(Eva Sorensen, Jacob Torfing) Theories of Democrat (BookFi) PDF

Theories of Democratic
Network Governance
Edited by
Eva Sørensen and Jacob Torfing
Theories of Democratic Network Governance
Also by Eva Sørensen
POLITICIANS AND NETWORK DEMOCRACY (in Danish)

ROLES IN TRANSITION (co-author with Birgit Jæger) (in Danish)
NETWORK GOVERNANCE: From Government to Governance (co-author with
Jacob Torfing) (in Danish)
Also by Jacob Torfing
DISCOURSE THEORY IN EUROPEAN POLITICS: Identity, Policy and

Governance (co-editor with David Howarth)
NEW THEORIES OF DISCOURSE
POLITICS, REGULATION AND THE MODERN WELFARE STATE
Theories of Democratic
Network Governance
Edited by
Eva Sørensen
Professor of Public Administration
Roskilde University, Denmark
and
Jacob Torfing
Professor of Politics
Roskilde University, Denmark
Editorial matter, selection, Introduction and concluding chapter © Eva
Sørensen and Jacob Torfing 2007
All remaining chapters © respective authors 2007
All rights reserved. No reproduction, copy or transmission of this
publication may be made without written permission.
No paragraph of this publication may be reproduced, copied or transmitted
save with written permission or in accordance with the provisions of the
Copyright, Designs and Patents Act 1988, or under the terms of any licence
permitting limited copying issued by the Copyright Licensing Agency, 90
Tottenham Court Road, London W1T 4LP.
Any person who does any unauthorized act in relation to this publication
may be liable to criminal prosecution and civil claims for damages.
The authors have asserted their rights to be identified
as the authors of this work in accordance with the Copyright,
Designs and Patents Act 1988.
First published in 2007 by
PALGRAVE MACMILLAN
Houndmills, Basingstoke, Hampshire RG21 6XS and
175 Fifth Avenue, New York, N.Y. 10010
Companies and representatives throughout the world.
PALGRAVE MACMILLAN is the global academic imprint of the Palgrave
Macmillan division of St. Martin’s Press, LLC and of Palgrave Macmillan Ltd.
Macmillan® is a registered trademark in the United States, United Kingdom
and other countries. Palgrave is a registered trademark in the European
Union and other countries.
ISBN-13: 978–1–4039–9528–5 hardback
ISBN-10: 1–4039–9528–1 hardback
This book is printed on paper suitable for recycling and made from fully
managed and sustained forest sources.
A catalogue record for this book is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Theories of democratic network governance / edited by Eva Sørensen
and Jacob Torfing.
p. cm.
Includes bibliographical references and index.
ISBN 1–4039–9528–1 (cloth)
1. Political planning. 2. Policy sciences. 3. Public – private sector
cooperation. 4. Public administration. I. Sørensen, Eva, 1957–
II. Torfing, Jacob.
JF1525.P6T43 2007
351.01—dc22 2006047274
10 9 8 7 6 5 4 3 2 1
16 15 14 13 12 11 10 09 08 07
Printed and bound in Great Britain by
Antony Rowe Ltd, Chippenham and Eastbourne
Contents
List of Tables, Figures and Boxes xi

Foreword xii
Notes on Contributors xiii
Introduction: Governance Network Research: Towards a

Second Generation 1
The rise of governance network research 3
The aims of this book 7
Defining governance networks 8
Merits and problems of network governance 11
First and second generation research 14
Plan of the book 20
Part I Governance Network Dynamics
1 Theoretical Approaches to Governance

Network Dynamics 25
Introduction 25
Historical institutionalism 31
Rational choice institutionalism 33
Social constructivist (or normative) institutionalism 35
Poststructuralist institutionalism 38
Similarities and differences 41
The structure of Part I 42
2 Mechanisms of Governance Network Formation – a
Contextual Rational Choice Perspective 43
Nils Hertting
Introduction 43
Interpretation, rational choice and mechanisms 45
Contextual mechanism: perceived interdependencies 47
v
vi Contents
Actor calculation mechanism: preference for

informal networks 50
The game mechanism: the problem of continuous
cooperation 51
Collective actors and vertical games 56
Conclusion 57
3 Virtuous and Viscous Circles in Democratic
Network Governance 61
B. Guy Peters
Institutionalization and deinstitutionalization 62

Factors associated with virtuous and
viscous spirals 65
Political factors 66
Functional factors 69
Social pressures 70
Other factors in explaining success 71
Characteristics of the members 71
Operating environment 73
Tasks 74
Summary and conclusions 74
4 Decentred Theory, Change and Network Governance 77
Mark Bevir and R. A. W. Rhodes
Introduction 77
Positivist approaches to network governance 78
Decentring network governance 80
The analysis of change in networks 81
Managing change in networks 83
Conclusions 87
Part II Governance Network Failure
5 Theoretical Approaches to Governance Network Failure 95

Interdependency theory 98
Governability theory 102
Integration theory 104
Governmentality theory 106
Contents vii
Similarities and differences 108

The structure of Part II 110
6 Closure and Governance 111
Linze Schaap
Introduction 111
Governance networks: open, closed, or both? 112
A systems theoretical contribution? 113
Governance networks and types of social systems 117
Two types of closure 118
Three explanations for closure 121
The relations between explanations for closure 123
Governing closed networks? 124
Governing veto power? 125
Governing closed frames of reference? 128
Governing closed policy communication systems? 129
Some concluding remarks 131
7 Consensus and Conflict in Policy Networks:
Too Much or Too Little? 133
Joop F. M. Koppenjan
Introduction 133
Consensus and conflict: an exploration of
two ambivalent concepts 135
The first face of policy networks: a surplus
of consensus 138
The second face of policy networks:
insufficient consensus 143
The true face of policy networks and its
implications for network governance 147
Conclusion: managing the consensus–conflict
dimension in network-settings 151
8 Network Governance: Effective and Legitimate? 153
Tanja A. Börzel and Diana Panke
Introduction 153
Networks as governance 154
The demand for effectiveness and legitimacy 156
Effectiveness and legitimacy: a trade-off? 163
Conclusion 165
viii Contents
Part III Metagovernance
9 Theoretical Approaches to Metagovernance 169

Introduction 169
Interdependency theory 170
Governability theory 172
Integration theory 175
Governmentality theory 178
Similarities and differences between the theories 180
Where to go from here? 181
10 Governing the Formation and Mobilization

of Governance Networks 183
Peter Triantafillou
Introduction 183
Governmentality and advanced liberal government 185
Mobilizing agency 187
Governing through the formation of autonomy
and interdependencies 190
Governing the performance of networks 194
Conclusion 196
11 Meta-governance as Network Management 199
Erik-Hans Klijn and Jurian Edelenbos
Introduction: a network management
perspective on meta-governance 199
Process design and management: setting up
and facilitating network interactions 201
Institutional design: changing the network 206
Good network management: skills and competencies 211
Research challenges 213
12 Governing Outputs and Outcomes of
Governance Networks 215
Laurence J. O’Toole, Jr
Framing the subject 215
Two notions of meta-governance 218
Meta-governance in action 221
Contents ix
Possibilities for public authorities to shape

network outputs and outcomes 223
Meta-governance via policy formulation 223
Assisting in the play of the game 224
Linking and segmenting games 225
Changing the game: active meta-governance
by public authorities 226
Conclusion 228
Part IV Democratic Network Governance
13 Theoretical Approaches to Democratic Network

Governance 233
Introduction 233
Governance networks and liberal democracy 234
Governance networks and postliberal democracy 236
Similarities and differences between the theories 245
Where to go from here? 246
14 Governance Networks and Participation 247

Allan Dreyer Hansen
The common good 249
Learning democracy 251
Equality 254
Conclusion 258
15 Networks and Democratic Ideals: Equality,

Freedom, and Communication 262
John S. Dryzek
Applying the standard democratic principles

to networks 263
Beyond lingering statism in democratic theory 264
Networks and the communicative aspect of
democratic theory 266
Who communicates 268
Beyond models of democracy 269
The contribution of governance networks to democracy 271
Conclusion 273
x Contents
16 Democratic Accountability and Network

Governance – Problems and Potentials 274
Anders Esmark
Democratic network governance? 274
Accountability as a democratic norm 276
Accountability and inclusion 278
First challenge: finding the holders and holdees 282
Accountability and publicity 284
Second challenge: sufficient publicity 287
Accountability and responsiveness 290
Third challenge: adequate responsiveness 293
Conclusion 295
The Second Generation of Governance Network

Theory and Beyond 297
Governance networks are here to stay 297

A multi-theoretical approach to network governance 299
Contributions to our understanding of
governance networks 303
Where next? 310
Bibliography 316
Index 343
List of Tables, Figures and Boxes
Tables
I.1 Overview of four different governance network theories 17

1.1 Overview of four different approaches to
institutional analysis 30
7.1 Implications of the level of consensus and conflict
within networks 151
9.1 Differences between the four governance
network theories 181
13.1 Four theoretical approaches to postliberal democracy 236
Figures
2.1 Perceptions, calculations and games in governance

network formation 57
6.1 Types of closure 120
6.2 Relationships between clarifications for closure 124
7.1 Policy games and arenas that cut across various
policy networks 148
Boxes
4.1 Ten lessons on how to manage your network 84

4.2 Local practical solutions 84
xi
Foreword
This volume is a part of a small series of books that aims to analyse how
governance networks contribute to the governing of our increasingly
complex and fragmented societies. The book series consists of three
books: Theories of Democratic Network Governance edited by Eva Sørensen
and Jacob Torfing, Methods in Democratic Network Governance edited by
Peter Bogason and Mette Zølner, and Democratic Network Governance in
Europe edited by Martin Markussen and Jacob Torfing. The three books
are self-contained volumes that can be read independently, but they are
all part of the same endeavour to develop a second generation of gover-
nance network research that focusses on new and important questions
about the dynamics of governance networks, the conditions for their
success and failure, the attempt to metagovern governance networks
and their democratic problems and potentials. The contributing authors
are either members of the Centre for Democratic Network Governance
that was established at Roskilde University in 2003, or have been associ-
ated with the Centre as guests or visiting research fellows. Anonymous
reviewers have provided valuable comments to earlier versions of the
chapters. Our student assistants have collected data and gathered mate-
rial for the books, and Andrew Crabtree and Jon Jay Neufeld have helped
to improve the language. We thank them for their excellent work.
Jacob Torfing
Series Editor
Roskilde
xii
Notes on Contributors
Mark Bevir, Professor, Department of Political Science, University of

California, Berkeley, USA.
Tanja A. Börzel, Chair of European Integration, Otto-Suhr-Institute for
Political Science, Centre for European Integration, Free University of
Berlin, Germany.
Allan Dreyer Hansen, Associate Professor, Department of Social
Science, Roskilde University, Denmark.
John S. Dryzek, Professor of Social and Political Theory, Australian
National University, Australia.
Jurian Edelenbos, Research Fellow, Department of Public
Administration, Erasmus University of Rotterdam, The Netherlands.
Anders Esmark, Associate Professor, Department of Social Sciences,
Roskilde University, Denmark.
Nils Hertting, PhD in political science and Research Fellow at the
Institute for Housing and Urban Research, University of Uppsala,
Sweden.
Erik-Hans Klijn, Associate Professor, Department of Public

Administration, Erasmus University of Rotterdam, The Netherlands.
Joop F. M. Koppenjan, Associate Professor, The Faculty of Technology,
Policy and Management, The Delft University of Technology, The
Netherlands.
Laurence J. O’Toole, Jr, Professor of Public Administration, Department
of Public Administration & International Affairs, The University of
Georgia, Athens, USA.
Diana Panke, PhD Student, Otto-Suhr-Institute for Political Science,
Centre for European Integration, Free University of Berlin, Germany.
B. Guy Peters, Maurice Falk Professor, Department of Political Science,
University of Pittsburgh, USA.
R. A. W. Rhodes, Professor, Research School of Social Sciences,
Australian National University, Canberra, Australia.
xiii
xiv Notes on Contributors
Linze Schaap, Leader of Centre for Local Democracy, The Faculty of

Social Science, Erasmus University, The Netherlands.
Eva Sørensen, Professor of Public Administration, Department of Social
Science, Roskilde University, Denmark.
Jacob Torfing, Professor of Politics, Director of Centre for Democratic

Network Governance, Department of Social Science, Roskilde University,
Denmark.
Peter Triantafillou, Assistant Professor, Department of Social Science,

