then have to compete with men who have more strong-that
is, more multiplex-ties to those men who decide about promotions and filling
vacancies. The result of these processes is that women who aspire to higher positions
within their organizations, come up against a glass ceiling they are unable to break.
Brass (1985a) has shown that for women, this sex-specific division is indeed not
conducive to being promoted into a managerial position.
Yet there do not seem to be large differences in the networks of managers at
different positions or of different sexes (Moore 1992; Burt 1996). Even the kind of
work contacts cited by managers as being the most important or most troublesome are
stable across kinds of managers. Managers as compared to ordinary employees,
however, do have larger and more open networks, with more work colleagues. They
also have stronger ties to their alters, but the alters are less strongly linked to each
other (Carroll and Teo 1996). The fact that there are clear differences between the
networks of managers and employees and no clear differences between the networks
of managers themselves suggests that managers (including female ones) are selected,
inter alia, for their network characteristics, making them more similar to their
Once they have a job as a manager, what are the returns of their network in their
job as a manager? Burt (1992) has demonstrated that male managers with a more
autonomous position within their networks are promoted earlier, whereas female
managers and males who have just entered the higher managerial ranks need a more
cohesive network or even a high-prestige sponsor if they are to succeed in getting
further promotions. In another study Burt (1996) described how these structural holes
(meaning that one's alters are disconnected from each other) contribute to higher
returns for executive managers at an investment banking division of a large American
financial organization. They receive larger bonus compensation payments since their
alters have no alternative for them, but they do have alternatives for their alters.
200 - Corporate Social Capital and Liability

Boxman, De Graaf, and Rap (1991) have showed that a larger, more diverse
external network of Dutch managers at large companies makes for a higher income,
quite independent of the managers' human capital and number of subordinates. In a
replication, Meyerson (1994) claims to have found that strong ties bring higher
compensation for managers at Swedish public firms. Carroll and Teo (1996) have
discovered that a greater number of colleagues in a person's network does not generate
more income for managers, but it does so for other employees. They account for this
by arguing that all managers have a sizable number of colleagues in their core network
and that if they did not, they probably would not last long as a manager.
Apart from Granovetter's small study, the job-finding process of managers and the
role that is played in job finding by their social networks has not been studied
extensively. This state of affairs makes it interesting to inquire more deeply into the
job-finding process of managers and the instrumentality of social networks in their
getting a job. We test a number of hypotheses (mainly taken from Granovetter's study,
as it is the landmark study on getting a job) in a large data set of Dutch managers. We
describe the job-finding process of the managers and how frequently they found their
current job through informal relations and, more specifically, through weak ties. In
addition, we examine whether managers with more social contacts do indeed use these
contacts to get their job. For a number of job-outcomes we also analyze whether there
are positive returns to using or having a better network, i.e. greater social capital.
In the last section of this chapter, we discuss whether an individual's social
network will also provide the firm with social capital.


Within the sociological and economic research literature on labor markets, at least four
traditions have paid attention to the influence of social networks on labor-market
positions. The theoretical background of the research literature on the networks of
managers is similarly heterogeneous, since it has been influenced by human capital
theory, status-attainment theory, structural-organizational ideas, and social capital
The status-attainment research studies the extent to which an individual' s chances
of upward or downward mobility depend more on his or her achieved characteristics
than on ascribed features such as family background. In the course of research, the
awareness has grown that the chances of finding a job depend not only on social
origin, education, and work experience but also on access to the kind of labor market
information that is provided by social networks. The main findings of the general
surveys incorporating these ideas are that the status of the contact person contributes to
the occupational attainment independently of human capital and that part of the
original effect of human capital is in fact to be attributed really to the social resources
used in the job-finding process (Lin, Vaughn and EnseI1981).
The structuralist tradition assumes that there are barriers between different parts of
the labor market, including cleavages between networks of relevant people. To find a
job in more attractive segments, like an internal labor market, individual capacities are
often not enough. One also needs social contacts that feed into this segment.
According to this idea of social closure, outsiders are excluded if company recruitment
Getting a Job as a Manager - 201

occurs along lines of preexisting social networks linked to the workplace-for

example, by way of employee referrals (Grieco 1987; Windolf and Wood 1988). In
addition to the social networks being instrumental as a channel of information,
institutional and organizational factors thus influence social networks at the work
place. Also see Marsden and Gorman (this volume) on internal hiring practices.
The neoclassic economic tradition of labormarket analysis has a number of
varieties-human capital theory, job-search theory, and transaction cost theory.
Human capital theory does not provide much information on the relation between
networks and labor market position. It states that human capital, as indicated by
education and work experience, constitutes the best explanation for attained
occupational position. Job-search theory does offer a suggestion (Stigler 1961, 1962;
for a review of job-search theory, see McKenna 1985; Devine and Kiefer 1991). It
assumes that the collection of information on labormarket opportunities entails certain
costs, like loss of time, money and other opportunities forgone. Various search
strategies have different costs. The people who are the highest achievers are those who
are able and willing to invest in the search for information. The job-search theory does
not devote much attention to the embeddedness of search processes by employees and
employers in the informal social networks of employees and employers (Granovetter
1985). Although the theory starts from the assumption that the use of personal contacts
lowers the search costs if people exchange their information with each other (pooling),
it does not specify the conditions under which the exchange will occur and with what
result. In reality, search processes do not emerge at random in markets of anonymous,
disconnected actors, but within the confines of preexisting networks. Interesting
enough, job-search theory provides a mechanism that could explain the effects of
social networks, but its weakest point is the lack of empirical testing.
A recent addition to this line of thought is transaction cost theory (Williamson
1994). It stresses that most of a [trm's costs are not production costs but transaction
costs. They include not only the ex ante costs of the search for attractive exchange
partners but also those of arriving at an agreement and the ex post costs of having to
enforce an agreement. Networks lower not only the search costs but also those of
contracting and enforcement (Granovetter 1985; Raub and Weesie 1990). Since they
engender mutual trust, denser networks will make the management of business and
labor relations less cumbersome, stabilize cooperation and indirectly improve
performance (see, however, Burt and Knez 1995).
The three traditions described above focus on the matching process between
people and positions. Neo-classic human capital theory, search cost theory and
sociological status-attainment research focus mainly on the role of individuals with
different characteristics. Structuralist theories look for the possible effects of industries
and organizations. Transaction cost ideas are only just beginning to influence
empirical research on the returns of work-related networks in the occupational career.
Network research has accepted suggestions from each of these three traditions in
research on networks and their returns in occupational attainment, and elements of
each can be recognized in studies on the networks of managers and their returns that
were discussed above. Insights from these three traditions come together in
Granovetter's Getting a Job. He explicitly links individual and positional factors. He is
interested in the latent structure of social networks and its influence on labormarket
202 - Corporate Social Capital and Liability

behavior. Although according to Granovetter information is mainly a byproduct of

other social interactions that come without costs, he assumes that informal search
efforts decrease search costs. This is why an employee's and an employer's interests
are served if they search through informal channels, especially weak ties, since they
provide access to information that is more likely to be new (Granovetter 1973).
Granovetter's ideas have constituted an important source for the recent fourth
tradition focused on networks and labormarket chances-social capital theory
(Bourdieu 1981; Flap 1988; Burt 1992). According to this theory, social capital
produces a better life. People who have access to more people and to people who have
more resources (including contacts) and who are more prepared to lend a helping hand
are more successful in goal attainment, including in the labormarket. What is different,
though, is that Granovetter sees networks mainly as a by-product of other kinds of
interaction, while social capital theory suggests that networks are also the result of
people investing in each other. Provided that there is a common future, people' invest
in particular ties and types of networks to the degree that they are instrumental. In
short, social capital is the present value of future help. Social capital theory specifies
the strength of the weak-ties argument in that weak ties do give access to people with
better second-order resources, but those people usually are less prepared to help.
We now present Granovetter's major hypotheses on the relation between
networks, job search, and the attained position, as well as three additional hypotheses
that have been the product of social capital theory proper. We specify Granovetter's
hypotheses and test them consecutively for our large data set of Dutch managers.


HI: Search through social networks decreases search costs.
People are more apt to find a job and to find a job with less effort if they search along
informal lines instead of formal ones, since it enables them to hear about vacancies
earlier. In addition to the advantage of timing-an advertisement informs potentially
everyone at the same time-informal contacts provide more in-depth information on
the particularities of the job in question and everything that goes with it, a kind of
information that cannot be found in advertisements. Employees can also make their
productivity and other capacities known through their social contacts to potential
employers and get an introduction via them. Contact persons feel an obligation not to
recommend people they know for bad jobs or refer poorly qualified people to
employers who are looking to fill a vacancy. Looking at the process from the other
side, employers can lower the costs of recruitment by searching informally for
candidates. They get more in-depth information on the true productivity of potential
candidates more quickly. This more intensive information prevents disappointment on
both sides. Moreover, it saves on the costs of control and training (see Grieco 1987).

H2: Better jobs in terms of income and prestige are found through social contacts.
People who enter the labor market want the best job they can get, and they will
mobilize their resources accordingly, including their social resources. Since social
contacts bring extensive and intensive information and possibly also other kinds of
support, better jobs are found using contacts.
Getting a Job as a Manager - 203

H3: Better jobs are found through weak ties.

Weak ties-ties that are less intensive and less frequent, like those with acquaintances
and colleagues-lead to the best jobs because weak ties can serve as bridges between
different social circles (cf. Granovetter 1973). Information travels across these bridges
over horizontal and hierarchical cleavages between different occupational groups and
levels and leads sooner to people in higher social positions, who are better able to
provide the relevant information on a vacancy or help in some other sense.

H4: Employees at later stages of their careers make more use of their informal ties,
especially their weak ties, to attain jobs.
This hypothesis builds on the above three hypotheses by adding the auxiliary
assumption that people acquire contacts, especially weak ties, on the job and even
more so by changing jobs. Under this assumption, it is logical to expect informal
channels to playa greater role in the later stages of an employee's career (see also
Granovetter 1988: 193).

H5: Employees at higher occupational levels make more use of informal channels
in the job-finding process.
Another plausible auxiliary assumption is that people in higher social positions have
larger networks and networks containing relatively more weak ties. They consequently
make more use of informal ties, especially weaker ones, to find ajob.