Roskilde University, Denmark.
Introduction
Governance Network
Research: Towards
a Second Generation
The structures and processes of public policy-making and societal

governance are rapidly changing. The many reports about the failure of
national and local governments to solve concrete policy problems and
exploit new opportunities through hierarchical command and control
have triggered an increasing use of market regulation in the provision
of public goods and services. Hence, privatization of public enter-
prises, construction of quasi-markets, contracting out of public services,
competitive deregulation and commercialization of the remaining
public sector in accordance with the principles of the New Public
Management doctrine have been in fashion since the early 1980s. The
limits to the neo-liberal quest for ‘less state, more market’ not only
include the standard problems of imperfect competition, unstable
and insufficient market supply, unchecked externalities and growing
inequality. The marketization strategy also fails to reduce the need for
state regulation which seems to grow rather than diminish in the face of
increased marketization. Last but not least, it fails to facilitate
collectively oriented and pro-active governance on the basis of joint
objectives and mutual trust. Despite of the many problems of the
neo-liberal marketization strategy, political decision makers around the
world continue to worship the market forces, mainly out of an ideolog-
ical concern for the facilitation of free choice and the promotion of
incentive-driven individual action. Whereas the critics of market regu-
lation call for a return to state steering, the true believers in competitive
market regulation claim that only more market will solve the problems
since only a perfect market can produce perfect regulation ( Jessop,
2000: 224 ff.).
1
2 Theories of Democratic Network Governance
Now, in the last decade, the heated ideological debate about whether
to base societal governance on either state or marked has been
challenged by new developments in societal governance. Hence, in
order to compensate the limits and failures of both state regulation and
market regulation new forms of negotiated governance through the
formation of public–private partnerships, strategic alliances, dialogue
groups, consultative committees and inter-organizational networks
have mushroomed. The empirical examples are numerous, so let us just
mention a few to get an impression of what we are talking about:
The construction of a huge, EU-funded shopping centre in major city

carried the promise of new jobs in construction and retail. In order to
ensure that as many of the unemployed people from the local area
would get a job a large network of public agencies, private business
firms, and voluntary organizations was formed. The actors in the
network formulated a joint plan for their efforts to train and recruit
local, unemployed people to the large construction firms and the
many new stores and shops.
A large national bridge-building project stirred up a lot of public

controversy about the environmental problems caused by the
increasing amount of heavy traffic, the destruction of a coastal habi-
tat, and the physical obstacles to the free water flow at sea. In order
to find a sustainable solution to the environmental problems the
national government was forced to negotiate with a large network of
EU-agencies, formal groups of environment experts, environmental
movements, interest organizations, local governments, etc. Public
hearings, joint fact finding, and on-going negotiations resulted in
substantial concessions that in many ways changed the original
project.
The Employment Guidelines formulated by the European Union

stipulate the over-all policy objectives on the basis of which the
member states draft their own National Action Plans and monitor the
results of their implementation of national initiatives in the field of
active employment policy. The Employment Guidelines are a result
of political negotiations which not only involve the European
Commission, several General Directories, and the European Council,
but also representatives from the member states, national and
transnational movements and interest organizations, and various
regional development associations.
Introduction 3
The empirical examples demonstrate the need to abandon the false

choice between state and market governance and to study the role,
function and impact of different kinds of governance networks. In particular
it is clear that policy making and public governance is no longer con-
gruent with the formal political institutions in terms of parliament and
public administration. The formulation and implementation of public
policy increasingly take place in and through interactive forms of
governance involving a plurality of public, semi-public and private actors.
This does not mean that central and local governments are ‘hollowed out’
since many of the old state powers are still in place and new capacities
are developed (Hirst 1994; Pierre & Peters 2000). The state still plays a
key role in local, national and transnational policy processes, but the
state is to an increasing extent ‘de-governmentalized’ as it no longer
monopolizes the governing of the general well-being of the population
in the way that it used to (Rose 1999). As such, the idea of the sovereign
state governing society top-down through comprehensive planning,
programmed action and detailed regulations is losing its grip, and is
being replaced by new ideas about a pluricentric governance based on
interdependence, negotiation, and trust.
The rise of governance network research
The study of governance networks is a novel research field that is founded

on the ‘discovery’ of non-hierarchical forms of governance based on
negotiated interaction between a plurality of public, semi-public and
private actors. Whereas interorganizational relations already became a
central research topic in the beginning of the 1970s (Evan 1976; Heclo
1978), the role of horizontal networks of organized interests in the
production of public policy and governance only came to the fore in the
beginning of the 1990s. The edited volumes of Marin and Mayntz
(1991) and Kooiman (1993) played a crucial role in the promotion of
governance network research. These widely read books were soon
followed up by the equally influential works by Scharpf (1994), March &
Olsen (1995), Rhodes (1997a) and Kickert et al. (1997). More recently,
edited volumes by Heffen et al. (2000), Pierre & Peters (2000), Bang
(2003) and Hajer & Wagenaar (2003) have further consolidated the field
of governance network research, which now constitutes an important
area of research within Public Administration.
In a nutshell, the argument prompting the rise of governance network
research is that policy, defined as the attempt to achieve a desired outcome,
is a result of governing processes that are no longer fully controlled by the
government, but subject to negotiations between a wide range of public,

semi-public and private actors, whose interactions give rise to a relatively
stable pattern of policy making that constitutes a specific form of regula-
tion, or mode of coordination (Mayntz 1993a, b). It is this pluricentric mode
of coordination that in the literature is dubbed governance networks.
The construction of, or reliance upon, governance networks is by no
means a new phenomenon. In many countries and policy areas there is
a long tradition for corporatist involvement of the social partners in the
formulation and implementation of policy, especially at the level of
national policy making and at the state level in federal systems. There
are numerous examples of public policy and governance being based on
political dialogue between public authorities and functionally defined
groups and organizations rooted either in the capitalist market economy
or civil society. However, the new thing is that political theorists and
central decision makers to an increasing extent tend to view governance
networks as both an effective and legitimate mechanism of governance.
The inclusion of relevant and affected groups and organizations in
governance networks help to overcome problems in terms of societal
fragmentation and resistance to policy change, and thus tends to make
the governing processes more effective (Mayntz 1993a). At the same
time, the participation of a plurality of stakeholders in the decision-
making process tends to enhance the democratic legitimacy of the public
policy and governance (Scharpf 1997). These practical and scientifically
justified insights are reflected in and underlie a large number of govern-
ment reports that recommend the use of governance networks at the
local, national, and even the transnational level (EU 2001a, b).
The increasing prominence of governance network research is rooted in
central insights developed in organization theory and political theory
(Klijn 1997). The conception of organizations as open systems adapting to
changes in the environment (Mintzberg 1979), and the recognition that
this environment consists of other organizations paved the way for a new
focus on the interorganizational exchange of information and resources
that takes place in and through relatively stable forms of interorganiza-
tional negotiation (Benson 1978; Aldrich 1979). Likewise, the recognition
of the limits of the corporatist and neocorporatist image of an ‘iron trian-
gle’ between the state, the trade unions and the employers organizations
spurred the attempt of political theorists to distinguish between different
kinds of policy networks, according to their degree of integration, stabil-
ity and exclusiveness (Heclo 1978; March & Rhodes 1992; Rhodes 1997a).
However, if both organizational studies and political theory gradu-
ally arrived at some protean notion of governance networks, the real
Introduction 5
break-through came in the field of policy analysis. Decision-making the-

orists pointed out the limits of the rational actor model in the face of the
complex streams, processes and interactions found with the increasingly
fragmented polity (Cohen et al. 1972) and implementation theorists
demonstrated the failure of central planning and programming in the
face of the resistance of street level bureaucrats, user groups and interest
organizations (Pressman & Wildawsky 1973; Lipsky 1980). The
inescapable conclusion was that the formation and implementation of
policy become more efficient if the key actors are somehow included in
the policy process (Sabatier 1986; Bogason 1991; Mayntz 1993b). The
study of inclusive, participatory and interactive policy making in
and through various forms of governance networks was facilitated, and
further stimulated, by the adoption of new methods of investigation in
policy analysis that recommended a ‘backward mapping’ of the political
actors that contributed to the production of a certain policy output
(Hjern & Hull, 1984). By asking the question of who were involved in
the decision-making process that produced this particular outcome, one
is bound to identify a network of social and political actors.
The recent academic focus on governance networks has an empirical
background in the widespread recognition of the increasingly frag-
mented, complex and dynamic character of society (Kooiman 1993;
Jessop 2002; Klijn & Koppenjan 2004). Fragmentation increases as a result
of the functional differentiation of society into relative autonomous
subsystems and the proliferation of relatively independent public and
private organizations. At the same time, the break-up of the sedimented
forms of political alignment and identification spurs the multiplication
of political actors and identities. Complexity increases as a result of the
growth and interweaving of ‘wicked problems’ characterized by blurred
and contextual understandings of the nature of the problem, the effects
of various solutions and the potential conflicts between the stakehold-
ers. It is further augmented by the concurrence of the growing demand
for knowledge-based decision making and the advent of new forms of
risk and uncertainty. Finally, new societal dynamics are created partly by
the multiplication and interconnection of spatial and temporal horizons
of action, and partly by the blurring and contestation of the boundaries
between institutions, sectors and regulatory scales. This leads to a con-
tingent articulation and interaction of different rationalities, procedures
and strategies for public policy making and governance, which in turn
give rise to new and unpredictable developments. The central decision
makers in public and private organizations at different regulatory scales
tend to see the new forms of interactive network governance as the
most suitable answer to the challenge posed by increasing societal

fragmentation, complexity and dynamism.
Leading politicians and entrepreneurial administrators have taken
network governance to their hearts. Institutionalized contacts and nego-
tiations with relevant and affect actors are frequent and the formation of
different kinds of boards, committees, partnerships and dialogue groups
where stakeholders are invited to contribute to public policy and
governance is a standard procedure. The increasing reliance on gover-
nance networks is a part of a new governmentality. Hence, the key
ambition of what is referred to as ‘advanced liberal government’ is to
‘govern at a distance’ by involving a plethora of ‘intermediary groups’
(citizen groups, professionals, voluntary organizations, social partners
and private firms) in the governing of society through the construction
of self-regulating networks of responsible actors (Rose & Miller 1992;
Dean 1999). The mobilization of the free action of resourceful and ener-
getic actors within a framework of norms and rules that ensures a certain
degree of conformity with the overall objectives of government can
relieve the state from some of it regulatory burdens and help to reach
various target groups with the deployment of less resources and less dis-
play of repressive control from above (Rose 1996).
The new ways of thinking and practicing governance through the
shaping of interactive networks of capable and responsible actors spurs
the academic interest in studying governance networks. Governance net-
work researchers are convinced that they are on to something, and their
own enthusiasm about the new forms of governance feeds back to organ-
izations, administrators and politicians. The result is a self-perpetuating
growth of a new research area which is still young and has not yet become
sedimented into a new paradigm with its own clear-cut definitions,
taxonomies and methods.
The new and rapidly growing field of governance network research
is cross-disciplinary, problem-driven, multi-level, comparative, and
stimulates interactive research. Political studies of institutions, power and
decision making are articulated with sociological studies of culture, com-
munication and social control and organizational studies of cognitive
frames, learning and resource exchange. Different theoretical approaches
are drawn upon in the attempt to address research problems derived from
studies of concrete, empirical cases of network governance. The research
strategy will often be based on retroduction rather than pure cases of
empirical induction or theoretical deduction. The analytical focus on pol-
icy output and the backward mapping of the key policy actors will often
lead to the unravelling of multi-level governance networks that include
Introduction 7
actors from different local, national and even transnational levels. To bet-
ter understand the distinctive forms, dynamics and outcomes of network
governance one can benefit from using a comparative research design
that facilitates comparison between most similar, or most dissimilar, gov-
ernance networks. Finally, the close study of interactive modes of policy
making based on a problem-driven, retroductive research strategy will
tend to stimulate interaction with the analytical object. Interaction with
the network actors can help to sharpen the research questions, to provide
access to hidden aspects of the network processes and to validate the ana-
lytical conclusions on a pragmatic basis.
The aims of this book
Governance network research is empirically oriented, but the study of

the conditions, functioning and implications of interactive forms of
network governance certainly carries an explanatory ambition, though
not in the classical sense of aiming to establish deterministic causalities
with a law-like character. The goal is to produce an open-ended, context-
bound knowledge that is relevant for the actors engaged in network
governance. Explanation is established through theoretically informed
empirical analysis of network-based policy processes and policy outcomes.
These processes and outcomes are results of the institutionally conditioned
interactions between the network actors; the particular form and character
of the network; the attempts by outside agencies to regulate its form and
functioning; and the external conditions for network governance includ-
ing the socioeconomic conditions, the predominant governmentality,
and the co-existence of different modes of coordination.
The theoretical underpinning of the explanatory ambitions is not
always as clearly delineated as one might wish. The scholars within the
first generation of governance network research seldom explicate their
theoretical point of departure and they tend to borrow concepts and
arguments from other scholars in the field, thus producing a somewhat
eclectic and confusing theoretical landscape. The result is not only the
lack of sufficient clarity and rigour, but also a failure to see the produc-
tive differences between the theoretical positions in the field of gover-
nance network research.
In order to solve this problem and enhance the theoretical self-
awareness of governance network theory, this edited volume aims to
renew and refocus the theoretical and analytical debate on governance
networks by raising a series of pressing questions about: 1. the dynamics
of governance networks; 2. the conditions for governance network
success and failure; 3. the forms and functions of metagovernance; and

4. the democratic implications of network governance. As such, the
book will address key questions like: Why and how are governance
networks formed, developed, reshaped and terminated? What are the
conditions for governance networks to produce public policy and
governance on the basis of stable, negotiated interaction between
interdependent, but relative autonomous actors? How is it possible for
political authorities of various kinds to regulate self-regulation governance
networks in order to minimize the risk of governance failure and
maximize the prospect of success? How can be assess the problems
and merits of governance networks in relation to normative standards
of democracy, and what is the result of such an assessment? Leading
scholars in the field of governance network research will attempt to
answer these fundamental questions by drawing on different theoretical
approaches while paying attention to the analytical tensions and
similarities between them.
The book is conceptual and problem-oriented in its focus, and
systematic and comprehensive in its analysis. As such, it aims to define
and clarify the central concepts that are used within governance net-
work research and apply them in analytical discussions of the problems
that are frequently encountered in concrete governance practices.
The problems and research questions are identified in relation to the
successive steps in the governance process and the answers draw on a
wide ranging spectrum of relevant theories.
The overall ambition of the book is to create a platform for the
development of a second generation of research into the problems and
potentials of new forms of interactive governance that tend to spread
faster and wider than most academics have hitherto recognized.
Defining governance networks
Today, everybody talk about ‘governance’ and ‘networks’. The problema-

tization of the idea of the sovereign power of government has prompted
a wide-spread circulation of the notion of ‘governance’, and the com-
plex interaction of economic, social and political actors, sites and
processes have stimulated the use of the ‘network’ metaphor. Both terms
are blanket terms that seem to cover many different phenomena.
As such, the notion of ‘governance’ is deployed in the World Bank’s
recommendation of ‘good governance’, international relations debates
about the possibility of ‘governing without government’, discussions
Introduction 9
about the institutional conditions for ‘economic governance’, the private

sector’s vision of new forms of ‘corporate governance’, the discourse of
New Public Management (NPM), and, more generally, in attempts to
capture the new patterns of public–private cooperation and partnership
(Rhodes 1997a; Kersbergen & Waarden 2004). The notion of ‘networks’
is an equally popular term than enters into many different conceptual
constellations. As such, there are frequent references to communication
networks, inter-firm networks, social networks, professional networks,
cross-border networks, terror networks, and network-centric warfare
( Jessop, 2002). In a somewhat different sense, Castells (1996) even claim
that we live in a network society.
To gain conceptual precision we will have to narrow scope of application
of the two new, fashionable, but notoriously slippery, terms. Hence, we
shall take the notion of ‘governance networks’ to refer to a particular
type of networks and a particular form of governance. As such, we shall
define a governance network as: 1. a relatively stable horizontal articu-
lation of interdependent, but operationally autonomous actors; 2. who
interact through negotiations; 3. which take place within a regulative,
normative, cognitive and imaginary framework; 4. that is self-regulating
within limits set by external agencies; and 5. which contributes to the
production of public purpose. This definition includes most of the fea-
tures commonly ascribed to governance networks (Rhodes 1997a: 53;
Jessop 2002: 228f.). It makes no claim to originality, but rather aims at
capturing the essence of that which is commonly referred to by the
notion of governance networks.
Let us attempt to unpack this dense definition of governance
networks by considering each of the defining aspects in turn. First,
governance networks articulate a number of private, semi-public and
public actors who, on the one hand, are dependent on one another’s
resources and capacities, and, on the other hand, are operationally
autonomous in the sense that they are not commanded by superiors to
think or act in a certain way (Marin and Mayntz, 1991). In order to
become a part of a particular governance network the political actors
must demonstrate that they have a stake in the policy issues at hand and
that they can contribute resources and capacities of a certain value to
the other actors. The relations of interdependency mean that the
network actors are horizontally rather than vertically related. However,
the horizontal relation between the actors does not imply that they are
equal in terms of authority and resources (Mayntz 1993b: 10f.). There
might be asymmetrical allocations of material and immaterial resources
among the network actors, but since participation is voluntary and the
actors are free to leave the network, and since the actors are mutually
dependent on each other, nobody can use their power to exert hierarchical
control over anybody else without risking to ruin the network.
Second, the members of governance networks interact through
negotiations that combine elements of bargaining with elements of delibera-
tion. The network actors may bargain over the distribution of resources
in order to maximize outcomes. But in order to facilitate the develop-
ment of negative and positive coordination (Scharpf 1994), this bar-
gaining must be embedded in a wider framework of deliberation that
facilitates learning and common understanding. However, deliberation
within governance networks will seldom lead to unanimous consensus
(Klijn & Koppenjan 2000: 146f.) since it takes place in a context of
intense power struggles that tend to breed conflict and social antago-
nism. As such, joint action will often rest on a rough consensus where a
proposal is accepted despite persistent disagreement.
Third, the negotiated interaction between the network actors does not
take place in an institutional vacuum. Rather, it proceeds within a
relatively institutionalized framework, which is more than the sum of its
parts, but do not constitute a homogenous and completely integrated
whole (March & Olsen 1995: 27 ff.; Scharpf 1997: 47). The institutional-
ized framework is amalgam of contingently articulated ideas, conceptions
and rules. As such, it has a regulative aspect since it provides rules, roles
and procedures; a normative aspect since it conveys norms, values and
standards; a cognitive element since it generates codes, concepts and
specialized knowledge; and an imaginary aspect since it produces
identities, ideologies and common hopes.
Fourth, the governance networks are relatively self-regulating since they
are not part of a hierarchical chain of command and do not submit
themselves to the laws of the market (Scharpf 1994: 36). Rather, they
aim at regulating a particular policy field on the basis of their own ideas,
resources and dynamic interactions, and do so within a regulative, nor-
mative, cognitive and imaginary framework that is adjusted through
negotiations between the participating actors. Nevertheless, governance
networks always operate in a particular political and institutional envi-
ronment that must be taken into account, since it both facilitates and
constrains their capacity for self-regulation.
Fifth, governance networks contribute to the production of public
purpose within a certain area (Marsh 1998). Public purpose is an expres-
sion of visions, values, plans, policies and regulations that are valid for
and directed towards the general public. Thus, the network actors are
engaged in political negotiations about how to identify and solve
Introduction 11
emerging policy problems or exploit new opportunities. Networks that

do not contribute to the production of public purpose in this broad
sense cannot be counted as governance networks.
Governance networks, as defined above, can take many different
empirical forms depending on the political, institutional and discursive
context in which they emerge. They might be dominated by loose and
informal contacts, but they can also be tight and formal. They can be
intraorganizational or interorganizational; self-grown or initiated from
above; open or closed; short-lived or permanent; and have a sector-specific or
society-wide scope. Finally, some governance networks might be preoccu-
pied with the formulation of policy, whereas others are preoccupied with
policy implementation. The multiplicity of governance networks attest to
the broad relevance of the concept for describing the contemporary forms
of societal governance.
Many actual forms of interactive governance will fail to meet all the
criteria specified in the definition of governance networks. However,
they may still carry many of the defining traits of governance networks
and thus form part of the broad family of governance networks. Exactly
where to draw the line between what constitute a governance network
and what does not is a matter of discretional judgement in relation to a
particular empirical case. But one should bear in mind that governance
network research is not preoccupied with classification, but rather aims
to provide a new perspective on how society is governed.
Merits and problems of network governance
Whereas the early governance network theorists saw governance networks

as a synthesis of state and market (Mayntz 1991: 11), later governance
network theorists tend to see governance networks as a distinctive
mechanism of governance that provides an alternative to state and mar-
ket (Rhodes 1997b: xii; Jessop 2002: 228 ff.). As such, governance net-
works distinguish themselves from the hierarchical control of the state
and the competitive regulation of the market in at least three different
ways. In terms of the relationship between the actors, governance networks
can be described as a pluricentric governance system as opposed to the
unicentric system of imperative state regulation and the multicentric
system of competitive market regulation (Kersbergen & Waarden 2004:
148). Hence, imperative state regulation is based on the undisputed
centrality and power of the state that turns everybody else into sub-
jects with clearly defined rights and obligations. Competitive market
regulation is based on an infinite number of self-interested actors who
are not bound by any common purpose or obligation. By contrast,

governance networks involve a large number of autonomous, but
mutually dependent, actors who interact in order to produce public
purpose.
In terms of decision making, governance networks are based on a
reflexive rationality as opposed to the substantial rationality that
governs imperative state regulation and the procedural rationality
that governs competitive market regulation (Mayntz 1993b: 13 ff.;
Scharpf, 1997: 46; Jessop 2000a: 6–7). Whereas imperative state regula-
tion aims to translate the substantial political values of the government
into detailed laws and regulations that are implemented and enforced
by publicly employed bureaucrats, competitive market regulation relies
on the invisible hand of the market forces that leads to a Pareto-optimal
allocation of goods and services in so far as the rules and procedures
ensuring free competition are carefully observed. By contrast, governance
networks make decisions and regulate various issues in and through
reflexive interaction that involves on-going negotiation between a
plurality of actors who build on their interdependencies in order to pro-
duce joint decisions and collective solutions in the face of persistent
conflicts between diverging interests, conceptions and world views
(Mayntz 1993b: 14; Scharpf 1994).
Finally, compliance with collectively negotiated decisions is neither ensured
by means of the legal sanctions of the state or out of fear of economic
loss on the market. Rather, it is ensured through the generation of gen-
eralized trust and political obligation which over time become sustained
by self-constituted rules and norms (Nielsen and Pedersen, 1988).
The distinctiveness of each of the three mechanisms of governance
implies that the governing of society must be based on a pragmatic choice
between state, market and governance networks, or of the right combina-
tion of these different modes of coordination ( Jessop 2002: 246).
Depending on the political task or policy problem at hand the central
decision makers must choose between different governance mechanisms
that each have their own particular strengths and weaknesses.
Today, governance networks are increasingly seen as a suitable
response to the question of how to tackle complex, uncertain and
conflict-ridden policy problems. Hence, both political scientists and
political decision makers praise governance networks for their potential
contribution to efficient governance. Efficiency gains derive from the
distinctive features of governance networks. First of all, governance
Introduction 13
networks are claimed to have a large potential for proactive governance

as the network actors can identify policy problems and new opportuni-
ties at a relatively early stage and produce flexible responses that allow
for adjustments to the complexity and variety of the concrete condi-
tions (Klijn & Koppenjan 2000: 114; Kooiman 2000: 155f.).
Second, governance networks are seen as important instruments for
the aggregation of information, knowledge and assessments that can
help qualifying political decisions. The network actors often have a deep
knowledge that is relevant for policy making and public governance,
and when the knowledge of all the actors is added up, it represents an
important basis for making an ‘intelligent’ choice of a feasible option
(Kooiman 1993: 4; Scharpf 1999: 20).
Third, governance networks are said to establish a framework for
consensus building, or, at least, for the civilizing of conflicts among
stakeholders. Governance networks tend to develop their own logic of
appropriateness that regulates the process of negotiation, the formation
of a rough consensus, and the resolution of endemic conflicts (Mayntz
1993b: 17; March and Olsen 1995: 27 ff.).
Finally, governance networks are supposed to reduce the risk of
implementation resistance. If the relevant and affected actors are
involved in the decision making process they will tend to develop a
sense of joint responsibility and ownership for the decisions and this
will oblige them to support, rather than hamper, the implementation
process (Sørensen and Torfing, 2003: 614).
Now, the problem is that the potential efficiency gains of governance
networks can only fully realized in well-functioning governance networks.
Changes in the composition of the network actors, the presence of
unresolved tensions and conflicts, weak and ineffective leadership,
frustration over the lack of clear and visible results, and external events
that disturb the policy process can destabilize governance networks and
turn them into malfunctioning talking shops. A careful metagovernance
of the self-regulating governance networks might prevent major dislo-
cations and mitigate the detrimental impact of various disturbances, but
it is a difficult task to optimize the functioning of governance networks
on all dimensions (Kickert et al. 1997: 9; Klijn & Koppenjan 2004b: 114).
Nevertheless, even apparently well-functioning networks might cause
problems for public policy making either by blocking new and innovative
policy initiatives or by trying to shift the costs of their own policy-
solutions to outsiders (Scharpf 1993). In this case governments and other
political authorities must use their powers to influence the composition,

conceptions and incentives of the network actors.
First and second generation research
Although the field of research is relatively new, it is possible to talk

about a first and a second generation of governance network research.
The first generation set out to convince us that something new was going
on. As such, it was primarily preoccupied with explaining why gover-
nance networks are formed; how they differ from the hierarchical rule of
the state and the anarchy of the market; and how they contribute to
effective and proactive governance within different countries and policy
fields and at different regulatory scales. The first generation did an excel-
lent job in linking the rise of network governance to new societal trends
(Kooiman 1993), in fleshing out the distinctive features of governance
networks vis-à-vis state and market (Mayntz 1993b; Rhodes 1996; Jessop
1998) and in analysing the function and effects of governance networks
in different countries, policy areas and levels of regulation (Mayntz &
Marin 1991).
Slowly but surely, the research agenda has moved beyond the preoc-
cupations of the first generation of governance network research.
Governance networks no longer represent something new and exotic;
they are something we must live with and make the best of. At this
stage, new and as yet unanswered questions have come to the fore and
constitute the research agenda of a second generation of governance
network research (Pierre 2000). The emergence of a new generation is
neither a matter of new and younger researchers entering the field nor a
matter of a clear break with the past; rather, it is a matter of a gradual
renewal and enlargement of the research agenda to include four pressing
questions:
1. How can we explain the formation, functioning and development of

governance networks?
2. What are the sources of governance network failure and the
conditions of success?
3. How can inter alia public authorities regulate self-regulating governance
networks through different kinds of metagovernance?
4. What are the democratic problems and potentials inherent to
network governance?
Introduction 15
Some governance network theorists have drawn our attention to the

dynamic aspects of governance networks, either in terms of the institu-
tional conditions for developing the identities and capabilities of the
actors and the political accounts and adaptiveness of the network as a
whole (March & Olsen 1995), or in terms of the isomorphic pressures on
different organizational fields for adopting particular organizational
designs or modi operandi (Powell & DiMaggio 1983, 1991). However, it is
also crucial to explore the factors triggering and facilitating the forma-
tion of governance networks as well as the reasons and conditions for
dismantling or reshaping obsolete or dysfunctional network structures.
Other researchers have pointed to the fact that governance networks
are no panacea, but regularly fail to sustain processes of negotiated inter-
action that lead to the production of public purpose. The recognition of
the possibility of governance network failure calls for a systematic
analysis of the political and institutional conditions for success with
regard to the formation of well-functioning governance networks that
contribute to societal problem solving (Scharpf 1994; Jessop 2002).
In continuation with this, many theorists have started to look at how
public authorities and other influential political actors attempt to regu-
late governance networks that are themselves characterized by a high
degree of self-regulation (Kickert et al. 1997; Milward & Provan 2000b;
Jessop 2002). This regulation of self-regulation is referred to as
‘metagovernance’, or ‘network management’, and there have been sev-
eral attempts to distinguish between different strategies, technologies
and tools of metagovernance (Klijn & Koppenjan 2004b; Sørensen 2005).
Finally, there is growing interest in normative questions about the
democratic legitimacy of governance networks (Rhodes 1997b; Mayntz
1999; Pierre and Peters 2000). The recent surge of governance networks
is a result of a pragmatic search for means of effective and proactive
governance. But governance networks cannot be legitimized merely by
reference to the quality of their output, i.e. to their problem-solving
capacity (Scharpf 1997). The input legitimacy of governance networks is
equally important and so is the possibility for ensuring democratic con-
trol and accountability. The assessment of the democratic performance
of governance networks is a difficult task since it is far from obvious
what the relevant standards of democracy are. New research suggests
that we should analyse and assess the ‘democratic anchorage’ of gover-
nance networks both in relation to different political constituencies
(elected politicians, the members of the participating groups and
organizations, and the affected citizens) and in an internal grammar of
democratic conduct (Sørensen & Torfing 2005a). However, more research