H6: Employees with better networks make more use of their contacts in finding a
Hypothesis 1 can be specified using the notion of social capital. If having a greater
number of contacts with alters who are higher placed is instrumental, then people will
employ these resources more frequently in finding a job.

H7: Employees with more social capital will have higher returns of their social
There is no reason to assume the advantages of social relationships stop after a person
has acquired a particular job. The basic idea is that informal social relations to relevant
others not only help get a job but also help while doing a job, and improve the returns
from this job (Han 1996; Flap, Snijders. and Van Winden 1996).2

H8: Employees with more social capital succeed in getting larger returns of their
human capital.
This hypothesis was formulated by Burt (1992), who argues that social capital
provides opportunities for applying one's human capital and thus promotes the returns
of human capital.


Our data set on Dutch managers is unique. Generally, data in the public domain on the
recruitment of managers and their ensuing careers are scanty. To our knowledge there
is no other sample of this size on the labormarket behavior of managers at many
different firms that includes multiple indicators of human and social capital.
204 - Corporate Social Capital and Liability

The actual research, a survey, was conducted in 1986-1987. A questionnaire was

sent to 4000 large companies with fifty or more employees. These 4000 companies
were randomly selected from the total population of 8746 large Dutch companies.
Packets were sent to the companies, each containing one questionnaire. Above the
company's address we printed the job title of a particular manager, for instance: To
the Managing Director of General Motors Netherlands.' We included job titles in five
categories of the most important decision-makers at Dutch companies: 1) managing
director, 2) personnel manager (human resources), 3) commercial manager (sales
manager or purchasing manager), 4) manager production/automation, and 5) financial
manager. For each category, we mailed 800 letters. The letter of introduction requested
that, if the company did not have the job mentioned, the questionnaire be completed
by the personnel manager or the commercial manager. Since at some firms, someone
in a middle-management position answered the questionnaire, these managers were
classified as an additional category. The response rate was 35.2 percent (n=1402). Due
to missing information for some respondents on specific variables, the number of res-
pondents used for the analysis in this chapter is 1369.
The nonresponse is rather high. Although we expected managers of larger
companies to be overrepresented in our sample because--e.g., smaller companies do
not always have a personnel manager-we cannot exactly determine whether this is
the case because we have information only on the size of the establishment our
respondent works at. However, based on information on the whole sample of
companies with more than fiftly employees, we know that 65 percent of them work at
companies with more than 100 employees; 55 percent work at establishments with 100
or more employees. This was to be expected, since it is possible to have large
companies with small establishments but not the other way round.
The key concepts in our theoretical arguments were operationally defined as

Job-Finding Method. The channel through which the current job was attained is
classified in two categories: 1) informal contacts (via relatives, coworkers, employers,
acquaintances, or somebody working at a headhunter's office), and 2) formal channels
(through advertisements and employment agencies). Of the three categories usually
used in labormarket research-formal, informal, and direct application-the third was
unfortunately not included. Rightly or wrongly, the reason for this omission was our
fear of inviting socially desirable answers on the part of the respondent. Saying you
found your job by yourself makes a much better impression than admitting that
somebody helped you. Formal and informal means of finding a job together account
for seven subcategories: 1) employment agency, 2) newspaper advertisement, 3)
information from a relative, 4) information from an acquaintance, 5) information from
a work contact, 6) being asked by the employer, and 7) being asked by someone from
a headhunter's office.
Strength of Tie with a Contact Person. The strength of tie categories were: 1)
strong tie (contact with a relative) and 2) weak tie (contact with a coworker,
acquaintance, employer, or a headhunters). Regrettably, the questionnaire did not
contain the category 'information from a friend,' so we had to work with the
assumption that every relationship with a nonrelative is weak. Although this is
Getting a Job as a Manager - 205

unfortunate, the hypotheses on the differential effects of strong and weak ties can still
be tested since family is generally considered to be the strongest of all the existing
types of ties.
Human Capital. Human capital was measured by two indicators-years of formal
education and work experience. The first indicator simply consists of the total number
of years at school. The second consists of the number of years worked after school.
Social Capital. Social capital was proxied by 1) the amount of work contacts with
people in other organizations, particularly people with the same level of education, the
same position at another company, and with people with many subordinates, 2) the
number of association memberships-for instance, Rotary or Lions, professional
associations, frequency of attending receptions (in five categories, ranging from no
memberships to four or more memberships), and 3) the number of family contacts in
managerial positions. The number of work contacts was actually measured on a
Mokken-scale (H=0.61; rho=O.79) (Niem611er, Van Schuur, and Stokman 1980). To
test hypothesis 7, we used the first two indicators-work contacts and memberships-
computed factor scores on social capital for all the managers, and divided the resulting
scale in four groups of comparable size (339 with hardly any social capital, and then
420, 304, and 296 managers with increasing amounts of social capital).
Income. Income was measured in gross annual income in Dutch guilders, with the
following categories: 1) less than Fl. 50000, 2) F1. 50000-70000, 3) F1. 70000-
100000, 4) Fl. 100000-150000, and 5) more than Fl. 150000. One Dutch florin or
guilder (1998) is worth about U.S. $0.50. For the analyses, we used the logarithm (In)
of the average category income to create a normal distribution. Except for the function
level variable, all other variables are quite normally distributed.
Function Level. Function level was measured by the logarithm (In) of the number
of a manager's direct and indirect subordinates, which takes into account differences
in the positions of managers in organizations of different sizes. It provides more
information than just the job title. For example, a commercial manager at a large
multinational is usually much higher functionally than a director at a smaller firm.
Job Satisfaction. Job satisfaction was measured by constructing a scale asking for
level of satisfaction with several aspects of the current job (H=0.41; rho=O.86).
Perceived Chances of Mobility. Perceived chances of mobility were measured
with a scale consisting of three items (H=O.51; rho=O.66).
Quality of Information on the Labor Market. This variable was established with a
scale of items on information on relevant vacancies, working conditions, job contents,
and possibilities of doing similar work at other organizations (Mokken-procedure on
four items produced a scale with H=O.60; rho=O.75).
Importance of Advertisements. The measurement of importance of
advertisements as a source of information on vacancies is straightforward.


Our research shows that personal contacts are a very important instrument for
managers and would-be managers not only to get a job, but also to get a better job.
0-03 N
- ~
Table 1. Way offmding currentjob for Dutch managers oflarger companies
Managing Personnel Commercial Manager Financial Middle
Directors managers managers production/ managers management ~

automation §:
1. Employment bureau 1% 1% 1% 1% 3% 4% Q
2. Respond to an advertisement 18% 28% 34% 32% 46% 38% ~
in daily /weekly
3. Advertisement in professionaljournal 7% 7% 5% 11% 5% 9"10 ~
4. Company-intemal advertisement 0% 3% 1% 0% 1% 0%

5. Family 12% 3% 7% 6% 4% 4%

6. Acquaintance 3% 7% 4% 4% 5% 4%

7. Work contact 6% 9% 4% 6% 4% 4%

8. Asked by employer 49% 36% 40% 40% 27% 30%

9. Asked by an employment agency 6% 9"10 4% 1% 4% 7%

Total percentage: 100% 100% 100% 100% 100% 100%

N (1369) (517) (167) (232) (278) (104) (71)
Getting a Job as a Manager - 207

Table 2. Network characteristics and information on the labor market among Dutch managers of
larger companies, 1987
Importance of
Information about advertisements for Perceived
labormarket information about mobility
opportunities the labormarket chances
Number of Tc= .08** Tc=-.l0** Tc= .06*
memberships (n = 1394) (n = 1347) (n = 1162)

Number of Tc= .25** Tc=-.l0** Tc= .13**

work contacts (n = 1380) (n = 1340) (n = 1149)

Number of Tc= .07** Tc= .OI Tc= .00

family-contacts (n = 1371) (n = 1330) (n=1143)
Kendall's Tau-test: * statistically significant < .05; ** statistically significant < .01

Sixty-two percent of the managers in our study did find their current job through some
kind of informal tie, and up to 75 percent of the managing directors found their current
job that way (Tables 1 and 4). The numbers are somewhat higher than those found by
Granovetter in his research (58 percent), but his figures relate to another country and
another time. Moreover, his respondents are not a random sample (Granovetter 1995:
7-10) and he did not inquire into the category 'asked by employer,' which is how,
according to our research, most managers (40 percent) in the Netherlands get ajob.
Granovetter (1995 : 19) reports that 14.8 percent of the managers he studied found
their job through direct application. The kind of frrms worked in by the managers who
were studied by Granovetter is unclear. Our study focuses on managers of firms with
fifty employees or more. Four out of ten of our respondents, and half of our managing
directors, stated that they were asked personally by their current employers to take the
job. Although around one-third of Dutch managers got their jobs through
advertisements, advertisements do not seem very important to Dutch managers, as 66
percent of our respondents do not think they are relevant as a source of information
(see Tables 1 and 2). In the recent National Organizations Study, 50 percent of all the
employers in the U.S. said they used advertisements when they recruit managers from
outside (Marsden 1995: 138). Referrals from current employees as well as referrals
from business or professional contacts are used less often to recruit new managers
from the outside. One difficulty interpreting the figures from that study is that
employers were not responding to a question about an actual hiring but about what
they typically do when they want to fill a vacancy through external hiring. Moreover,
employers often use two or more recruitment methods simultaneously.3
To determine whether social networks lower the costs of search (Hypothesis 1),
we looked into the influence of networks on the amount of labormarket information,
the perceived importance of advertisements, and the perceived chances of upward
mobility. If it is cheaper to acquire information on the labor market through informal
channels, then people with more extensive and diverse social networks should have
more labormarket information, and they should also perceive more opportunities for
mobility. Advertisements would also probably be less important to them. Table 2
208 - Corporate Social Capital and Liability

shows that people who have more work-related contacts, attend receptions and other
social gatherings more frequently, are member of more associations and have more
family members in managerial positions do indeed have more information about the
labor market. In addition, they also have a lower opinion of advertisements as a source
of relevant information on vacancies, and they see better chances for themselves on the
labor market. The associations found also make clear why financial managers and
middle managers have a greater use for formal channels than informal ones (see
Tables 1 and 4); they have fewer contacts at other companies and they attend fewer
receptions and the like than do managing directors or commercial managers. Although
the indicators are only proxies for search costs, our results nevertheless confirm the
hypothesis that networks lower the costs of job search.
The hypothesis on better placement through social networks (Hypothesis 2) has
been corroborated by our findings. Table 3 shows a statistically significant4 positive
association between informal ways of finding a job and income earned, level of the
attained job, number of subordinates, and job satisfaction. If a managerial job is found
through some kind of informal contact, this job will be at a higher level, bring a better
income, entail responsibility for more subordinates, and give greater job satisfaction.
Multiple regression analysis of income on education, work experience and use of
personal contacts for finding a job has resulted in the following statistically
significantly beta-coefficients: education = .42, work experience = .35, and use of
informal contact to find the job .13 (R2 .20). =
Hypothesis 3 on the strength of weak ties fared less well when confronted with
the data. According to Table 3, it does not find support in our data. There is no
statistically significant relation between the strength of a tie with the contact person
and the income acquired (Tallc = 0.03). Moreover, managers who found their job
through a weak tie are also not more satisfied with their job, nor do they have more
subordinates or higher-level management positions. Yet, as can be seen from Tables 1
and 4, strong ties are rarely used to find a managerial job. Strong ties are more often
used at smaller firms (see Table 4), a fact that is probably related to the fact that small
firms are more often family finns.
The test of the career-cycle hypothesis (Hypothesis 4) shows no statistically
significant associations between a more advanced career and a greater use of informal
ties in general or weak ties in particular. The third panel in Table 4 shows that work
experience also does not have a statistically significant effect on the use of informal
ties in general nor on the use of weak ties in particular.