on the democratic problems and potentials of governance networks is
needed in order to develop adequate standards for a democratic audit of
network governance.
In sum, the second generation of governance network research raises
a series of important questions to be critically examined and answered
on the basis of different theoretical perspectives. The contributions to
this edited volume will draw on a wide range of theoretical approaches
within institutional theory, governance network theory and democratic
theory. Institutional theory will help to answer the question about the
formation, functioning and development of governance networks
that is dealt with in Part I. Governance network theory will help us to
provide some answers to the question about the sources of governance
network failure and the question of how to metagovern self-regulating
governance networks. These questions are dealt with in Parts II and III,
respectively. Finally, Part IV will introduce democratic theory in order to
answer the normative question about the democratic problems and
potentials of governance networks. Naturally, the clear division of
labour between the three bodies of theory will not be strictly maintained
since the research questions posed by the authors of the individual
chapters often require a pragmatic combination of concepts and
arguments drawn from all three bodies of theory.
The three bodies of theory are defined broadly, but in order to provide
an initial mapping of the three different theoretical terrains, we shall
distinguish between four theoretical positions on the basis of their
different conceptions of social action and societal governance. As such,
the four theoretical landmarks differ in two important respects. First,
they differ according to whether they conceive rational calculation or
culture-bound rule-following to be the driving force of social action. Is
social action driven by self-interested individuals’ rational calculation of
costs and benefits, or is it shaped by rules, norms and values intrinsic to
particular cultures and historical contexts? Second, they differ according
to whether they perceive persistent conflicts or smooth coordination as
the defining feature of societal governance. Some theories tend to view
conflict and power struggles as a constitutive, and yet potentially
destablizing, feature of societal governance, whereas other theories tend
to downplay the role of conflict and instead conceive indifference or
inappropriate action, caused by the lack of coordination, as the major
obstacle to societal governance. The four theoretical landmarks provide
central reference points, not only for people who want to expand a crit-
ical dialogue between different theoretical approaches, but also for those
Introduction 17
Table I.1 Overview of four different governance network theories
Calculation Culture
Conflict Interdependency theory Governmentality theory

[Rhodes 1997a, b; [Foucault 1991
Kickert et al. 1997; Dean 1999; Rose and
Jessop 1998, 2002] Miller 1992]
Coordination Governability theory Integration theory
[Mayntz 1991, 1993; [March and Olsen 1995;
Scharpf 1993, 1994, Powell and DiMaggio 1983,
1997; Kooiman 1993] 1991; Scott 1995]
who want to move beyond the established theories and explore

new land.
Since governance network theory constitutes the theoretical backbone
of this book, we shall briefly present an overview of the four theoretical
positions that emerge when we divide the main theories in the field of
governance network theory according to the analytical distinctions
between calculation-culture and conflict-coordination. A synoptic
overview of the theoretical mapping of paradigmatic theories that are rel-
evant for the study of governance networks is presented in Table I.1.
We have named the four theoretical positions presented in the
four-fold table interdependency theory, governability theory, integra-
tion theory and governmentality theory (Sørensen & Torfing 2005b).
We have added references to particular works of leading theorists that
we conceive to be emblematic for the different theoretical positions.
We should like to stress that some of the theorists referred to in the table
might not even perceive of themselves as governance network theorists.
Whereas governance networks are clearly at the centre of the works of
the theorists associated with interdependency theory and governability
theory, the theorists associated with integration theory and governmen-
tality theory only vaguely, and in passing, refer to ‘relatively stable, self-
organizing networks’ (March and Olsen 1995: 70), ‘organization fields of
relevant actors’ (Powell & DiMaggio 1983: 148), and ‘the construction of
multiple forms of agency through which rule is accomplished’ (Dean
1999: 209). Nevertheless, we deem that they are all making important
contributions to grasping the intricacies of network governance. We
should also like to stress that many theorists cited in the table, and sev-
eral of the contributors to this book, aim to transgress the analytical dis-
tinctions that define the theoretical positions displayed in Table I.1.
However, we believe that the heuristic value of establishing some central
theoretical reference points merits our one-sided and reductive attempt

to simplify the theoretical terrain.
To give an initial flavour of how the four theoretical positions
contribute to the analysis of interactive governance we shall briefly
consider how they define governance networks, explain their forma-
tion, and account for the forces that link the different actors in a gover-
nance network. Interdependency theory has developed in a critical
dialogue with corporatist and policy network theory and has a certain
affinity with historical institutionalism. It defines governance networks
as an interorganizational medium for interest mediation between inter-
dependent, but conflicting actors each of whom has a rule and resource
base of their own. Governance networks are results of the strategic actions
of independent actors who interact because of their mutual resource
dependencies and thereby counteract the institutional fragmentation
caused by NPM. Governance networks are formed through incremental,
bottom-up processes, but are often recruited as vehicles of public policy-
making by public authorities. The network actors seek to realize differ-
ent interests through internal power struggles, but they are held
together by their mutual interdependence, which facilitates negotia-
tion, compromise and joint learning processes.
Governability theory was developed mainly by people associated with
the Max Planck Institute in Cologne. It starts off from a critical investi-
gation of the changing conditions for governing society and has a certain
affinity with rational choice institutionalism. It defines governance net-
works as a horizontal coordination between autonomous actors who
interact in and through different negotiation games. The formation of
governance networks is seen as a functional response to the increasing
societal complexity, dynamics and diversification that undermine the
ability to govern society efficiently through the traditional means of
hierarchy and market. Governance networks are formed through the
construction of game-like situations that enhance horizontal coordina-
tion, and they are held together partly by the anticipated gains from
resource pooling and joint action and partly by the development of
mutual trust and institutional rules that help to overcome collective
action problems. Network-based governance tends to enhance the
governability of modern societies.
Integration theory is a name for the organization sociological studies of
political governance and socioeconomic regulation that take place in
increasingly disintegrated polities. Although it is not always fully recog-
nized by the emblematic theorists themselves, integration theory has a
certain affinity with hermeneutics and social constructivism. It defines
Introduction 19
governance networks as a relatively institutionalized field of interaction

between relevant and affected actors that are integrated in a community
defined by common norms and perceptions. Governance networks may
serve as means for improving the effectiveness and legitimacy of societal
governance, but they can also be seen as a normative response to the
twin problems of totalitarian over-integration and individualistic
under-integration of social agency. They are formed through a bottom-
up process whereby contacts that are established due to the recognition
of interdependence are evaluated and extended on the basis of institu-
tionalized logics of appropriateness. Over time, governance networks
develop their own logic of appropriateness, which is often influenced by
isomorphic pressures. The network actors become normatively and
cognitively integrated through the construction of solidarity and
common identities and frames of reference.
Governmentality theory is the name of a broad research programme
initiated by the poststructuralist, political philosopher Michel Foucault
in the last years before his death in 1984. The theory of governmentality
focuses on how we collectively think and organize the governing of
society. It explores how the present governmentality tends to enrol a
plurality of actors in the governing of particular spheres, groups and
individuals, and thereby contributes to a further decentring of power
and governmental agencies that multiply and form networks with a
‘rhizomatic’ character, to use Deleuze & Guattari’s (1987) famous
metaphor. Governmentality theory is not a network theory per se, and it
does not have any clear definition of governance networks. However, it
implicitly conceives governance networks as an attempt of an increas-
ingly reflexive and facilitating state to mobilize and shape the free
actions of self-governing actors. Governance networks are construed as
a political response to the failure of neo-liberalism to realize its key goal
of ‘less state and more market’. The problematization of neo-liberalism
leads to the formulation of a new governmentality programme, associ-
ated with advanced liberal government that aims to shift the burden of
government to local networks in which the energies of social and polit-
ical actors are mobilized and given a particular direction in order to
ensure conformity. Governance networks are held together and framed
by particular technologies and narratives that recruit social and political
actors as vehicles of power.
There are many family resemblances between the four theories of
network governance, but there are also significant differences. Both simi-
larities and differences will be further explored in Chapters 5 and 9 which
aim to compare the four theories of network governance with regard to
what they have to say about success and failure network governance and
the attempts to metagovern self-regulating governance networks.
Institutional theory and democratic theory can also be theoreti-
cally mapped by using the four-fold Table I.1 presented above. This is
demonstrated in Chapter 1 that compares different theoretical positions
within the new institutionalism with regard to how they explain the
dynamics of governance networks, and in Chapter 13 that compares dif-
ferent democratic theories with regard to their conception of the demo-
cratic problems and potentials of governance networks.
Plan of the book
As already hinted above, the book is divided into Parts I–IV, each
devoted to answering one of the four big questions posed by the second
generation of governance network research. Each part begins with an
overview chapter that aims to flesh out and compare the theoretical
answers to the question at hand that can be found, or construed, on the
basis of the analytical mapping of the four paradigmatic positions
within the relevant body of theory. In each of the four parts, the
overview chapter is followed by three independent chapters that aim to
provide more detailed answers to different aspects of the overall research
question. The authors are free either to associate themselves with one of
the four paradigmatic theories, combine different theories in an innova-
tive way, or transgress the analytical scheme by drawing upon other
relevant theories or developing a new theoretical approach. The three
chapters following the overview chapter are structured in accordance
with a stylized description of the different phases in the political process
of network governance: first, the governance network is formed; then, it
gives rise to negotiated interaction between a plurality of relevant and
affected actors; and finally, it produces some kind of public value.
Part I focuses on governance networks dynamics. The overview
chapter is followed by a discussion of mechanisms of governance
network formation (Chapter 2, Nils Hertting); an analysis of virtuous
and viscous circles in democratic network governance (Chapter 3, Guy
Peters); and a study of attempts to manage change in networks (Chapter
4, Mark Bevir and Rod Phodes).
Part II analyses the conditions for governance network failure by dis-
cussing three crucial dilemmas that confront the network actors and the
metagovernors. First, there is a discussion of the sources and impact of
different forms of closure and openness (Chapter 6, Linze Schaap).
Introduction 21
Then, there is a study of the negative and positive effects of conflict and
consensus (Chapter 7, Joop F.M. Koppenjan). Finally, there is discussion
of the relation between effective and legitimate network governance
(Chapter 8, Tanja Børzel and Diana Panke).
Part III addresses the important question of metagovernance. The
review chapter is followed by a discussion of the formation and
mobilization of governance networks (Chapter 10, Peter Triantafillou); a
study of how to metagovern networks through network management
(Chapter 11, Erik-Hans Klijn); and an analysis of how outputs and
outcomes of networked policy processes can be influenced through
different forms of metagovernance (Chapter 12, Laurence J. O’Toole).
Part IV of the book assesses the democratic implications of network
governance. A brief overview of different approaches to the analysis of
democratic network governance is followed by a discussion of the
democratic norms for participation in governance networks (Chapter 14,
Allan Dreyer Hansen); the demands for democratic dialogue and com-
munication (Chapter 16, John Dryzek); and the challenges to democratic
accountability (Chapter 16, Anders Esmark).
The conclusion takes stock of the answers provided by the various
authors and assesses the need for further research of the research
questions that have been posed in this volume. It also discusses the
future directions of governance network research that might expand, or
move beyond, the research agenda of the second generation.
This page intentionally left blank
Part I
Governance Network Dynamics
1
Theoretical Approaches to
Governance Network Dynamics
Introduction
Governance networks contribute to the production of public policy and

governance. Political visions, policy ideas, comprehensive plans,
informal norms and detailed regulations are often crafted, or at least
influenced, through policy processes involving relevant and affected
actors from state, market and civil society. The networked policy output
is a contingent result of negotiated interaction between a plurality of
interdependent, and yet operationally autonomous, actors. The form
and character of the policy output depends on the form and character of
the horizontal interplay between the network actors. The negotiated
exchange between the various actors changes over time and varies from
governance network to governance network. Basically, network-based
governance is a complex and potentially chaotic process in which
numerous interests, identities and rationalities fuse and collide.
However, the interaction between the network actors does not take
place in a void, but rather proceed within a relatively institutionalized
framework that facilitates and constrains political interaction and thus
affects the production of the outputs and outcomes of governance
networks. The institutionalized framework consists of more or less
sedimented rules, norms, cognitive paradigms, and social imaginaries
that are constructed in and through negotiated interaction. The fact
that governance networks are stabilized by the contingent and tentative
formation of rules, norms, etc. does not turn governance networks into
organizations in the sense of relatively formal and unified institutional
actors. Governance networks might recruit public agencies, social
partners, social movements and other forms of political organizations,
and they might even aspire to form, or become, an organization.
25
However, governance networks are not themselves organizations since

they generally lack the defining characteristics of organizations in terms
of an overriding and unifying objective, a political leadership capable
of imposing formal sanctions on the participants, and a chain of com-
mand permitting governing by decree. The network actors’ autonomy and
interdependency prevent the exercise of hierarchical monitoring and
control that is normally associated with Weberian-type organizations.
Governance networks cannot even be defined as institutions in the
strict sense of well-integrated systems of social interaction based on
relatively fixed rules, norms and procedures. Governance networks are
marked by a profound institutional ambiguity since, at the beginning,
there are no clearly defined and commonly accepted rules, norms and
procedures and no formal constitution that predetermine how legiti-
mate decisions are made (Hajer & Versteeg 2005). Governance networks
bring together a wide range of political actors who have different rule
and resource bases. The network actors engage in an intercultural and
pluricentric interaction marked by conflicts, ambiguity and uncertainty.
In the outset, the actors might try to establish a mode of interaction that
facilitates cooperation and joint decision-making. They might even try
to stabilize this mode of interaction through the construction of formal
and informal rules and the development of shared norms, values and
conceptions. However, this does not transform governance networks
into institutions in the sense of well-integrated social systems that are
based on common rules, norms and compliance mechanisms that
enable a stable reproduction of the social pattern of interaction and the
role and identity of the actors. Governance networks are complex and
dynamic systems in which centripetal and centrifugal forces constantly
undermine each other so that order and stability only exist as a partial
limitation of disorder and instability.
Governance networks bring together a range of relevant and affected
actors in contingent interactions that over time may become subject to
processes of institutionalization and de-institutionalization that con-
struct a regulative, normative cognitive and imaginary framework for
negotiation and joint decision making. The institutionalized framework
both enables and constrains the interaction of the network actors.
Hence, it might facilitate interaction by ensuring the codification of
political compromises, lowering the transaction costs of horizontal
coordination, developing the identities and capacities of the actors, and
recruiting individuals, firms and organizations as free, active and
responsible actors. At the same time, it might constrain interaction
by regulating conflicts, influencing the perception of problems,
Theoretical Approaches to Governance Network Dynamics 27
preferences, costs and benefits, defining appropriate actions, and

constructing particular rationalities and technologies. The institutional-
ized framework of interaction is dynamic and incomplete as the net-
work actors will aim to influence and adjust it by invoking elements and
aspects of their own rule and resource bases that will gradually modified
and changed in the course of interaction. Public authorities might aim
to subject governance networks to a particular set of formal rules, norms
and procedures (metagovernance), but the network actors will often
oppose these, or find ways of getting around the formal constraints that
are imposed upon them. The rules, norms, etc. that are constructed
through the self-regulated interaction of the network actors will often
have an informal character and in periods of turbulence and change
governance networks will be hold together by story telling and common
symbols rather than clearly defined rules and norms.
In sum, governance networks are neither organizations nor institutions
in the strict and narrow sense of the terms, but relatively institutionalized
frameworks of negotiated interaction within which different actors strug-
gle with each other, create opportunities for joint decisions, forge political
compromises and coordinate concrete actions. Conflicts sustained by
the cultural, social and political differences between the relatively
autonomous actors prevent governance networks from being transformed
into stable political institutions. However, the relative institutionalization
of governance networks, which facilitates and constrains negotiated inter-
action, permits us to draw on institutional theory in order to understand
the dynamics of governance networks by taking into account the mutual
conditioning between the interaction of the network actors and the ten-
tative rules, norms and procedures that are developed in the course of
action. Institutional theory is indispensable for any attempt to grasp the
functioning and development of governance networks as it helps us to
understand the complex interplay between political agency and the
emerging structures of their interaction.
In this first part of the book we shall draw on theories associated with
the new institutionalism. Whereas the old institutionalism focused on
the formal and constitutionally defined political institutions in terms of
parliaments, governments, and legal systems (see Finer 1954; Duverger
1959; Johnson 1973), the new institutionalism has a much broader
institutional focus that covers not only the formal political institutions,
but also the less formal institutionalizations of the patterns of interaction
between different political actors (Rhodes 1995: 54; Peters 1999: 18).
The new institutionalism claims that in the world of politics ‘institu-
tions matter’ and it sets out to explore the interaction of politics and
institutions through theoretically informed case studies of the genesis,

impact and transformation of formal and informal institutions.
The new institutionalism conceives institutions as a structural feature
of social systems that provides a certain degree of order and stability to
social interaction by means of regulating and affecting the beliefs and
behaviour of the actors. This common conception of the role of institu-
tions hides a number of competing theoretical approaches that tend to
disagree about the nature of institutions, how they affects human
behaviour and how they are constructed and transformed. Although the
analytical labels might differ, it is common to distinguish between three
theoretical positions within the new institutionalism: historical institu-
tionalism, rational choice institutionalism and social constructivist (or
normative) institutionalism (Campbell 1994; Hall & Taylor 1996; Peters
1999). There might be partial overlaps and cross-overs between the three
institutionalist approaches (Torfing 2001), but it is nevertheless possible
to identify some clear differences in their basic analytical assumptions.
The historical institutionalist approach focuses on the institutional
mediation of conflicts and claims that political struggles over scarce
goods may result in political compromises that are institutionally
codified and tend to regulate and influence the future political struggles
giving rise to a path-dependent development (Hall 1993; Pierson 1994;
Thelen & Steinmo 1992). Historical institutionalism sets itself out from
the other approaches by its emphasis on political conflict and power
struggles, but it contains a certain ambiguity in its conception of the
social and political actors. Hence, on the one hand, it sides with rational
choice institutionalism in its understanding of human action as basically
driven by self-interested calculations. On the other hand, it sides with
social constructivist institutionalism in claiming that the interests of the
rational actors are influenced by the rules and norms, cognitive para-
digms and learning processes that are embedded in the institutional
framework of negotiation. However, despite the partial endogeniezation
of the actors’ interests and preferences, the rational calculations and
power resources of the actors gain the upper hand in the explanation of
political outcomes. The interests and preferences might be affected by
the feasible strategies of the actors and their institutionalized interaction,
but when first they are fixed, the actors act rationally in order to maximise
their outcome within the limits set by the institutional framework.
Rational choice institutionalism focuses on the choices made by
individual actors with pre-given preference who are acting on the basis
of a bounded rationality within an institutional setting that defines the
range of options, their pay-offs and the kind of games that are likely to
emerge between strategically interdependent actors (Ostrom 1990;

Dunleavy 1991; Scharpf 1997). In contrast to historical institutional-
ism, the starting point is individual rather than collective actors. The
actors act rationally to maximize personal utility within the constraints
set by the institutional framework of their interactions. The absence of a
common utility function might lead to conflicts between the self-inter-
ested actors. However, the fundamental problem in relation to gover-
nance is not the presence of political conflicts between different
political groups, but rather individual defection or noncompliance with
the rules defined by the institutional framework. However, the institu-
tional design of the right incentives in terms of rewards and punishment
(Calvert 1995) and the repetition of rational games over time might
facilitate greater cooperation and mutual compliance (Axelrod 1984).
Social constructivist (or normative) institutionalism is in many ways
the antidote of rational choice institutionalism (Powell & DiMaggio
1991; Scott 1995). The interests and preferences of the actors are intrinsic
to the identity of the social and political actors which is shaped and
reshaped in a particular institutional context. The actors do not act on
the basis of a logic of consequentiality, but on the basis of a logic of
appropriateness (March & Olsen 1989). They match the institutionally
embedded rules, norms and cognitive paradigms with their own iden-
tity and the situation in which they are placed, and they are acting
appropriately on the basis of their own constitutive interpretations of
the institutionally defined rules, norms, etc. (March & Olsen 1995). The
emphasis on the institutional embeddedness of the actors, and the
broadening of the definition of institutions to include ideas, knowledge,
conceptual codes, symbols, rituals, and other cultural artefacts (March &
Olsen 1989), means that the actors can be conceived in terms of neither
an utility-maximizing ‘economic man’ nor a power-maximizing ‘political
man’. Social actors are normative creatures as their identity, capacities
and aspirations are shaped by the social and political communities to
which they belong. Normative integration of the social actors to
competing conceptions of the common good is seen as an important
condition for democratic governance.
Social constructivist institutionalism is rooted in interpretative sociology
and organizational studies. Its emphasis on the endogenous character of
the actors’ interests, conceptions and operational rationalities is shared
by poststructuralist theories of politics, governance and institutions.
However, these poststructuralist theories tend to be highly skeptical
about the prospect of normative integration due to the ineradicable
presence of power and social antagonism. Although poststructuralism
can hardly be said to provide a clear and coherent theory of institutions,

there has been an increasing emphasis on the interplay between politics
and institutions (Clegg 1990; Dyrberg & Torfing 1992). Foucault (1979,
1990) has always stressed the intrinsic link between power and institu-
tions, and also Laclau (1990) has paid attention to the institutional
sedimentation of politically constructed social relations. As such, it is
possible to add a fourth theoretical approach to the three approaches
mentioned above (Campbell & Pedersen 2001). The poststructuralist, or
discursive, approach to institutional analysis tends to see institutions as
vehicles for local and decentred power strategies that aim to work upon
the free individuals through a combination of discipline, normalization
and governmental self-regulation. Power involves the production of
particular forms of subjectivity that takes place in and through the
articulation of discursive forms of knowledge and truth that over time
become sedimented into institutionalized rules, procedures, rationalities
and totalizing world views.
If we add the poststructuralist perspective on institutional analysis, we
have four theoretical approaches that can be mapped by using the same
analytical distinctions as we used in the mapping of the theories of
network governance presented in the introductory chapter. The emerg-
ing four-fold table is shown in Table 1.1 where a couple of emblematic
works have been added in order to indicate what hides behind the
analytical labels defining the four theoretical approaches to institutional
analysis.
There is a close connection between theories of governance networks
and institutional theory. In fact, each of the four theories of governance
Table 1.1 Overview of four different approaches to institutional analysis
Calculation Culture
Conflict Historical institutionalism Poststructuralist

[Hall 1986, 1989; institutionalism
Peters, et al. 2005; [Clegg 1990; Dean 1999;
Rothstein 1992; Thelen & Foucault 1991a; Rose 1999]
Steinmo 1992]
Coordination Rational choice Social constructivist (or
institutionalism normative) institutionalism
[Moe 1990; Olson 1965; [March & Olsen 1989, 1995;
Ostrom 1990, 1991; Riker & Powell & DiMaggio 1983, 1991;
Ordeshook 1973] Scott 1995]
networks that were presented in the introduction is underpinned by a

particular institutionalist approach. Hence, interdependency theory is
underpinned by historical institutionalism, governability theory by
rational choice institutionalism, integration theory by social construc-
tivist institutionalism, and governability theory by poststructuralist
institutionalism. The neat correspondence between the theories of
network governance and the institutionalist approaches is to some
extent founded on a tautology since in most cases the representatives of
the particular theory of governance networks are more or less identical
with the representatives of the corresponding institutionalist approach.
However, it is by no means a perfect match, and the link between
governance network theory and institutional theory is not always
explicit. Nevertheless, the link is important since there is much to learn
about the dynamics of governance networks from an encounter with
the different strands of the new institutionalism. Let us take a closer
look at the theoretical insights into governance network dynamics that
are provided by the four institutionalist approaches.
Historical institutionalism
Institutional theory can help us to understand the crucial role that

institutions play in the formation, functioning and transformation of
governance networks. Historical institutionalism defines institutions as
the ensemble of formal and informal rules, norms and procedures that
regulate the political action of collective actors. Institutions are
conceived as both the medium and outcome of the political struggles of
a plurality of interest-based actors. Political struggles about the formula-
tion and implementation of public policy take place within an institu-
tional framework which is itself a result of the codification of political
compromises obtained in past political struggles.
When it comes to explaining the formation of governance networks,
institutions play a dual role. On the one hand, governance networks are
developed in an attempt to overcome the institutional and organiza-
tional fragmentation caused by the implementation of the principles of
New Public Management (Rhodes 1997a: 45; 2000: 54). The increasing
‘agentification’ of the local public sector is countered by the formation of
governance networks that facilitate crosscutting coordination between
individual agencies. As such, governance networks are the aggregate
result of local actors’ strategic attempt to cope with the consequences of
institutional reform. On the other hand, when the fragmented, local
actors aim to establish horizontal relations of negotiated coordination
with other actors, the formation, or reactivation, of particular rules,

norms and procedures might help to highlight and reinforce the mutual
dependency between the actors and to regulate the access to the gover-
nance network (Kooiman 1993). Hence, institutionalized norms about
‘holistic governance’ and ‘teamwork’ encourage the actors to recognize
their interdependency, and institutionalized rules and procedures
determine who can be considered as legitimate members of governance
networks.
Historical institutionalism has even more to offer when it comes to
explaining the functioning of governance networks. The negotiated
exchange between the network actors will over time result in the codifi-
cation of political compromises and experiences into a loosely defined
set of rules, norms and procedures that channel, guide and sustain
further interaction within the network and thereby enhance its robust-
ness. The institutional framework helps to stabilize the interactions
within the governance network by providing rules about legitimate
decision making, furnishing norms that support the development of
generalized trust, and establishing procedures for the regulation and
resolution of conflicts. The institutionalization of norms and values
might also change and modify the identities and preferences of the
actors so that the risk of future conflicts is reduced. However, the
institutional framework is not always supporting the development of
well-functioning governance networks. The institutional codifications
might constrain new and innovative policy making because the regulative
and normative framework reflects old and long-forgotten compromises
and fails to take into account the new societal conditions. The institu-
tional framework might also be too weak to tame the fierce conflicts and
power struggles that emanate from an interest-based policy making.
Finally, the institutionalization of divisions between central and peripheral
members of the governance network may sustain, or even strengthen, an
asymmetric distribution of resources between the network actors. This
might lead to a marginalization of certain network actors that will be
tempted to leave the governance network because they have no chance
of being heard. Such a defection will tend to reduce the legitimacy of the
governance network.
The most significant contribution of historical institutionalism is prob-
ably the argument about the path-dependent transformation of gover-
nance networks (Hodgson 1994; Krasner 1984). Institutionalization is
always a matter of degree and in the highly institutionalized governance
networks deliberate attempts to bring about radical changes in their
form and functioning are likely to fail. The contingent codification of
past conflicts, and the institutionalization of strategic responses to past