Table 3. Information channels used by Dutch managers of larger companies and

characteristics of the job

Attained Number of Work

PQsition Income subordinates satisfaction
Informal contacts Tc = .21 ** Tc = .15** Tc=.l2** Tc= .09**
vs. formal channels (n = 1364) (n = 1364) (n = 1359) (n = 1368)

Non-family Tc= .01 Tc = .03 Tc=.OO Tc= .02

vs. family (n= 843) (n = 843) (n= 840) (n = 844)
Kendall's Tau-test: ** statistically significant < .01.
Getting a Job as a Manager - 209

Concerning Granovetter's hypothesis on the effect of the job level on search

behavior (Hypothesis 5), more positive findings can be reported. Table 4 demonstrates
that managing directors usually make more ample use of personal contacts to find a
job than lower-level managers. Personnel managers and commercial managers are
special in that, together with managing directors, they have more work-related contacts
at other companies and professional associations. For example, 47 percent of the
personnel managers are members of a professional association and 44 percent have
contacts with a professional recruitment agency. Most of the commercial managers (67
percent) state that they find advertisements unimportant as a source of information
because they hear that information from others. The relation we have noted between
the executive level and the use of weak ties is contrary to Granovetter's hypothesis.
Although weak ties are more important to most managers than strong ones as an
avenue for locating a job, they seem to be especially important to people in lower
management positions.
We have also examined some special categories of managers. Managers with
more human capital in the form of education generally make more use of informal ties.
This human capital effect is even more clear in their use of weak ties. No statistically
significant differences have emerged between male and female managers in how they
acquired their present position. But since women are a very small part (4 percent) of
our research group, not much weight can be given to this statistically nonsignificant
difference. Of greater importance seem to be our findings on how a job was found by
managers in large and in small establishments, combined with what was found on
internal and external recruitment. Table 4 shows that the size of the establishment
combined with whether recruitment has been internal or external makes a statistically
significant difference. Managers who change firms and start at a smaller establishment
of another frrm make more use of social contacts than those who are going to work at a
smaller establishment of their own frrm. One explanation could be that companies with
large establishments have ample relations with small suppliers or organizations
servicing them. These contacts probably promote informal mobility among firms (see
also the chapter by Higgins and Nohria).
The next question about the job-finding process is whether people with more
social relationships have also more often found their present job through an informal
channel. The answer to this question is important, not only as a preliminary test of the
network-as-resource argument (Hypothesis 6) but also because a positive finding
would indicate that our implicit assumption on the causal order of social capital and
job outcomes has some truth to it-ties are used to find jobs and attain a decent
income, rather than that ties are the result of a job and a particular labormarket
position). This validation is needed because the research design was cross-sectional,
meaning that the respondents' social capital was not measured before they entered
their current job but that their social capital, as well as their human capital and income,
were all measured in the same period.
Table 5 shows that managers with an average or large number of work contacts at
other frrms have more often reached their current positions via informal channels.
The same can be observed of managers with relatively numerous club and association
memberships. Although the differences are not very large-the extreme categories
differ only by 10 percent-they are statistically significant. This does not hold true for
210 - Corporate Social Capital and Liability

Table 4. Group-specific differences in employment of social contact in finding a job among Dutch
managers oflarger companies

Percentage employing social contacts Percentage

Total external internal total
Population recruittnent recruittnent population
(N=1402) (N=776) (N=618) N=1402)
Women 63 62 67 94
Men 62 54 71 88
Formal education
Primary-extended primary 63 63 64 79
Grammar 67 **) 54 **) 79 **) 83 **)
Higher vocational 54 48 64 91
University 71 69 75 94
Work experience
0-10 years 58 50 73 88
11- 15 years 48 36 67 89
16-25 years 62 55 72 90
26-50 years 63 58 70 86
Size offirm or establishment
< 100 employees 64 64 **) 74 84 **)
> 100 employees 60 52 70 92
Managing director 74 72 76 84
Personnel manager 62 **) 55**) 69 89
Commercial manager 59 50 69 94
Manager production/automation 57 49 73 92
Financial manager 45 40 60 90
Middle manager 49 49 52 91
** statistical significance of chi -test <.01.

the association between the number of managers in one's own family and the use of
informal job entrance means (this absent association is not shown here). So, we have
found some support for the network-as-resource argument. These results reinforce the
confidence we have in our indicators of social capital.
The final question on the returns on social capital on the job (Hypothesis 7) and
the related one on the relative returns of human compared to those of social capital, is
answered in Table 6. This Table presents the average annual incomes of managers in
four categories of human and social capital. More social capital clearly produces a
higher income. There is also a clear positive income effect of having a better
Burt's hypothesis that social capital enlarges the returns of human capital
(Hypothesis 8) was refuted. An examination of successive rows of the same Table
Getting a Job as a Manager - 211

Table 5. Proportion of Dutch managers finding a job through social contacts by two indicators of
social resources, 1987 (n=1359)

Percent Number of F-test

informal valid cases (significance)
Work contacts with managers
in other organizations
1. none 54.6 240
2. very few 55.4 276
3. moderate 67.9 274
4. many 65.0 546
(1336) 5.57 (.0008)
Memberships in clubs.
professional organizations. etc.
1. none 63.3 128
2. one 54.9 446
3. two 63.8 423
4. three 66.3 252
5. four or more 71.3 87
(\336) 3.80 (.0043)

shows that human capital and social capital interact: social relations help to earn more
income at any level of human capital. The returns of social capital are about equal for
all the educational categories but, as is shown in the successive columns of Table 6,
the returns of human capital in terms of income decrease at higher volumes of social
relations. The same pattern emerges, although somewhat less clearly, if human capital
is measured by years of work experience: social capital does not multiply the returns
of human capital. This is particularly the case for the managers with the most social
capital. The difference in mean annual income between managers with primary
school education and those with an university education is smallest for the managers
with the lar~est social network (for more intricate analyses, see Boxman, de Graaf, and
Flap 1991).


Our questions addressed the extent to which managers employ informal contacts to
find their job and whether informal contacts benefit managers in terms of income and
the like. Our research among Dutch managers made it clear that managers generally
rely on social contacts to find their jobs, especially at the higher executive levels. This
finding gives further credence to the conjecture of a U-shaped association between job
level and the use of social contact, a conjecture that explains the apparent contradiction
between general social surveys that establish a negative association between job level
and the use of informal job-finding methods, and our finding that managers, a
well-defined group of high-status occupations, mainly get their jobs through social
212 - Corporate Social Capital and Liability

Table 6. Average gross yearly income of Dutch managers of larger companies in Dutch guilders (x
1000) by social capital and formal education. 1987 (n=1359)

Social capital
low high
Formal education (I) (2) (3) (4)
1. Primary/low vocational 73 (n=45) 86 (n=41) 88 (n=26) 128 (n=18)
2. Extended primary-grammar 81 (n=llO) 100 (n=107) 104 (n=80) 114 (n=75)
3. Higher vocational 91 (n=14O) 99 (n=206) 104 (n=127) 117 (n=136)
4. University 108 (n=44) 132 (n=66) 132 (n=71) 142 (n=67)

Dutch managers do receive higher earnings if they have used informal job-finding
methods. and their work satisfaction is higher as well. Moreover, the ones at higher
executive levels have more frequently found their job informally. Contrary to
Granovetter's hypothesis, weak ties do not bring better jobs, although they are the
most important way managers have found their present job. His ideas that weaker ties
are used most widely by higher-level managers and that managers at more advanced
stages of their career make more use of informal job finding methods have also been
The specifications of Granovetter's hypotheses with the help of the social capital
theory have held up well in our empirical analysis. Managers equipped with greater
number of social ties (more memberships and external work contacts) make more
ample use of their social contacts to find a job. In their jobs as managers, these
contacts provide them with social capital as they earn a higher income if they have
more social ties. Social capital and human capital both contribute to a higher income.
Burt's argument that social capital multiply the returns of human capital has, however,
been refuted. Human capital and social ties interact in their contribution in a particular
way to income: social ties bring the social capital of higher returns at any level of
human capital, but for managers with a larger volume of social ties, more human
capital does not add to their income. Human capital and social capital are only partly
interchangeable as regards earning an income as a manager: social capital can act as a
substitute for human capital, but not the other way round. Contrary to the hypothesis
forwarded by Burt (1992), human capital is not most valuable for people with the most
social contacts but for those who are practically without any social resources at all; it is
least valuable for managers with many social resources. Elsewhere we have
established that human capital produces social capital, but the effect is not very strong
(Boxman, De Graaf, and Flap 1991).
Apart from the confirmation that informal searches lower the costs of collecting
labormarket information, there is more direct evidence to support the idea that is at the
basis of most of the arguments on networks and labor market chances: managers with
larger and better networks are better informed about the labor market and have a lower
opinion of the value of advertisements.
A number of our findings call for further comments. The refutation of the weak
ties argument stands out, or rather the question why. Of course our operationalization
of strong and weak ties may be held responsible for the absence of any positive effects
of weak ties in our study. But one should bear in mind that according to other studies,
including general social surveys among the workforce, informal ties do not always
Getting a Job as a Manager - 213