events, will tend to create a certain lock-in effect, according to which
particular rules, norms and procedures ‘freeze’ the patterns of interaction,
even when it results in a sub-optimal functioning of the governance net-
work. As such, the complexity and opaqueness of governance networks
might spur the development of self-sustaining mental models that
define how the negotiated interaction in the network should be
organized and evolve (Pierson 1998). The institutionally conditioned
path-dependency will not prevent changes in the governance network
and the networked policy making, but only permit evolutionary and
incremental changes that require a lot of political pressure and tend to
favour the preservation of status quo over the dissolution and transfor-
mation of the old path. Revolutions and radical reforms are rare, but
might occur at breaking points in history where the institutional
equilibrium is punctuated by new events that problematize and dislo-
cate the old path and create a branching point where new paths might
be chosen as a result of intense political struggles (Hall 1986).
Rational choice institutionalism
Historical institutionalism claims that institutions facilitate and

constrain action, but it is not always entirely clear about how it is that
institutions actually affect action. This is certainly not a problem in
rational choice institutionalism which clearly specifies how institutions
matter. Institutions are defined more narrowly as the formal and infor-
mal rules and the associated compliance system that are found within a
particular arena of action. The institutional rules determine the external
conditions for the rational choice of individual actors. As such, institu-
tions tend to determine the actual range of choices and options of the
actors; the pay-offs of the different options, or combination of options;
the available information about options, costs and benefits; and the
structure of the game played by the rational interdependent actors.
The development of modern society is said to provide favourable
conditions for the formation of governance networks and for explaining
the forms and functioning of these in terms of the rational choice of
individual actors. The functional differentiation of society into plurality
of relatively autonomous systems, sub-systems and organizations means
that no single public or private actor is capable of governing society
alone (Kooiman 2000: 142). Therefore, governance must necessarily rely
on horizontal coordination between actors coming from different sys-
tems, subsystems and organizations. The rational actors tend to realize
this and they begin to form horizontal networks of societal coordination

because it is rational for them to do so. The actors have a relative auton-
omy, but they are dependent on each other’s resources and expertise in
order to ‘get things done’. As such, they will tend to form governance
networks on the basis voluntary contracts, whenever they perceive the
potential benefits to be higher than the costs.
The formation of governance networks is hampered by the presence
of collective action problems as the actors might choose to free ride.
However, collection action problems might be overcome in institutional
settings where there is a known tradition for giving real influence to
governance networks; where participation gives access to important and
scarce resources and immediate benefits in terms of political recognition
and social status; where the entry rules ensure that the most important
actors are included; and where the transaction costs of networking are
lowered by mutual trust.
Collective action problems are not only hampering the formation of
governance networks, but might also jeopardize their general functioning.
Opportunistic action, and/or indifference to the negative externalities of
one’s own actions, provides a constant threat to horizontal coordination
through negotiated interaction. It is tempting for the network actors to
try to reap the fruits of the network while contributing as little as possi-
ble to its general functioning and disregarding the costs that are gener-
ated by the self-interested action of the winning coalition within the
network. Therefore, the construction of well-functioning governance
networks will depend on the creation of institutional rules, norms and
incentive structures that will favour games that build relations of trust
and cooperation rather than distrust and competition. Hence, ‘split-a-
dollar’ games are to be preferred to ‘prisoners-dilemma’ games, although
repeated prisoners dilemma games that facilitate learning might also do
the trick. In general, the preferred games are those that emphasize
the positive interdependence of the actors and increase the costs of
non-cooperation.
Rational choice institutionalism generally claims that institutions are
difficult to transform, or dissolve, because they often assume the
character of a private good for the involved actors. This argument also
applies to governance networks which may provide a means for the par-
ticipating actors either to gain access to particular benefits and resources
that would otherwise be inaccessible, or to shift the costs of particular
plans or regulations to the external environment. In situations where
governance networks have assumed the character of a private good for
the actors, institutional transformation requires deliberate attempts to
change the preferences of the actors. When first the individual preferences
of the actors are changed, it will be rational for them to change the
institutional rules and norms that regulate their negotiated interaction.
However, rational choice institutionalism has little to say about how
preferences can be changed and often falls back on the somewhat naïve
assumption that external changes in the environment will automatically
alter the preferences of the individual actors.
Social constructivist (or normative) institutionalism
Social constructivist institutionalism also has a relatively clear

understanding of how institutions affect the action and interaction of
network actors. Institutions are defined broadly as not only the ensemble
of formal and informal rules, norms and procedures, but also the know-
ledge, values, codes and conceptions that informs and support them
(March & Olsen 1989: 22). Institutions affect the identities, perceptions,
capacities and routines of the actors which in turn determine their
actions and interactions. Institutions are not reduced to an external
constraint on the rational action of individuals, but define the repertoire
of appropriate action of different kinds of agency. Rational action might
be appropriate in a certain institutional context, but mostly rational
calculation is used post hoc to justify action which is formed by a
particular logic of appropriateness (March & Olsen, 1995: 8, 29).
However, the emphasis on rule-governed action does not mean that the
actors are structurally determined by the institutional context of their
actions. The matching of the particular situation to a certain rule (March &
Olsen, 1989: 25f.) and the interpretation of the matching rule both call
for a constitutive intervention of human agency who must select a rule
and determine its content on the basis of a legal reasoning that aims to
compare the present situation with past experiences (March & Olsen,
1995: 32f.).
Social constructivist institutionalism provides both a micro-level
understanding of how institutions affect the formation of governance
networks and macro-level understanding of how institutional designs
that favours the formation of governance network are spread. The
micro-level theory of network formation emphasizes the interaction
between interdependent organizations and the institutional logic of
appropriateness on which these organizations are based (March &
Olsen, 1995: 107 ff.). Organizations tend to contact other organizations
because they are dependent on each others resources and capacities.
Who they might contact is not so much determined by a rational
calculation of the possible outcome of that particular contact, but rather

depends on institutional norms specifying who it is appropriate to
contact. The actors constantly evaluate their interorganizational con-
tacts and this is done on the basis of the rules, norms, values and
conceptions that are found within their particular group, or organiza-
tion. The contacts that are deemed positive on the basis of the internal
logic of appropriateness will tend to become repeated and this might
over time lead to the formation of a governance network. This might
even institutionalize its own logic of appropriateness. If the network is
not already organized around a particular public task, or policy problem,
it might later become the locus of public governance and policy making.
March and Olsen’s bottom-up explanation of the institutional condi-
tions for network formation is supplemented by Powell and DiMaggio’s
(1983) top-down explanation of the spread of contingent organizational
designs through isomorphic pressures that force organizational fields to
adopt a particular organizational principle which, today, might well be
that of interactive network governance. Public and private organizations’
concern for legitimacy often overrules their concern for efficiency.
Organizations that operate in uncertain environment may seek to obtain
legitimacy by giving in to isomorphic pressures. Coercive isomorphism
involves the adoption of organizational designs recommended by
higher level political authorities from which the organizations in the
organizational field receive important resources. Mimetic isomophism
involves voluntary attempts to copy designs from other organizations
that appear to be successful and legitimate. Finally, normative isomor-
phism involves adopting organizational designs favoured by the profes-
sional groups inhabiting particular organizations. The combination of
different isomorphic pressures provides a strong homogenizing force.
However, Powell (1991b: 194 ff.) has later brought attention to the fact
that heterogeneity and ambiguity in the environment of the organization
field might weaken the isomophic pressures and their homogenizing
force. In much the same vein, Røvik has emphasized the need for
organizational translation of organization designs through operational-
ization, selection, combination and fusion of their different elements.
To this Friedland and Alford (1991) have added that there will often be
conflicts between and within organizations about which organizational
designs to pursue and which to reject. All in all, social constructivist
institutionalism provides us with a well-developed theory about the
factors that might facilitate or constrain the homogenizing spread of
organizational designs that, for example, favours the formation of
governance networks.
Institutions also affect the functioning of governance network by

providing the conditions for developing the identities and capacities of
the network actors (March & Olsen, 1995: 49f.). Identity is defined as
the sum of our often simplified images of ourselves and others, our
interpretations of these images, and our attempts to live out these inter-
pretations in and through our social and political practice (March &
Olsen, 1995: 74). Likewise, capacities are defined as the sum of the
rights, resources, competences and organizational know-how which the
individual and collective actors possess (March & Olsen, 1995: 92). Well-
functioning governance networks require the construction of solidarity
among the actors; the creation of identities that are compatible with and
supports democratic political processes; and the civilization of conflicts
between different solidarities and identities (March & Olsen, 1995: 63
ff., 72 ff., 178 ff.). The identities of the network actors are shaped and
reshaped through socialization, education, reflection, action and inter-
action which is conditioned by the logic of appropriateness that has
been developed within the governance network.
The development of democratic political identities must be combined
with attempts to develop the capacities of the network actors through
political empowerment whereby rights are created, resources are
generated and distributed, and competences are enhanced (March &
Olsen, 1995: 91 ff.). However, the transformation of rights, resources
and competences into political action that makes a difference requires
the development of the political reflexivity and know-how of the actors.
It goes without saying that the institutional conditions for network
governance is crucial for the political empowerment of the network
actors. In addition, we should not forget that the development of the
political capacities of the actors also depends on the development of
their political identity. In fact, capacities create identities, and identities
create capacities (March & Olsen, 1995: 103).
As for the transformation of governance networks, social constructivist
institutionalism firmly rejects the belief in the ‘efficiency of history’ that
guarantees a perfect match between institutions and their external envi-
ronment (March & Olsen, 1995: 10f., 40 ff.). There is no mechanism
whereby changes in the environment automatically induce changes in
the preferences of the network actors which in turn lead to intentional
reforms that construct an optimal institutional design. The rationalist
account of institutional change completely overlooks how institutions
condition the actors’ perception of their environment and how organi-
zations aim to control and shape their environment. The alternative
historical institutionalist account depicts political institutions as stable
configurations that have their own internal dynamic that is protected

from the environment through ignorance, distancing, or rejection
(Krasner 1988). Institutions live their own quiet life until the day where
the gulf between the demands from the environment and the internal
rules and procedures has become too big and the institutions are either
destructed or transformed through big and fatal crises. March and Olsen
(1995: 185 ff.) tend to agree with the main argument advanced by the
historical institutionalists, but they urge us not to forget that institu-
tions also have some relatively stable procedures for how to gather
information about and assess changes in the environment and eventually
transform and adapt the institutional rules and norms in accordance
with the institutionally embedded logic of appropriateness.
Poststructuralist institutionalism
Poststructuralist institutionalism does not constitute a unified theoreti-

cal paradigm and fails to provide a comprehensive account of the
formation, functioning and transformation of institutions, organizational
fields, or governance networks. In fact, poststructuralist institutionalism
does not even aim to develop such an account. Rather, it seeks to offer a
certain problematization of the institutions of governance by means of
revealing their political ‘origin’ in a myriad of political struggles, their
discursive conditions of possibility, and their power effects. Foucault
(1991a: 103) criticizes the liberal and Marxist obsession with the state
and claims that instead of searching for the essence of the state, we
should analyse how the unity, competences and limits of the state are
produced in and through the dominant ‘governmentality’ that com-
prises a collective and institutionalized way of knowing how to govern
and be governed. The notion of governmentality is developed through a
genealogical study of how government, in the sense of the ‘conduct of
conduct’, has been problematized in modern society and how this
problematization has formed the basis of the formation and institution-
alization of an art of government that circumscribes and conditions
concrete acts of government.
The art of government is founded on particular discursive constructions
of the subjects, objects, means and telos of government, and the discur-
sive constructions are invested in particular institutions that establish
and distribute particular roles, identities, rules, norms, calculations,
conceptions and imaginaries (Dean 1999). The institutions of govern-
ment are not a result of conscious choices on the part of the actors, but
rather the result of the sedimentation of discursive strategies that are
taken granted and recursively validated in the course of action (Laclau

1990: 33–35).
The institutions of governance are vehicles for the exercise of power,
but power is not analyzed in terms of repression, prohibition and taboo,
but rather conceived as a field of crisscrossing discursive strategies that
produce and regulate particular subjectivities. Previously the citizens of
the feudal state were subjected to a sovereign power based on centralized
authority, laws, and physical punishment, but in modern society there
was an increasing reliance on the deployment of disciplinary techniques
aiming to create docile bodies and normalizing practices aiming to
shape and regulate the soul by means of defining what is normal and
what is pathological. Recently, the increasing emphasis on the govern-
ment of the population has taken the form of subtle attempts to
mobilize and shape the freedom of individual actors and different target
groups on the basis of a particular institutionalized understanding of
how this freedom should be exercised and what its exercise should
accomplish (Foucault 1991a: 103). Power is exercised through the
construction and regulation of governance networks.
In line with the historical institutionalist approach, poststructuralist
institutionalism tends to view the institutions of governance as the
medium and outcome of political power strategies. However, poststruc-
turalist institutionalism claims that political strategies are ‘intentional,
but non-subjective’ in the sense that they always aim to achieve or
accomplish something, and yet cannot be seen as the result of the
choice, decision, or strategic manipulation of an individual subject or
collective agency (Foucault 1990: 94f.). This deliberate decentring of the
subject clearly takes us beyond the actor-centred theory of historical
institutionalism.
Poststructuralist institutionalism rejects objectivist explanations of
governance networks as a functional response to the gradual evolution
of the inner logics of modern society. It also rejects the alternative expla-
nation of governance networks as the result of the rational decisions of
interdependent actors. Governance networks are to be explained in
terms of the conditions of possibilities that are provided by the current
governmentality that aims to govern at a distance through the mobi-
lization and regulation of self-regulating individuals, organizations and
networks. The present governmentality is developed in response to the
problematization of both state rule and market regulation. The new
governmentality program advances a global strategy for interactive
governance that simultaneously supports, and is supported by, local
strategies and experiences. Both the global and the local strategies of
governance, and the governance networks they engender, are stabilized

by institutional forms that are conceived in terms of ‘regimes of prac-
tices’, defining what can be said and what can be done, which rules to
impose and which reasons to give and what to take for granted in the
interaction between a multiplicity of actors (Foucault 1991b: 75).
Foucault (1980: 194) sometimes invokes the notion of dispositif, or
apparatus, as a name for the relatively institutionalized discursive strate-
gies that facilitate the contingent articulation of dissimilar subjects and
objects within a relatively unified grid, assemblage, or network.
In relation to the functioning of governance networks, poststructuralist
institutionalism draws attention to the productive effects of the power
strategies invested in particular institutions. Institutions contribute to
the construction of particular subjectivities. The social and political
actors are not naturally inclined to interact in and through governance
networks. Neither are they automatically endowed with the capacities,
sentiments and self-perceptions that are required for governance net-
work to function properly. The actors involved in network governance
must be constructed in and through institutionalized power strategies
that hail them as active, responsible, self-regulating, and even demo-
cratic, actors. The institutionalized power strategies also play a crucial
role for the functioning of the governance network as a whole. They will
often help to construct a particular narrative that defines the overall
purpose and mission of the governance networks and offers more or less
technical ways of defining the problems at hand and selecting the
feasible options (see Hajer 1993, 1995). The institutionalized narrative
creates an operational closure by including and excluding various issues,
particular forms of knowledge and certain actors.
Poststructuralist institutionalism offers two different ways of under-
standing the institutional dynamics of governance networks. The first
emphasizes the so-called tactical polyvalence of the institutionalized
discourses of governance. The tactical function of discourse is neither
uniform nor stable, and the discursive elements can come into play in
various and opposed strategies (Foucault 1976: 100f.). In other words,
the ultimate undecidability of social meanings and identities permits
gradual changes of the institutional framework of action through con-
stant displacements, modifications and resistances (Laclau 1990: 26 ff.).
The other way to understand institutional transformation is through the
notion of problematization and dislocation (Laclau 1990: 39 ff.). New
events that cannot be integrated, domesticated, or otherwise accounted
for by the relatively institutionalized discursive system will tend to
reveal its limits and might even cause its partial breakdown. This will
open a strategic terrain for hegemonic struggles aiming to construct a

new order by providing a new reading principle that offers a particular
understanding of the crisis and the possible ways to solve it. As such,
gradual transformation or ruptural change can both be explained by the
theories informing poststructuralist institutionalism.
Similarities and differences
The main difference between social constructivist institutionalism and

poststructuralist institutionalism is that the latter tends to see institu-
tions as an integral part of crisscrossing power struggles, whereas the
former conceives institutions as a means for the integration of frag-
mented actors and organizations. There is also an important analytical
difference between the emphasis on the individual interpretation of
the institutional rules in social constructivist institutionalism and the
emphasis on the discursive construction of subjectivities in poststruc-
turalist institutionalism. However, both the social constructivist and the
poststructuralist institutionalism highlight the role of norm-based
action and the institutional shaping of the identities of the network
actors. As such they reject the rationalist theory of action that underlies
both rational choice institutionalism and historical institutionalism.
The gulf between the rationalist and the anti-rationalist strands of
institutionalist theory is wide and impossible to bridge. However, we
should remember that what really divides the four theories is not the
question of whether or not action is rational. Both March and Olsen
(1989, 1995) and Foucault (1976, 1991b) will claim that action can be
informed by a certain rationality, but they will insists that there are
many different rationalities and that these are contingently constructed
within particular institutional contexts.
Historical institutionalism is sometimes portrayed as sort of compro-
mise between rational choice institutionalism and social constructivist, or
normative, institutionalism (Hall & Taylor 1996). However, the emphasis
of interest-based calculations seems to bring historical institutionalism
closer to rational choice institutionalism than to the culture-orientated
institutionalisms that tend to emphasize the institutional and discursive
conditions for the formation of the actors identities and preferences.
Truly, there are attempts to insist on the role of learning and the impact
of norms, values and policy paradigms (Hall 1993; Campbell 1997). But
the role assigned to institutions in these arguments can still be analysed
within a rationalist perspective as a matter of constraints on the range of
feasible options. Hence, historical institutionalism shares with rational
choice institutionalism the emphasis on the transformative capacities of

social and political actors. Only, historical institutionalism tends to
think of political actors in terms of collectively organized actors and
tends to relax the assumption of the pre-given interests of the actors
much more than most rational choice institutionalists are prepared to
do. Finally, historical institutionalists insist that institutional con-
straints are often very difficult to remove in the short run and therefore
tend to preserve status quo, or only permit incremental changes.
The structure of Part I
The chapters in Part I will further scrutinize the role of institutions for
the formation, functioning and transformation of governance networks.
Drawing on insights from the theories associated with the new institu-
tionalism the three chapters will address crucial question about the
necessary and sufficient conditions for the formation of governance
networks; the development of virtuous and vicious circles in the func-
tioning of governance networks; and finally the factors determining the
endurance or change of governance networks over time.
2
Mechanisms of Governance
Network Formation – a
Contextual Rational
Choice Perspective
Nils Hertting
Introduction*
Where do governance networks come from? How do they emerge and

why do they arise? Until recently these issues have not been prioritized
in the network governance literature. Focus has been on the outcomes
and political consequences of governance networks. Still, it seems appro-
priate to claim there is a lack of elaborated theoretical arguments about
governance network formation (Hay & Richards 2000). If, however, gov-
ernance networks are ‘here to stay’ and network governance should be
‘taken seriously’, we need a better understanding of the mechanisms of
governance network formation. If the ‘second generation of network
governance research’ specifically address meta-governance strategies
(cf. Sørensen & Torfing in the Introduction), a crucial element should be
systematic attempts to answer the questions above. It is the purpose of
this chapter to develop some theoretically informed arguments about the
formation and institutionalization of governance networks from what
I call a contextual and ‘thin’ rational choice perspective.
According to a historical account, network modes of coordination
evolve as a response to functional and organizational fragmentation and
differentiation in public policy (Rhodes 1997a). First, the post-war strat-
egy of implementing welfare policies through functionally specialized
organizations and welfare sectors caused ‘over-organization’ (Hjern and
Hull 1982). Secondly, NPM-strategies, including ‘agentification’ of
public authorities and ‘marketization’ of the public sector increased
43
fragmentation even more (Rhodes 1997a; Pierre & Peters 2000). The
historical account claims that these processes not only undermine
hierarchical coordination but also create a need for new modes of gover-
nance to secure coordination, resource mobilization and problem solving.
In other words, lack of coordination on the societal level, generates new
governance arrangements in order to make coordination more efficient.
According to this approach – Bevir and Rhodes call it ‘positivist’ in
Chapter 4 in this volume – the functioning of governance networks as
institutional mechanisms for coordination and resource mobilization
explains why they exist. Such a functionalistic account of governance net-
work formation makes explicit questions about the micro-mechanisms of
governance network formation irrelevant; the functional consequences
have been reason enough (Elster 1989). Consequently, the policy network
concept often refers to structural relations of interdependencies within a
policy field as well as to the informal institutions for coordination that
may evolve in such a context at the same time.1
It is the merit of such grand functionalistic theories to sum up and
condense historical trends. However, in order to produce more detailed
ideas and insights about crucial mechanisms for governance network
formation such an approach is not adequate. From a functionalistic con-
ceptualization it seems that differentiation, fragmentation and interde-
pendencies are sufficient mechanisms to bring about some kind of
network mode of coordination. They are not. Such an account obscures
the role of more or less strategic real-world actors in governance network
formation and, as a consequence, the potential cooperation problems
and dilemmas of organizing governance networks (Blom-Hansen 1997;
Hay & Richards 1998).2
To better understand how and why governance networks are formed
and how and why attempts to form such networks sometimes fail it seems
reasonable, I argue, to ask what conditions are crucial for the formation of
governance networks from the perspectives of interdependent actors.
Even though ‘meta-governance’ strategies might help the development
of governance networks (Sørensen & Torfing, Chapters 1 and 9 in this
volume), governance networks are appealing as institutions capable of
making and implementing legitimate policies in the absence of top down
authority. Hence, it is of fundamental theoretical interest to ask under
what conditions we should expect governance networks to evolve from
purposeful horizontal interactions among real-world actors (Scharpf
1994: 43). That is, when and how should we expect networks, capable of
producing efficient and legitimate policies and implementation, to evolve
endogenously among actors without external meta-governors?
Mechanisms of Governance Network Formation 45
Interpretation, rational choice and mechanisms
In this chapter I adopt a non-formal, context-oriented and ‘thin’

rational choice perspective to address these questions. The basic idea is
that governance networks are products of interactions among more or
less rational actors that invest in institutional arrangements to improve
their capacity to implement various policy ideas. It is ‘context-oriented’
because I claim rational choice is a method for interpreting how politi-
cal actors understand and give meaning to their actions in specific con-
texts or situations. It is ‘thin’ (Ferejohn 1991) since the idea of
rationality is deliberately defined in a broad way.
The rational choice approach has been questioned and appears
controversial in governance networks studies (Rhodes 1997a: 174f.; Toke
2000). While rational choice analysis is often applied to stable and
highly institutionalized political settings, governance networks often
evolve in some kind of institutional vacuum between formal organiza-
tions. In such dynamic situations, uncertainty about the ‘rules of the
game’, the set of players and their action alternatives and outcomes
seems to limit the value of rational choice (cf. O’Toole 1993). That is, the
nature of governance networks does not fit nicely with mainstream
rational choice analysis. Sharing the argument for taking real world
actors seriously and understanding governance network formation
endogenously and ‘bottom up’, Bevir and Rhodes therefore argue for a
‘decentred’ approach and ‘to focus the social construction of policy net-
works through the ability of individuals to create meaning’ (Bevir &
Rhodes, Chapter 4 in this volume).
Following scholars like James D. Johnson (1990), John Ferejohn
(1991), Bates et al. (1998), Bo Bengtsson (1998) and most recently Colin
Hay (2005), however, I am arguing that rational choice analysis is a
method for understanding the meaning that actors give certain actions
in specific situations. Such a rational choice approach might be seen as
a framework for reconstructive and interpretive studies.3 The idea that
actors act rationally is a ‘methodological commitment’ (Popper 1965:
362), not a hypothesis about individuals and their behavior. As a
methodological device, the concept of rationality should be broad,
almost empty (cf. Ostrom 1990).
According to such a ‘thin’ concept of rationality, actors act on the
basis of a perceived rather than an objective reality (Elster 1989: 31;
Ferejohn 1991: 282). The preferences of rational actors may reflect sub-
jectively defined interests and cultural or normative orientations. What
is a rational action, therefore, cannot be understood without analyzing
the meaning that actors attribute to a specific action in a specific social,

political or administrative context (Scharpf 1997: 19f.). Such a rational
choice method modestly assumes that actors promote governance net-
works because they believe such institutional arrangements will help
them accomplish some kind of goal or meaning. This is not to make the
public choice assumption that actors always try to maximize their self-
interest, regardless of norms and rules that go with their position. On
the contrary, given the importance of organizations, professions and
other more or less institutionalized actors it seems reasonable to expect
that network formation strategies are very much based on the institu-
tional and normative frameworks that define the roles of individual
actors, such as social workers, school principals, national bank presi-
dents, ministry of environment civil servants etc. Hence, rational choice
analysis requires cultural knowledge and political anthropology (Bates
et al. 1998: 628). The point is that rational choice analysis provides us
with a tool for investigating ‘what rational can mean’ for actors in dif-
ferent situations and contexts ( Johnson 1990: 118).
Theories are tools and one tool cannot fit every social science research
endeavor. There should be nothing controversial about that. To stress the
interpretive nature of rational choice analysis is not to claim that a con-
text-oriented and ‘thin’ rational choice perspective is always appropriate.
Rational choice makes a virtue of being ‘anti-holistic’. It reduces the inter-
nal complexity of political actors in order to understand outside political
situations and contexts from the perspective of ‘thinly’ and context-
bound rational actors. Understanding individual processes of learning
and cognition requires different analytical devices (cf. Tsebelis 1990: 39f.).
Mechanisms: incentive and opportunity structures

The ambition in the rest of this chapter is to discuss three types of
mechanisms of governance network formation by adopting such a
context-oriented and ‘thin’ rational choice approach. Following Hedström
& Swedberg (1996), a mechanism oriented analysis is characterized by
three core features. First, a mechanism oriented analysis strives to estab-
lish immediate and close links between input and output elements of
analytical entities. Second, the mechanism orientation includes a focus
on relatively specific phenomena; a mechanism oriented analysis is
limited in its scope. Third, the mechanism approach is based on the
principle of methodological individualism, that is, the idea that actors with
meanings and intentions produce those processes and changes we try to
understand in social science, not casual laws beyond the reach or under-
standing of social actors (Coleman 1990: 8).4
From this third principle it logically follows that mechanisms are not
decisive or deterministic conditions. Instead, I suggest, a mechanism
should be defined and understood as an incentive and opportunity struc-
ture.5 That is, when a mechanism is working in a specific social situation
the actors perceive they have the opportunity and the incentive to act in
a certain way. Again, such incentives need not necessarily be economic
or related to self-interests. The more specific arguments about network
formation mechanisms are organized around three related questions:
1. What are the contextual incentives for network formation?