lead to better jobs for every group, and weak ties hardly ever do. For a review, see
Flap (1991), Breiger (1995), and Granovetter's (1995) afterthoughts in the second
edition of his 1974 book.
Probably the answer should be sought in the fact that there are structural and
institutional restrictions that have influenced the search process as well as in the search
outcomes. Important examples of disturbing factors include the situation of contact
persons who act as go-betweens, the search and recruitment behavior of companies
that search for candidate managers, and the situation in a particular sector of the labor
market. As to the first factor, it is an empirically well-established relationship in
studies on networks and social stratification that contact persons in higher social
positions bring better returns. So perhaps the determining factor is not the informal tie
as such or the strength of a tie but the person at the other end-what he stands for or
the resources he or she provides access to. For example, individuals of higher social
standing signal higher value: perhaps they are more trustworthy as judges of quality in
others, or perhaps they do indeed have the necessary contacts with the employer who
does the hiring and is looking for a candidate.
Furthermore, recruitment behavior differs between organizations and sectors of
the labor market, depending on tradition or on what is needed on the job. A particular
kind of search behavior by a job candidate-for example, an informal search-might
have quite different outcomes, especially when a job is such that the person who
occupies that job can do a lot of damage, employers recruit informally and preferably
through stronger ties, thus enlarging the returns on informal search. For these jobs,
they need more subtle information on the quality of a candidate to lower the risks of
damage (Flap and Boxman 1996).6 By hiring someone informally, the employer has
some assurance that the contact person will somehow vouch for the individual he has
recommended and that the newly hired person will get along with sitting personnel.
Tacit skills needed at the job will also be more easily transferred to the newcomer if
there is some preexisting tie (cf. Grieco 1987; Marsden 1995). This explanation seems
appropriate for the job-finding process of managers, especially given that 40 percent of
the managers and about 50 percent of the managing directors in our sample were asked
to apply for the job by their employer.
The value of informal search, or for that matter the value of a tie is contingent.
Employers should recruit informally. If contact persons are involved, the contacts
should have relevant information or contacts to still other people who do, and they
should be prepared to share that information. Granovetter sometimes gives the
impression that news flows along the ties in a network like water in a system of canals
(cf. Frenzen and Nakamoto 1993). But contact persons are not always prepared to pass
on job information to anyone they know because in doing so they vouch for the person
they give this information to.
The career hypothesis has also been refuted. As to why managers at later stages of
their career do not make greater use of their social contacts to find a job, it is
conceivable that the number of relevant contacts does not increase with the number of
years at the job. This can be the case when, for example, the job does not bring with it
many contacts, the organization operates in a market segment with fewer contacts, or
the employee has been working too long at the same company thereby ossifying his or
her network so that fewer persons elsewhere know about the employee's qualities (cf.
214 - Corporate Social Capital and Liability

Granovetter 1988). Yet another possibility is that for certain higher-level occupations,
the specific labor market is small and circumscribed. The people in that market all
know each other and the readiness to provide information about a particular vacancy to
others might decline if a would-be information provider fancies that job himself
(Spector 1973).
Much still remains to be explored in research on the instrumental role of networks
in the life of managers. For example, longitudinal studies could establish in greater
detail the extent to which a particular network comes with the job or is a prerequisite
for getting and doing the job. Another question that might be answered with
longitudinal data is whether individuals with more social capital hop faster from one
job to another than those with less social capital.
A second item on the research agenda is the existence of organizational and
institutional variation, which might make for different returns of social capital. The
instrumental value of social relations is contingent partly on the institutional and
organizational context and partly on the strategy chosen by the management of the
firm. Our analysis has been practically devoid of any institutional context. Elsewhere
(Boxman, De Graaf, and Flap 1991) we did check for company size, number of
subordinates, and market sector (manufacturing or service), but that did not greatly
alter the results presented above. Nevertheless, an important question we only touched
on is: how do institutional conditions alter these effects? Bauer and Bertin-Mourot
(1991) argue that in France, state-controlled firms and financial institutions often
perform more poorly than do private banks because they parachute into top positions
managers who are from outside and who lack ties to the people working at the firm or
to relevant business partners outside the firm. In fact, specific institutional
arrangements, such as appointment procedures and rules of succession, are responsible
for this.
Most importantly perhaps, future efforts should concentrate on the mechanism
that accounts for the social capital returns on social structure. What is the mechanism?
There are many ways for social connections to provide advantages resulting in a
higher income and not just at the moment of job-finding. Is it scarce information, for
example, about the situation in the particular market the company operates on? Is it
learning from the good and bad examples of other managers at other companies? Or
should we look at the recruitment process? How do employers recruit managers? Do
they 'buy' managers who are rich in social networks? Do they have more trust in the
quality of a managers if referrals come from particular others? Do certain networks
keep others from opportunistic short-term profit-making (Meyerson 1994)? Do
managers indeed experience higher rates of return on their social contacts as a reward
for the higher returns they and their contacts bring to the firm? Or are some relations
more helpful in doing a job while others are needed to get a better compensation, at
least partly independently of how the job is done (cf. Belliveau, O'Reilly, and Wade
1996; Lazega, this volume)?7
All this brings up the question we referred to above-the question of the
relationship between individual-level and firm-level social capital. Managers usually
use their own networks to promote company goals. D'Aveni (1991) demonstrated that
managers with an extensive informal network in the business community can help
ward off business failure and bankruptcy. But the agency problem of managers who
Getting a Job as a Manager - 215

act opportunistically and misuse corporate social structure to advance their own goals
to the detriment of the goals of the company they work for (cf. Meyerson 1994) makes
it clear that individual-level and fIrm-level social capital are not identical. The latter is
also not a simple aggregate of the former (see Leenders and Gabbay, this volume;
Pennings and Lee, this volume) because the collective good aspect of corporate social
capital creates the danger of underinvestment in corporate social networks. Managers
sometimes do not share business and other work contacts with each other because they
are competing for status, money, or other company bound rewards. Property rights to
social capital are unclear, as can be seen when employees are lured away by a
competing fInn, and take along their contacts with clients or suppliers. This is done,
for example, by account managers at investment banks (cf. Eccles and Crane 1988), or
by editors at publishing houses, who take along the authors whose work they have
edited in the past (Powell 1984). Many companies have formal and informal rules
forcing managers and other employees to share their contacts with the other employees
(cf. Lazega, this volume).
It cannot be denied that companies do benefIt from the private part of a manager's
network as well. It was recently demonstrated by Podolny and Baron (1997) that, for
managers, having a tight core network of persons with whom to consult in the event of
personal problems is conducive to promotion. Earlier research has shown that married
managers are evaluated by their superiors as being more productive and are given
higher functions (Korenman and Neumark 1991). A wife who does not have ajob and
can devote more time to supporting her husband is especially advantageous to
individual managers (Pfeffer and Ross 1982}-and probably also to some extent to the
company he works for.
In a wider sociological context, it is also interesting to contemplate the
possibility that the networks of managers help them reproduce the positions of their
family group or class. Wright and Cho (1992) demonstrate that the lines of authority
are quite permeable, at least as far as friendships between managers and
nonmanagers are concerned. These lines are far less divisive than those of property
and expertise. Wright and Cho account for this by noting the many opportunities for
interaction that are forced on managers and other employees while at the workplace.
In what could be called a semiquantitative, comparative study, Bourdieu and De
Saint Martin (1978) describe how different factions of the French elite-either
connected to state-owned or state-controlled enterprises or to large private fIrms
largely controlled by a small number of families-try to preserve and reproduce
their status and social capital in the next generation by way of various strategies. For
the former, educational credentials plus friendly connections with other highly
placed offIcials are their main resource. The latter typically try to hold on to their
position by having more children and having them marry children from other
families in the private faction of big business. 8

We would like to express our gratitude to P. de Graaf and 1. Schmidt for contributions to earlier versions
of this chapter. Vedior Personnel and Advice 8.V. at Almere Haven allowed us to collect the data. Please
address all correspondence to H. flap. ICSlDepartment of Sociology. Utrecht University. Heidelberglaan
1.3508 TC Utrecht. Email: h.flap@fsw.ruu.nl.
216 - Corporate Social Capital and Liability

I. A large literature exists on the networks of managers filed under the heading 'interlocking
directorates.' With few exceptions, the numerous quantitative studies of interlocking directorates barely
touch on the importance of networks of individual managers. Stokrnan, Van der Knoop, and Wasseur
(1988) is such an exception. Their study of large corporations in the Netherlands suggests that the pattern
of interlocking between large firms and banks is not only the outcome of corporate actors looking after
their interests by using their directors and CEOs, but also the outcome of the actions of individual
managers, who, looking after their own interests, create and use their own network.
2. As was noted above, our dataset does not include information about direct applications. In his study
on the job-finding process of professional, technical, and managerial workers, Granovetter (1995: 19)
reports that 14.8 percent of the people in the managerial category (there were only 81 managers in his
sample of 282) succeeded in finding their current job through direct application. For a discussion of the
difficulties under the heading 'direct application,' see Granovetter (1995: 154-156). There are indications
that direct application brings in less income compared to the use of informal channels (see Faase 1980).
3. If we combine informal job-finding methods, human capital, and social capital into one multivariate
analysis, all three of them have a statistically significant effect on the income managers attain.
4. Statements about statistical significance refer to the .05 level.
5. Their National Organizations Study in the U.S. showed that traditions within a particular firm seem
to be a stronger detenninant of recruitment methods than characteristics of the firm or the occupation
(Marsden 1995: 149).
6. There is also the possibility that the strength of a tie is not measured correctly if one equates
colleagues with weak ties or for that matter friends with strong ties. In our research on how people got
ahead in their occupational careers in former East Germany, people classified 24 percent of their ties to
workrnates as strong and 15 percent of their friendship ties as weak (VOlker and flap 1998).
7. These questions lead to a wide range of other questions. For example, what do the networks of
managers mean to the goal achievement and performance of their department or of the company as a
whole (e.g., O'Aveni 1991)? Are different types of networks needed to accomplish different company
goals (Briider! and Preisendorfer 1997)? And is social capital also managerial capital-that is, can it be
willfully created and can it be steered (Kanter and Eccles 1992)? For a review of these questions, see
flap, Bulder and Volker (1998) and Leenders and Gabbay (this volume).
8. A somewhat similar idea of Bourdieu is that there is a division between different types of jobs.
Higher jobs, especially, differ according to whether they belong to the economic domain or the cultural
domain. Earnings advantages are likely to be found among people who choose education and occupations
in the sector of their origin and in occupations where the criteria for measuring work performance are
unclear. Recent research by Hansen (1996) in Norway demonstrates that this indeed the case and is also
true for managers in the cultural and in economic domains.
The Changing Value of Social Capital
in an Expanding Social System:

Lawyers in the Chicago Bar, 1975
and 1995 12
Rebecca L. Sandefur
Edward O. Laumann
John P. Heinz

Social capital is 'some aspect of a social structure' (Coleman 1990: 302) that acts as
a resource that individuals may appropriate and use for their own purposes. In this
paper, we examine the economic value of ties to local professional elites:
specifically, the income returns to Chicago lawyers of contacts among the elite of
the Chicago bar. Contact with the elite of the bar represents a channel through which
rank and file lawyers may 'tap in' to the social structure of the bar and acquire
valuable resources. The information and influence lawyers access through contacts
with notables are properties of the corporate organization of the bar, and, as such,
are benefits of corporate social capital. We briefly outline a theory of social capital
and suggest ways in which elite ties may act as social capital. We then discuss
changes in the social organization of the bar and suggest how these changes may
affect the value of the social capital represented in elite contacts. We reason that
acquaintance with elites will become more valuable because of the relative scarcity
of such contacts in larger social systems. In analyses of factors affecting lawyers'
incomes, we find evidence consistent with the hypothesis that ties to elite system
members are more valuable in a larger system.

In addition to the obvious reasons for hiring a lawyer-i.e., to acquire expertise
about the law and its procedures-lawyers are sources of a broader range of
information about the community or the business environment, and they help
corporate clients span the boundaries between their own organizations and other
organizations. Lawyers are not the only agents employed by corporations for the
218 - Corporate Social Capital and Liability

latter purposes, of course, but lawyers make an important contribution to these

functions. When a business hires an attorney or a law firm, it gains access to
courtroom skills and legal knowledge, which are aspects of the lawyers' human
capital, and it also gains access to the lawyers' social capital. One component of this
social capital is lawyers' connections to other lawyers, both within and outside their
own firms (cf. Lazega, this volume). Contacts with other lawyers provide useful
information about opportunities, contingencies and strategies, and they give access
to persons who may be willing to use their influence to benefit the lawyer, whether
by dropping the lawyer's name in tones of approval and regard or by intervening
more directly on the lawyer's behalf. These connections may benefit the lawyers'
clients, as well as the lawyers themselves.
This paper considers the returns that accrue to a particular form of social
relationship, ties to local professional elites, held by lawyers in the bar of a major
American city. 1 Our analysis builds on the fact that these urban lawyers constitute a
large, loosely-bounded social system. The social system of the bar differs from the
social system of a large formal organization such as a large corporation in the
distinctness of its boundaries; but, the bar is not unlike many large corporations in
the contemporary United States in its size and in the diversity of work that is
conducted within its boundaries (cf. Burt's (1992) study of a U.S. high technology
firm that is 'the size of a small city'). At the same time that the private practice
lawyers we consider in this paper are members of the corpus of the bar, they are
themselves members of corporate formal organizations: law firms ranging in size
from a solo practitioner and her secretary to large firms employing hundreds of
lawyers in offices around the country. The social capital we will consider is social
capital from sources outside the lawyer's office, but it is social capital indigenous to
the social system of the local profession.
Lawyers who possess valuable social capital, providing access to information
and ties to influential others, should benefit from this capital in two ways that will be
reflected in the rewards that accrue to their work. First, lawyers with superior social
connections should have an advantage in attracting clients, since they have the
benefit both of access to information about opportunities and of contacts potentially
willing to refer business to them. Second, lawyers with superior social capital should
be more valuable to their clients. Access to well-placed contacts gives lawyers
access to resources useful not only in finding and courting new clients, but also in
dealing with the legal matters of existing clients. Successful legal strategy requires
not just technical knowledge of the law, but also other salient knowledge, such as
knowledge of others' personalities, of others' likely strategies, of others' history of
past behavior in similar circumstances. Such extra-legal knowledge aids lawyers in
outwitting and outmaneuvering their opponents in the adversarial contests carried
out in the texts of legal briefs and in conference rooms and court rooms. Well-
connected lawyers whose associates can pass along such useful extra-legal
knowledge should be more valuable to their clients than equally competent lawyers
without access to such useful contacts.
Below, we briefly outline a theory of social capital that identifies important
differentiations in its forms by examining the mechanism through which it has its
effects and the way in which it is acquired. In light of these observations, we suggest
The Changing Value of Social Capital in an Expanding Social System - 219

specific ways in which ties to the elites of one's profession may act as social capital.
We investigate whether ties to elites are valuable to lawyers in increasing their
incomes net of other characteristics and organizational contexts which might aid
them in attracting clients and in being superior lawyers. We then ask whether the
value of the social capital of elites ties changes when the social system in which it is
embedded changes.

Our conceptualization of social capital draws from the work of James S. Coleman
(1988, 1990). In Coleman's formulation, social capital is appropriable social
structure; it is 'some aspect of a social structure' that acts as a resource that
individuals may use for their own purposes (Coleman 1990: 203). Such social
structure may exist in relatively discrete forms, such as organizations, or in more
diffuse forms, such as extended families, communities, or other loosely bounded
social systems. Always, it consists of relationships. These relationships may be
components of formal organization, such as the relationships of classmate,
department head, co-worker, and instructor; or, the relationships that constitute
social structure may be defined by other criteria, such as the relationships of
neighbor, lover, uncle, co-conspirator, and friend-of-a-friend. These relationships
may be characterized by both their structural form and the content that inheres in
them; and, aspects of both their form and their content will condition their
productivity as social capital.
An individual's stock of social capital consists of the collection and pattern of
productive relationships in which she is involved and to which she has direct access,
and further of the location and patterning of her associations in larger social space.
That is, her social capital is both the contacts she herself holds and the way in which
those contacts link her in to other patterns of relations. Social capital thus exists in
an amazing multitude of forms; nevertheless, any form of social capital may be
productive through one or more of three types of benefits: information, influence
and control, and social solidarity. Information benefits arise when social
relationships provide access to relevant, timely and trustworthy information of use to
the actor in question (Burt 1992; Laumann and Knoke 1987). The influence and
control benefits of social ties are obverse sides of one coin: the ability to influence
others (Parsons 1963; Coleman 1990) and the ability to be free of others' influence.
Solidarity benefits arise when there is some degree of mutual trust and commitment
among a group of individuals. In this paper, we confine our discussion to the first
two types of benefits. 2 Our conception of social capital differs from the definition
given by Gabbay and Leenders (this volume), who emphasize a distinction between
social ties and social capital. In their view, social capital comprises the beneficial
resources actors draw from their social networks, rather than the relationships that
constitute those social networks. In our view, the aspects of social structure and
social relationships that are potentially valuable through their provision of beneficial
resources are forms of social capital; beneficial resources provided by forms of
social capital are what we term benefits (Sandefur and Laumann 1998b).
220 - Corporate Social Capital and Liability

As presented by Coleman (1988, 1990), social capital is defined by its function;

thus, any form of social capital has a quality that may be called its valence. That is,
forms of social capital which provide one benefit or are useful for one purpose may
not provide a different benefit and may be useless or actively harmful for other
purposes. This variable characteristic of benefit or liability, of positive or negative
'charge,' may change given the context in which a particular form of social capital is
activated and the purpose for which it is employed. Among lawyers, for example,
the solo practitioner is free of formal collegial entanglements and so has the
opportunity to be a self-determining entrepreneur (Seron 1996). At the same time,
however, he is a relative isolate, which means that, while he is relatively free of
others' control, he may lack others on whom he can rely when he faces difficulties.
Thus, Arnold and Kay (1995) in a study of lawyers' professional misconduct, find
that solo practitioners are more likely than lawyers who work in law firms both to he
charged with acts of professional misconduct, and to be convicted of them. The
misbehaving solo practitioner might have been saved from disgrace if he had been
prevented by vigilant colleagues from committing his crime. Post hoc, he might
hope to mobilize others to collude to hide his misdeeds or to try to sway the
decisions of members of the disciplinary committee considering his disbarment.
Thus, social structure that is social capital for some purposes is social liability under
other circumstances.
The third characteristic of social capital to note is the variation in its mode of
acquisition. Some forms of social capital are acquired relatively passively (such as
family connections acquired by birth), while others may be actively sought (such as
acquaintance with influential colleagues). The acquisition and cultivation of such
key relationships can be part of a deliberate strategy of personal advancement, as
well as the unintended by-product of ordinary informal social contact.

The Social Capital of Ties to Professional Elites

Lawyers who work in private practice are rewarded with income accruing as a result
of evaluation by two audiences, their employers and their clients. Lawyers can
increase their incomes by being superior performers and by appearing to be superior
performers. Contacts who are knowledgeable about the bar or its clients can be
valuable by providing a lawyer with information that helps make her human
capital-her talents, abilities, and experiences-more productive (Burt 1992).
Contacts who are influential in the bar or with its clients can promote a lawyer's
reputation with clients and among other lawyers. Elite lawyers in a city bar can act
as such informed and influential contacts.
The elite of the bar are not simply prestigious or respected members of the
profession; they are leaders of various constituencies within the bar and liaisons with
people and organizations outside it (cf. Laumann and Pappi 1976; Heinz et al. 1982;
Heinz et al. 1997). As leaders, they are consulted because they are knowledgeable
and their acquaintance is sought because they are influential. A notable lawyer is not
just then a prominent member of his local profession, he is also a 'relay point' among
his associates. Considered from a network analytic perspective, the notable lawyer's
prominence reflects his advantageous structural position as a leader of some
segment of the bar or the bar at large-his position at the center of a pattern of
The Changing Value of Social Capital in an Expanding Social System - 221

branching relationships. Through his ties to constituents and his relations with other
prominent members of the bar, he can synthesize a picture of the bar and its clientele
that a less prominent lawyer would not have the information to construct. Further,
his status as a leader may give him access to particular, possibly sensitive items of
information--either about the larger bar or about matters of interest to those whose
work intersects with his own. Thus, contacts with elites can provide a rank and file
lawyer with information about the general 'lay of the land' and about specific
Acquaintance with a member of the elite can also benefit a lawyer through the
influence elites may use on her behalf. Such sponsorship by an elite or 'notable'
lawyer may benefit a rank and file lawyer in three ways. First, the judgment by a
respected member of the profession that one is a valuable colleague or a key player
enhances one's reputation among peers and superiors. Second, the elite lawyer can
circulate his favorable evaluation among his other constituents and other prominent
lawyers; in such a way, his position in the network of relations in the bar can
augment the benefit of his favorable evaluation.3 Thus, one's reputation benefits
both from the seal of approval of a respected judge and from the ramifying nature of
the elite lawyer's social network. The third way in which an elite lawyer may use his
influence to benefit a constituent is by a specific intervention. For example, 'making
a call' is a common practice in which elites use both their social connectedness and
their influence to benefit their constituents by intervening in another's decision
(about, for instance, whom to hire, whom to promote, or how much to reward
Even quite unintentional actions by elites, such as the passing mention of a
name to a potential client or a gossipy story about a competitor can benefit a rank
and file lawyer. There is no doubt great variation in the intensity and type of elite
sponsorship and support. In the analyses to follow, we do not have measures of the
specific mechanisms through which the social capital represented in elite contacts
has its effects. We simply argue that, on the whole, such contacts should provide
access to information and to someone potentially willing to use his influence on
one's behalf.

Social Capital and Changes in the Social System

By this reasoning, elite contacts would be a form of social capital in a city bar with
1500 lawyers or a bar with 15,000 lawyers. But, we argue, in addition, that the rate
of income return on such capital will change as characteristics of the system change.
That is, since social capital is an aspect of social structure, as the social structure
itself changes, the value of its appropriable components may change. In particular,
we hypothesize that the value of elite ties will be greater in larger systems than in
smaller ones because the social capital of elite ties is likely to be more scarce in
larger systems.
Let us consider elites for the moment as groups with some degree of internal
social cohesiveness (rather than a priori designated proportions of the population,
such as the top 1%). For example, consider an elite such as the generally recognized
group of respected scholars in some academic field, for instance the top researchers
of poverty in sociology or of Anglo-Saxon in English. In addition to being aware of
222 - Corporate Social Capital and Liability

one another's work, these scholars meet at conferences and on review panels and
they correspond about their students, their research and other concerns of their field.
Each of them may not personally know all of the others, but each is acquainted with
a (probably very high) proportion of her colleagues. Consider a situation in which
such more or less cohesive elites are relatively open to communication with rank
and file system members. The hypothetical respected scholars exemplify such an
elite, as do the leaders of a community or town (Laumann and Pappi 1976) and
prominent lawyers in a city bar (Heinz et al. 1982; Heinz et al. 1997).
As a social system expands, such elites are unlikely to grow at a corresponding
rate. Two different approaches suggest support for this proposition. Mayhew (1973)
presents a formal proof of the proposition that the rate of elite expansion will be
slower than the growth of the social system the elites inhabit. Employing Mosca's
(1939) definition of a ruling elite as some numerical minority of the popUlation,
Mayhew shows that the ruling elite of a social system will comprise a decreasing
proportion of the popUlation as the system increases in size. A similar prediction
follows from a network analytic perspective. Again, let us assume that a certain
degree of social connectedness among members of the elite is necessary for the elite
to persist as a relatively defined group. As an elite grows in size, the ability of
individual members of the elite to maintain ties of a specified intensity with other
members will decrease. Any individual can maintain only a certain number of
relationships; as the number of potential interaction partners increases, the average
proportion of those potential partners known by individuals will likely decrease.
Individuals' 'carrying capacity' for relationships thus places an upper bound on the
size of an elite with a specified degree of internal cohesion.
From the perspective of 'ordinary' system members, the question may be framed
somewhat differently. In a smaller social system, acquaintance between a rank and
file member and a member of the elite is more likely than in a larger one, if only
because acquaintance between any two randomly selected members is more likely.
Because elites comprise a smaller proportion of the population in larger systems, the
chance probability of a tie between a randomly selected member of the population
and a member of the elite is lower. To the extent that members of the elite possess
special influence or other resources that they may use on behalf of their
constituencies, access to such resources is then more restricted (in the sense of being
less widespread) in larger systems. Thus, the resources are more scarce and
consequently may be more valuable.

Changes in the Social Organization of the Bar and the Value of Elite Contacts
Since 1970, the number of lawyers in the United States has roughly doubled; the rate
of expansion of the bar has far outstripped growth in the population (Abel 1989:
Table 23). This tremendous growth in the number of lawyers has been attended by
changes in the way law practice is organized. Lawyers and their work have become
increasingly incorporated into organizations such as large law firms. Many lawyers
now occupy work roles more similar to that of employees than to the tradition of
free-standing professionals. Job placement has become increasingly mediated by law
school placement offices and by groups of firms that join forces in their search for
qualified students (Abel 1989: 224). Law firms themselves have grown larger, and
The Changing Value of Social Capital in an Expanding Social System - 223

have experienced attendant organizational changes that arguably erode the

professional autonomy once characteristic of work in the law (Galantar and Palay
1991). Thus, lawyers' opportunities to recruit clients and to increase their own
incomes have become increasingly structured by the formal organizations that train,
sort, employ and reward them.
Reflecting national trends, the Chicago bar has doubled in size in the past
twenty years and law practice has become increasingly concentrated in large
organizations. In a random sample of Chicago lawyers taken in 1975, 19% of
licensed lawyers were solo practitioners, while a similar survey conducted in 1995
found that solo practitioners constituted only about 13% of lawyers. The twenty
years saw no change in the proportion of lawyers working in law firms (67%), but a
great increase in the average size of the firms in which lawyers were employed (37
lawyers in 1975, 178 lawyers in 1995). As these firms have grown, they have in
many cases formalized procedures for hiring, evaluation, and promotion. In such a
context, Chicago lawyers' income attainment chances may have become
increasingly 'bureaucratized: in the sense that a greater proportion of lawyers are
subject to formalized procedures applied with supposedly less partiality.
One could argue that the increasing formalization of lawyers' income attainment
chances should lead to a decrease in the importance of elite contacts and other
particularistic factors, in favor of the increasing importance of formal qualifications,
such as education. Such would be a classic Weberian argument about bureaucracy
(Weber 1946). However, one might also argue that key contacts would remain
important or in fact increase in value with the increasing bureaucratization of
mobility contests. Even in more formalized settings, the information accessed by ties
to elites should still be valuable to lawyers in increasing their incomes by providing
notice of opportunities, contingencies or complications. And, clients and employers
will continue to encounter situations in which there is some uncertainty about
someone's past or future performance that available information 'in the file' cannot
resolve. In such cases, elite influence or the 'reference' provided by association with
elites can help resolve this ambiguity in favor of the lawyer who counts elites among
his acquaintance. Further, since employers (and, likely, some clients) have
implemented formal procedures intended to reduce the role of non-productive
factors in hiring and reward decisions, they will find themselves faced with several
formally equivalent candidates. Here again, elite influence can be a crucial factor in
determining who among formally identical lawyers is to receive scarce and valuable
opportunities and rewards. In sum, it is not clear that increasing rationalization will
lead to a reduction in the value of elite ties.
Below, we present a preliminary test of the two hypotheses that we have
outlined above. First, we ask whether, net of more evident productive
characteristics, elite contacts are associated with higher income. Second, we ask
whether the value of elite ties increases as a social system expands. We consider the
specific case of the Chicago bar as a large, loosely-bounded social system in which a
group of elite lawyers is identified as leaders of the bar and of various constituencies
within it (Heinz et al. 1997). Employing two surveys of the bar, the first conducted
in 1975 and the second in 1995, we investigate changes in the value of contacts with
elite lawyers that coincide with changes in the social organization of the bar.
224 - Corporate Social Capital and Liability


The 1975 study of Chicago lawyers includes 777 randomly selected respondents
with offices within the city limits (Heinz et al. 1982). The 1995 survey includes 788
randomly selected lawyers drawn from the same area (Heinz et al. 1997). The
response rates for both surveys were just over 82% of the target samples. The
lawyers in each survey were asked a series of questions about their career history,
their educational and family background, the nature and structure of their current
work and work experiences, their social and political attitudes, their social
participation, and their relationships with specified other groups of lawyers.
Analyses presented below are for private practice lawyers (those working in law
firms or as solo practitioners). The sample includes 522 such lawyers in 1975, 35 of
whom were dropped from the analyses due to missing data, resulting in a sample of
487 respondents. In 1995, the sample includes 521 such lawyers, 33 of whom were
dropped due to missing data, resulting in an analysis sample of 488 respondents. 4
In consultation with informants knowledgeable about the Chicago bar, the
principal investigators identified lawyers who appeared to be representative of
several salient elites. The elites selected were drawn from a large range of social and
professional categories: bar association officers, political partisans, type of work and
kind of clientele, practice setting, gender, race, and ethnic background (Heinz et al.
1997). Lawyers who were currently in public office were excluded from the
identified elites, to avoid conflating official power or influence with prominence
among colleagues. We refer to these selected elite lawyers as the 'notables.' The
1975 questionnaire included the names of 49 notables, 6 of whom were dropped
from subsequent analyses. 5 The 1995 survey included the names of 68 notables,
three of whom were dropped from subsequent analyses because they proved to be
relatively unknown (for details, see Heinz et al. 1997). The notables of 1995 are
more diverse demographically than those included in the 1975 study, as is consistent
with the increase in the demographic diversity of the Chicago bar. The 1995 list of
notables also includes lawyers who work as the internal counsel of non-legal
organizations and those who work for legal aid organizations, while the 1975 list did
As part of each survey, respondents in the random sample were given a list of
the notables and asked to indicate which notables they knew at two specific levels of
association. On their first pass through the list, they were asked to mark those
notables with whom they were 'personally acquainted.' Next, they were asked to
mark the notables of their acquaintance who they believed would take the time to
advise them if asked. Analyses in this paper will focus on the notables nominated at
the second, stronger level of connection.
A variety of aspects of lawyers' personal and work characteristics affect their
income attainments (d. Hagan and Kay 1995; Abel 1989; Sandefur and Laumann
1998a). In this analysis, we are interested in the relationship between ties to elites
and income net of these other characteristics. In our models, therefore, we include
measures of respondents' law school quality, achievement in law school, practice
setting, practice type, and position in their firm. We include a control for age, which
is here a proxy for experience in the law. The relationship between age and income
is modeled by a term for age and a term for age-squared. This latter term permits the
The Changing Value of Social Capital in an Expanding Social System - 225

positive effects of age (and, by proxy, experience) to decrease over time, and so acts
as a statistical control for the obsolescence of skills acquired early in one's
professional career. We also include a control for gender; female lawyers have been
shown to receive lower pay than male lawyers, on average and net of certain
productive characteristics (cf. Hagan and Kay 1995). We are unable to include a
measure of hours worked, since this question was not asked of respondents in 1975.
The distribution of ties to notables is quite skewed in both periods, and we have
no reason to believe that the relationship between the number of notables known and
income is linear; therefore, we constructed dichotomous measures-whether the
respondent reported knowing 1,2,3, or 4 or more notables well enough to ask them
for advice. In the regression analyses, nomination of 0 notables is the omitted
category. Age is the respondent's age calculated from her self-reported year of birth.
As a measure of the lawyer's ability (or human capital) as indicated by her law
school performance, we include a pair of dummy variables indicating whether the
respondent was in the top 10% of her class and/or on the law review of her law
school or in the top 11-25% of her class. In the regression analyses, class ranks
below the top 25% are the omitted category. Whether the respondent attended an
elite law school (e.g., Harvard), a prestigious law school (e.g., Northwestern
University), or a local law school (e.g., John Marshall Law School in Chicago) is
indicated by a trio of dummy variables; regional law schools (e.g., University of
Illinois) are the omitted category in the regression analyses.6 Practice setting is
measured by a dummy variable indicating whether or not the respondent is a solo
practitioner; effects presented in the regression analyses have as their referent
respondents working in law firms.
Practice type is measured slightly differently in the two surveys. In the 1975
survey, respondents were asked what proportion of their income they earned from
work with clients who were businesses. Two dichotomous variables indicate
whether 25% or less of their income or 75% or more of their income comes from
work for businesses (as opposed to non-profit corporations or personal clients). In
the regression analyses, the omitted category is lawyers receiving 26-74% of their
income from clients who are businesses. In the 1995 survey, respondents were asked
what proportion of their clients were businesses; using this information, we
constructed a pair of dichotomous variables parallel to the indicators constructed for
1975.7 A lawyer's position in her firm is modeled by a dummy variable indicating
whether or not the respondent is a partner. The respondent's gender is modeled by a
dummy variable indicating whether or not the respondent is a woman. 1975 income
is converted to 1995 dollars using the Consumer Price Index, and its natural log is
taken before computation of the regression equations. Modeling the natural log of
income reduces the impact of outlying observations (in this case, respondents with
very high incomes) on the coefficient estimates, and it permits straightforward
comparisons across time. In the regression equations, the estimated metric
coefficient of a variable predicting income may be interpreted as the proportionate
change in the dependent variable given a one-unit increase in the predictor. The
magnitude in dollars of the effect of a change in the value of an independent variable
will depend upon the value at which it and the other independent variables in the
model are evaluated (Hauser 1980; Stolzenberg 1980).
226 - Corporate Social Capital and Liability

Table 1. Descriptive statistics for analysis variables

1975 1995
n 487 488
Logged Income (st.dev.) 11.6 (.73) 11.4 (.83)
I Notable Advisor 18% 11%
2 Notable Advisors 14% 13%
3 Notable Advisors 10% 6%
4 + Notable Advisors 27% 17%
Age (st.dev.) 44.7 (14.0) 42.0 (10.4)
Top 10% of ClasslLaw Review 33% 29%
Top 11 - 25% of Class 29% 32%
Elite Law School 23% 14%
Prestigious Law School 16% 14%
Local Law School 45% 43%
Solo Practitioner 27% 20%
Clients 0-25% businesses 31% 24%
Clients 75-100% businesses 47% 58%
Partner 42% 40%
Female 3% 21%

Table 1 presents means and standard deviations for continuous measures and
percentages for dichotomous measures used in the regression analyses. Recall that
this sample is restricted to private practice lawyers. It is worthwhile to discuss
briefly some of the changes in the Chicago bar between 1975 and 1995. The most
striking change is in the proportion of women. In 1995, 29% of all and about 21 % of
private practice lawyers in Chicago were women, nearly ten times their presence in
1975. (Currently, women are over-represented in government employment and
underrepresented in private practice.) Businesses constitute a larger proportion of
the client base of private practice lawyers in 1995 than in 1975. Solo practitioners,
who serve mostly individuals and small businesses, have declined as a proportion
both of lawyers in general (see above) and of lawyers in private practice. The share
of Chicago lawyers who graduated from elite schools decreased. This is likely
related to two factors. As law firms grew during the 20 years between surveys, they
broadened recruitment beyond graduates of top schools. Further, as the number of
applicants to law schools increased during the same period, elite schools held their
enrollments relatively constant while lower-ranked law schools increased the size of
their incoming classes. Thus, changes in both supply and demand are reflected in the
educational backgrounds of Chicago lawyers.
Acquaintance with notable lawyers is clearly much more common in 1975 than
in 1995. Fifty-three percent (53%) of lawyers know no notables in 1995, while only
31 % of lawyers lack such connections in 1975, even though the list of notables was
longer (50% longer) and more inclusive in 1995. As we suggested earlier, ties to
elites should be a rarer commodity in a larger system. A different and convenient
summary measure of the degree of interconnectedness in the bar is a simple measure
The Changing Value of Social Capital in an Expanding Social System - 227

Table 2. Regressions for logged income. 1975 and 1995. Metric coefficient
estImates. standard errors In parentheses.
1975 1995
Intercept 8.49*** (.37) 7.73*** (.49)
1 Notable -.08 (.08) .03 (.09)
2 Notables .00 (.08) .25** (.09)
3 Notables .02 (.10) .13 (.11)
4 Notables .35*** (.08) .43*** (.08)
Age .11*** (.02) .13*** (.02)
Age2 -.00*** (.00) -.00*** (.00)
Top 10% of Class .08 (.06) .25*** (.07)
Top 11-25% of Class .06 (.06) .21*** (.09)
Elite Law School .02 (.09) .21* (.09)
Prestigious Law School -.09 (.09) .10 (.09)
Local Law School -.00 (.08) -.15* (.07)
Solo Practice .15 (.08) -.38*** (.09)
Clients 25% or less business .01 (.07) .02 (.09)
Clients 75% or more business .18** (.07) .24** (.08)
Partner in firm .52*** (.07) .45*** (.08)
Female -.21 (.15) -.14* (.07)
Adjusted R2 .40 .50
*p < .05. ** P < .01. *** P < .001.
Italic-face type indicates a statistically significant difference between the
two periods at the level of p < .05. Bold-face type indicates a statistically
significant difference between the two periods at the level of p < .10.

of the density of ties between elites and rank and file lawyers. In this case, the
density is simply the ratio of the number of observed nominations of elites by rank
and file members to the number of possible nominations; as such, it represents the
degree to which possible connections have been realized. In the Chicago bar, the
density of ties between elite lawyers and the rank and file has declined considerably.
In 1975, the sample density of such ties at the level of simple acquaintance was
.117; the density at the stronger level of connection was .065. In 1995, the sample
density of ties of simple acquaintance was .059, while the density of advisor ties was


Table 2 presents results for OLS regressions of logged income in 1975 and 1995.
Estimated metric coefficients and their standard errors are presented separately for
each model in each year.9 In the body of Table 2, statistically significant differences
between the coefficients in the two periods are indicated by differences in typeface;
statistical significance within each period is indicated by the familiar asterisk
The findings support our first hypothesis that the social capital represented by
ties to elite lawyers would be positively associated with higher incomes net of other
characteristics. In both periods, counting four or more notable members of the bar as
advisors is positively associated with higher income (p < .001); in 1995, counting
two notable members of the bar as advisors is positively associated with higher
income (p < .01). The pattern of findings is also consistent with our hypothesis that
228 - Corporate Social Capital and Liability

elite ties would be more valuable in the larger bar. The income returns to the highest
degree of elite connection appears to have increased slightly, but this increase is not
statistically significant. The income returns to less extensive connections to the elite
have increased to a degree that is statistically significant (p < .05).10 In neither
period does 1 tie to a member of Chicago's legal elite evidence an appreciable
relationship to income.
To get a sense of the 'real money' value of ties to notable lawyers, let us
compare the predicted incomes of similar lawyers with and without ties to local
legal elites. In 1975, the predicted income of a thirty year-old male graduate of the
elite Harvard Law School, who graduated in the top 10% of his class and worked as
an associate in a law firm where his clientele consisted of a roughly equal balance of
businesses and persons would be essentially the same whether or not he counted two
notable lawyers among his closer acquaintances. In 1995, an identical lawyer would
gain roughly $20,000 by acquaintance with two notable members of the bar. To take
a different example, consider a fifty year-old man who graduated in the middle of
his class from a local law school and operated a solo practice where he served
largely personal clients. In 1975, if he counted four or more elite members of the bar
among his acquaintance, his predicted income would be 44% higher (in 1995
dollars) than that of an identical lawyer with no notable contacts. In 1995, a similar
lawyer gains 58% more income by acquaintance with four or more elite lawyers.
The increase in income associated with elite contacts varies in size depending on
characteristics of the lawyer and her practice, but the returns to notable ties are
considerable and, on the whole, larger in the larger system than in the smaller.
Specific other findings in these models are interesting and worthy of discussion,
both as they inform us about changes in the determinants of lawyers' income and as
they suggest the workings of other forms of social capital. Earlier, we suggested that
work in the law have become increasingly structured by organizations such as law
firms and law schools, and that procedures for hiring, promoting and rewarding
lawyers have become more formalized. This observation is borne out by inspection
of the effects of class rank and law school prestige on lawyers' incomes. Changes in
the relationship between legal education and income are striking. When partnership
status, client base, law school, practice setting and gender are controlled, class rank
has no direct effect on lawyers' incomes in 1975. In 1995, however, class rank is a
statistically significant predictor of income, net of all controls (p < .001). The effects
of law school prestige are more than twice as large in 1995 as in 1975 (p < .10 for
the test of statistically significant difference between the two periods for the effect
of elite law school attendance). It appears that income attainments may have become
more keyed to the human capital signaled by law school performance and attendance
at a well-regarded institution.
Net of age, elite ties, legal education, client type and partnership status there is a
negative relationship between solo practice and income in 1995 (p < .001), while the
positive coefficient in 1975 does not attain statistical significance. In both periods,
solo practitioners have, on average, lower incomes than lawyers in law firms
(Sandefur and Laumann 1998a). Part of the deficit experienced by solo practitioners
stems from the fact that their clients tend to be individuals and small businesses. The
most lucrative of private practice legal work, work for large corporations, is
The Changing Value of Social Capital in an Expanding Social System - 229

performed by lawyers in large law firms. Thus, even when solo practitioners have
businesses among their clients, these clients tend to be small, locally owned
businesses that require neither the volume nor type of work done for large
corporations. In addition, solo practitioners must rely on referrals and advertising to
attract clients,l1 while firm lawyers have the benefit of association with an
organization with an established reputation and of formal organizational contacts
with other lawyers who can share their expertise and who may make referrals that
benefit them (cf. Lazega, this volume).
Given these general considerations about the different situations faced by solo
practitioners and lawyers in firms, there are two possible reasons for the difference
between the two periods in the effect of solo practice on income. It may be the case
that, in 1975, the best lawyers in Chicago were more evenly distributed between
positions in firms and solo practice, while in 1995 the best lawyers were more
strongly attracted to work in firms. Our measures of lawyers' ability are restricted to
measures of their education, and so do not capture all the attributes that make for a
successful lawyer. Thus, stronger selection of talented lawyers into firms in 1995
than in 1975 might account for the different relationships between solo practice and
income. An alternative explanation has to do with the increasing concentration in
law firms both of opportunities to do lucrative work and for professional
advancement. Solo practitioners may be at a greater disadvantage in the larger, more
formally organized bar not only because of the greater organizational resources and
more lucrative client bases of larger firms, but because the career mobility pathways
provided in large firms are now more remunerative than those open to solo
Unsurprisingly, partners in law firms make more money than associates in both
periods. As we noted above, work for business clients is more lucrative than work
for personal clients and non-profit organizations; thus, those lawyers whose practice
consists largely of work for businesses have higher incomes. In 1975, the negative
effect on income of being a woman does not reach statistical significance (there
were very few women in the Chicago bar in 1975). In 1995, the negative effect on
income of being female is statistically significant, net of controls for practice setting,
client base, education and elite contacts.

Further Considerations
The results from our statistical analyses of the relationship between social ties to
elites and income in the two periods are consistent with our suggestion that such ties
are useful in increasing income and with our hypothesis about the increasing value
of elite contacts in expanding social systems. Nevertheless, competing explanations
for our findings deserve attention, and we turn now to discussion of three classes of
competing explanations.
The first set of alternative explanations has to do with causal ordering. We have
discussed our findings as though elite contacts were a 'cause' of income attainments.
Of course, the relationship is most likely non-recursive; that is, the causal arrows run
in both directions. More successful lawyers are likely more attractive associates for
notable lawyers and are perhaps more likely to come into contact with them in the
course of their work and work-related socializing. At the same time, acquaintance
230 - Corporate Social Capital and Liability

with notable lawyers likely affects income attainments in the ways we have
discussed, and perhaps in other ways which our theory did not address. Further, it is
unlikely that most of the lawyers in our sample became acquainted with the elites
they nominated only in the year of the survey; thus, for many (if not most) lawyers,
we are investigating the relationship between contemporary income and
relationships formed some time before income was measured. It is highly unlikely
that the opposing hypothesis about causal order-success in law as measured by
income affects the likelihood of elite acquaintance, but the social capital of elite
acquaintance has no effect on income-is the truth. Nevertheless, if such were the
case, these findings would then tell a story about the social stratification of the bar
that would be more about the social distribution of success than about the resources
which contribute to success.
The second set of alternative explanations concerns the issue of selection or
omitted variables in an analysis. This concern is ubiquitous in social scientific work.
Any or all of our findings which we have interpreted as the 'effect' of one factor on
another may in fact be the result of selection into certain roles and circumstances on
the basis of some other factor we did not model or did not measure. For example, if
the most motivated lawyers go to work for firms, while the less motivated lawyers
become solo practitioners, then all or part of the effects of firm practice which we
have argued are due to organizational factors are in fact due to unmeasured
characteristics of the individuals who enter firm practice. Likewise, if the most
motivated and consequently successful lawyers are those who are most likely to be
associated with members of the elite, then what we interpret as the effects of social
capital are simply the social organizational results of individuals' traits. If such
selection processes differ in the two periods under study, then effects that we have
argued are due to specific changes in the social and formal organization of the bar
could in fact be due to other changes in the bar's organization that affect the ways in
which particular individuals are selected into particular circumstances. In this paper,
we can only appeal to the plausibility of our theory to counter this type of alternative
explanation. We note that, if selection were the only operative factor, the findings
would remain an interesting descriptive characterization of changes in the social
organization of the bar, though our explanation would require reconsideration.
The last set of potential competing explanations for our findings has to do with
the identification of the notables themselves. We have operated under two
assumptions about the notables identified by the survey directors. First, we have
assumed that the notables were correctly identified in each study, or at least that any
error in identifying them was not systematic in either period. Our arguments do not
require that the notables be an exhaustive or perfect sample, but they do require that
they comprise an accurate and reasonably representative characterization of the elite
of the Chicago bar. If any error in identifying the notables is systematic-for
instance, if the leaders of some sector of the bar are excluded-then our findings
may be compromised by the invalidity of the notables as representatives of the bar's
elite in one or both periods.
It is difficult to know for certain how the exclusion of some proportion of the
bar's elite would affect our findings. However, let us consider the case in which this
exclusion did occur in both periods. Some substantial proportion of respondents in
The Changing Value of Social Capital in an Expanding Social System - 231

each period then possess elite contacts which we have not measured. In the
regression analyses, these respondents who do have the unmeasured advantage of
notable ties are included in the reference category against which the effect of notable
ties is presented. This should bias the regression coefficients for notable contacts
downward in both periods, a bias which is not fatal to our hypothesis about different
effect sizes in the two periods.
The second assumption about the notables is specific to the second period under
study. While the Chicago bar doubled in size between 1975 and 1995, the number of
notables that respondents were permitted to nominate was only increased by an
additional 50% or so (65 notables in 1995 versus 43 notables in 1975). We argued
above that the elite would not grow at the same rate as the bar; however, even if
such is the case, that does not mean that the list of notables for 1995 was increased
by a factor corresponding to the true increase in the size of the bar's elite. If the
process by which elites were selected in the second period more strongly excluded
relatively unimportant prominent lawyers, then our findings could be an artifact of
the greater importance of the individuals chosen for the 1995 study elite, rather than
a result of the greater importance of elite contacts.
We are very doubtful that the second survey's group of notables results from
more stringent skimming of the cream of the bar. Neither group of notables was the
result of an 'attempt to create a list of the most notable, successful or influential
lawyers in Chicago. Rather, the list[s] include[s] a selection of lawyers, of varying
types, who are prominent in one respect or another, but not necessarily more
prominent than others' (Heinz et al. 1997: 447). We find no reason to believe that the
selection procedures resulted in a systematically more influential and informed
group of notables in the second period. The second possibility, that the selection
procedure did not result in skimming, but did fail to increase the list of notables by a
sufficient amount in the second period, is less cause for concern. If such were the
case, we would again have a number of respondents who have the unmeasured
benefit of notable ties, but are included in the reference category against which the
effects of notables ties are estimated. Such an occurrence would bias the coefficients
downward in the second period, in a direction unfavorable to our hypothesis. Thus,
if such bias were present, the relationships we report in support of our hypothesis
would appear weaker than they actually are.

Contact with the elite of the Chicago bar represents a channel through which rank
and file lawyers may 'tap in' to the social structure of the bar and acquire valuable
resources. The information and influence lawyers access through contacts with
notables are properties of the corporate organization of the bar, and, as such, are
benefits of corporate social capital. The business firms and individuals who retain a
lawyer gain not only from the lawyer's human capital-her skills, abilities, talents,
and industry-but also from their social capital, which she may use as a resource in
her work for them. The findings of this paper suggest that the social capital of ties to
legal elites is associated with valuable income rewards for the lawyers who possess
it. Changes in the organization of a social system-in this case, a tremendous
increase in the size of the Chicago bar-are linked to changes in the value of given
232 - Corporate Social Capital and Liability

forms of social capital. Our findings are consistent with our hypothesis that the
greater scarcity of contact with elites would lead to increases in the value of elite
As formal organizations have become increasingly salient in structuring
Chicago lawyers' access to opportunities and rewards, formal qualifications, such as
educational credentials and law school performance, have become more important
predictors of lawyers' incomes. Employment in law firms, which provides access to
organizational resources, such as office equipment, support staff and colleagues, and
to a more lucrative client base, seems to confer greater benefits in increasing income
for contemporary lawyers than for lawyers of 20 years ago. Yet, at the same time
that the distribution of rewards appears to have become in some ways more
'universalistic' and structured by formal organization, the economic value of ties to
specific others, such as those between a rank and file lawyer and legal elites, has not
declined and, in fact, appears to have increased. In the contemporary urban bar, the
extra-organizational social capital provided by ties to local professional elites
remains valuable, both to the lawyers who have such capital, and, by implication, to
the firms in which they work and to the clients who pay for their services.

This research was supported in part by grants from the American Bar Foundation and the National
Science Foundation (#SBR-9411515). Additional support for Heinz was provided by Northwestern
University's Institute for Policy Research. We thank Charles E. Bidwell. Ronald S. Burt. Jeffrey A.
Hayes, Ray Reagans. Ross M. Stolzenberg. Christopher B. Swanson. Jeffrey Y. Yasumoto. Ezra
Zuckerman. and an anonymous reviewer for helpful comments and useful discussions. This paper is a
revision of a section of a paper presented at the Conferen