2. What strategic calculations and choices should we expect of the single
interdependent policy actor?
3. And, finally, what interactions or games should we expect to support or
constrain the formation and institutionalization of governance net-
works?
In the final analysis I discuss the contribution a contextual and ‘thin’

rational choice perspective can make to the second generation of
governance network studies (Torfing & Sørensen, Introduction in this
volume) and our understanding of governance networks formation.
Contextual mechanism: perceived

interdependencies
The concept of interdependency is crucial to theories about governance

networks. ‘Networks develop and exist because of the interdependency
between actors’ (Klijn 1997: 31). But what is interdependency and what
can it mean to real-world actors?
Resource and strategic dimensions of interdependencies

In order to understand how interdependencies are interpreted as an
incentive structure for governance network formation, we should high-
light two aspects of how actors perceive such dependency relations.
Analytically, a distinction between resource dependencies and dependency
relations that produce strategic externalities should be useful (cf. Hertting
2003: 61).
In the network literature, the concept of resources and resource
exchanges has been crucial. According to inter-organization theory, all
organizations are dependent on certain fundamental and critical
resources. Functionally specialized organizations need to exchange
resources with others in order to achieve their goals: Actor A is dependent
on Actor B if A, in order to bring about a preferred action, needs a

resource that B controls. If B at the same time need a resource that A
controls, A and B are interdependent (Scharpf 1978; Benson 1982). Such
dependencies are managed by resource exchanges and actors may invest
in trustful network relations in order to make such exchanges more effi-
cient. Actors in governance networks are organized by a joint interest in
securing critical resources in a policy system (Klijn 1997: 31).
Strategic aspects of interdependencies, in contrast, are not managed
through material or financial resources exchange. The actors have the
necessary resources to implement their actions, but the outcome is
dependent on the actions of other actors. If actor A’s action is producing
outcomes in interaction with actions taken by actors B, C and D, then
A’s capacity to implement its own policies is dependent on the strategic
choices of B, C and D. We may claim that such interdependencies are
found on the outcome rather than the action level.
To talk about information as a resource is not sufficient reason for
incorporating this aspect of interdependencies. Information exchange is
necessary but not always sufficient. Such dependency relations require
control exchange (Coleman 1987) to be successfully managed. When the
outcomes of the strategies of a set of actors interfere with each other,
every single actor in the set is motivated to transfer a part of its right to
autonomously choose and design its strategies to the rest of the set of
actors, in exchange for a similar right to control their choice of actions.
By explicating this aspect of dependencies, we are stressing the fact that
perceived conflicts may be motives for a set of actors to develop
governance networks (cf. Stoker 1991: 49f .).
Two points have been made. First, the experience of interdependency is
equivalent with that of lack of implementation capacity. Interdependent
actors cannot implement their goals on their own (Hanf & O’Toole 1992).
Second, we need to look beyond resources and include strategic external-
ities in order to understand how actors may perceive interdependencies
and why they will try to develop governance networks. The distinction
between resource and strategic dependencies is analytical. In real world
governance, these are two related aspects of dependencies. However, as
modern welfare societies have increasingly become products of political
projects (cf. Wildavsky 1979) we should expect interdependencies to
increasingly include this strategic dimension.
Mutually perceived, mutually recognized and

changing interdependencies
It is perfectly compatible with a ‘thin’ rational choice approach that
experiences of mutual dependencies need not necessarily be objectively
true. It is perceptions together with desires that stimulate actors to build

institutional arrangements, for tomorrow including networks. As a
consequence of its reductionistic character, the rationalistic perspective
cannot explain how these perceptions come into existence at the
individual level (cf. Wildavsky 1991). Rational choice analysts need to
understand perceptions but not to understand their origin. If we are
to understand the complete chain of events behind network formation,
however, we should also strive to understand the power struggles
between actors to define and re-redefine the policy issue and construct
the dominant discourses of specific networks (cf. Toke 2000; Hajer
2003).6
It seems reasonable to argue that network formation is easier when
the actors share a mutual understanding and shared perception of the
policy problem (Zafonte & Sabatier 1998). From what has been said
above, however, a completely shared understanding of the cause-and-
effect relations should not be necessary to the emergence of governance
networks. The actors may recognize and agree about the existence of
interdependencies without sharing specific beliefs about the issue at
hand. This is one of the characteristics of issue networks as opposed to
policy communities (Rhodes & Marsh 1992).
Even though a completely shared understanding is not necessary,
dynamics and changing perceptions will probably create uncertainty
among actors. Empirically, such uncertainty induced by dynamic
perceptions may threaten the formation of governance networks (see
below); analytically, it may reduce the relevance of rational choice
(O’Toole 1993). Nevertheless, if political actors themselves are able to
deliberately form governance networks on the basis of a collective
understanding of their relations, anthropologically oriented rational
choice analysts should also be able to grasp them and, hence, make
relevant interpretations (Scharpf 1997: 40f., Bates et al. 1998). Hence, to
become a fruitful method, context-oriented rational choice rest on some
basic ideas of social constructivism.
Symmetrical and asymmetrical interdependencies

The consequence of structural interdependencies is that actors
experience a lack of implementation capability. Having said this, it is
essential to note that the importance of such dependencies is not always
equally distributed. Interdependencies may be more or less asymmetrical.
In a time when the complex and interdependent nature of society is
emphasized, it seems appropriate to stress that a dependency relation
between a set of actors can be described as mutual while the relative
importance of the issue concerned can be asymmetrical (Hernes 1975;
Scharpf 1978).7 Therefore it is important to relate dependencies among

actors to their evaluation of the specific problem or issue concerned. It is
possible to find situations where all actors are aware of strong mutual
dependencies, while the intensity of their motives for handling the situa-
tion through some kind of coordination still differs slightly. As we will see
below, even small asymmetries may cause big problems when it comes to
institutionalizing governance networks among limited rational actors.
Actor calculation mechanism: preference

for informal networks
Formally interdependent actors can be independent, but functionally they

lack autonomy. They need some kind of exchange to be able to imple-
ment their goals. Yet, this is the very stuff of bargaining and negotiations
(Elster 1990); this is not sufficient to an understanding of why and when
interdependent actors strive for governance networks. Why should they?
The need for institutions

The rationale of the negotiator is to find better strategies through
exchange and cooperation with others without giving up her autonomy.
The dilemma of the negotiator, however, is that the search for more
efficient collective strategies may conflict with the distribution of the
value of these strategies (Scharpf 1997: 137ff.). Innovative strategies
follow from creativity, deliberation, open communication and mutual
trust. At the same time it may very well be a rational strategy for a single
actor to hide or lie about its beliefs and preferences in order to create a bet-
ter starting point for negotiations. That is, negotiations have an intrinsic
cooperation problem: if both actor A and B strategically manipulate the
information about their preferences and capabilities this will lead to
imperfect negotiations and lack of coordination in the policy process.
Having said this, it seems reasonable to expect that actors that repeatedly
find themselves in this dilemma will value highly some kind of institu-
tional framework (cf. Williamson 1996). If the actors perceive interde-
pendencies as quite stable and find a repeated need for negotiations we
should expect them to invest in institutions that make the single
negotiation or deliberative activity more efficient. Hence, efforts to
establish governance networks may be understood as strategies for
establishing an institutional framework for more efficient negotiations.
Hence, network formation is institutional design; that is, investments
into the future in order to compensate for limited rationality and oppor-
tunistic behavior and to secure action capacities (cf. Tsebelis 1990).
The strategy of informality and generalized trust

Institutions need not be formal. Governance networks are not. But why
would (thinly) rational and interdependent actors prefer informal
modes of coordination based on trust? One answer is that perceived
interdependencies make ‘contingent consent’ (Levi 1990) possible. If
interdependent actors believe they will need efficient channels for
resources and control exchanges in the future they can make their
cooperative strategies contingent on cooperative responses. Hence, gov-
ernance networks and generalized trust could be seen as the outcome of
such contingent cooperation.8
The informal character of governance networks is not only compatible
with but also an inevitable outcome of this strategy. Governance net-
works are preferred to more formal institutional arrangements because
they make a ‘cheap’ exit strategy possible. In informal institutions actors
may easily withdraw their institutional consent if others do not ‘deliver’.
This strategy of informality and generalized trust is somewhat
paradoxical given the focus on interdependencies in the network litera-
ture. If the exit possibility is the reason why collective actors prefer
network-like governance institutions, then we would expect stable
governance networks to evolve among actors that have at least some
capacity to implement their goals on their own or have other potential
network partners to cooperate with (cf. Blumberg 2001). A governance net-
work seems to be a stable institutional outcome among not completely
interdependent actors. The value of informality presupposes a potential
to act alone or together with other partners.
The game mechanism: the problem of

continuous cooperation
Governance networks can be seen as the product of institutional strategies

among a set of interdependent actors in order to solve problems
with communication, information, opportunism and transactions in
complex negotiations. It is, then, not sufficient to outline the network
formation rationale for the single actors. From an actor perspective
governance networks as institutions might be regarded as a collective
good (cf. Kickert & Koppenjan 1997; Carlsson 2000). In the social
sciences it is well known that individual preferences for such collective
goods are not sufficient for their provision. In that respect, governance
networks are just like organizations. Governance networks are
institutional outcomes on an aggregated or collective level.
Governance network as continuous cooperation

What makes governance networks special, compared to other institutional
mechanisms, is the continuous need for cooperation among actors.
While cooperation may take place within a hierarchical context, a typi-
cal hierarchy coordinates through top down directives; and while the
typical market coordinates through the aggregative outcome of a large
number of autonomous and anonymous spot-contracts, we also know
that negotiations and constellations may take place within market con-
texts. Network governance, however, needs cooperation to become a
working mechanism of coordination (Zintl 1993; Scharpf 1997: 47).
Consequently, we may conclude that formation should not be concep-
tualized as a distinct and temporary phase in a sequence of network
governance. A governance network is never formed, established or institu-
tionalized once and for all. To work as an institution for coordination,
interdependent actors need to continuously constitute the network.9 They
have to demonstrate their recognition of and identification with the net-
work while they ‘use’ it in on the levels of policy production and operative
actions (cf. Ostrom 1990: 53f and O’Toole, Chapter 12 in this volume).
Below I will discuss three specific cooperation problems on the
constitutional or formation level of governance networks identified within
game theory. Referring to the previous discussion on interpretive rational
choice, a ‘game’ is neither a dependent, nor an independent variable.
It is an ideal type used to understand what it means to be rational in
different game-like situations. Hence, the interactions are interpreted as
reflections of how actors perceive and construct the governance process.
In order to do so, in empirical research we need information about
the set of actors that are involved in handling interdependencies, their
perceived implementation capacity and available strategies, and their
evaluation of expected institutional outcomes as products of different
combinations of available strategies in the situation (Hermansson 1990,
Ostrom 1990, Scharpf 1997).
Here, three out of many formally defined games are highlighted as
tools for understanding why the formation of a governance network
may be troublesome or even fail although every actor would gain from
such an institutional framework.
The free riding problem

The first cooperation problem to be discussed is ‘free riding’ (cf. Olson
1971). This is the situation where everybody would gain from coopera-
tion, but each actor would gain even more if everybody but he or she is
cooperating. Game theoretically, this situation is typically defined as a
Prisoner’s Dilemma. In such a situation not to cooperate is a ‘better’

choice, regardless of what choices the rest of the actors make. Within the
vocabulary of game theory, not to cooperate is a dominant strategy
(Elster 1989, Hermansson 1990).
The free rider problem in network governance is a strategic situation
where every actor will gain from the existence of effective governance
networks. For each and every actor, however, the best situation is when
everybody else is engaged in establishing and sustaining of such a net-
work institution while he or she is enjoying the fruits of the network
without contributing. If every actor follows this strategy there will
always be a lack of efficient governance networks in the policy process.
And since non-cooperation is a dominant strategy in Prisoner Dilemma
situations this will always be the case.10
But is this a relevant analytic model? Is it possible to gain from the
fruits of governance networks without supporting them? To some extent
it seems so. A free riding strategy in governance network formation
might be to communicate false and misleading information on own per-
ceptions, preferences and potential actions in the process of constructing
a common knowledge about interdependencies. This is an attempt to
exploit the dependency and trust of others in order to improve one’s own
position. If this is the game real actors play we have a situation where
perfect network governance would be a collective good never realized.
But there are also less flagrant forms of free riding: actors may not
seriously invest time and energy in collecting and disseminating infor-
mation on their positions; within organizations, higher level principals
may send networking agents without resources, knowledge and author-
ity to run the contacts with the rest of the network. ‘Conflict avoidance’
might also be interpreted as a free riding strategy (Hertting 2003: 326f.).
The Prisoners Dilemma is known for its cynical view on the possibilities
of cooperation; and rightly so. The model configures a situation where
the rationality of the individual leads to irrationality for the collective.
The problem is not lack of information about the perceptions, possibili-
ties and preferences of others. No matter how they act, non-cooperation
is always best for the individual and mutual non-cooperation is the only
equilibrium. Hence, it is not only hard to establish cooperation; it will
also be difficult to maintain it if, by some means, it is established (Elster
1989: 105; Hermansson 1990).
The assurance problem

The problem with the Prisoner Dilemma model is not the cynicism as
such, however. The problem is that this stylized game logic has become
the model for cooperation problems. But every cooperation problem is

not a Prisoners Dilemma. There are also less obvious problems identified
within game theory.
What is known as the Assurance Game describes a situation logic that
threats cooperation despite a clear joint interest in mutual cooperation.
In such a game each actor would get the highest pay off if he or she and
everybody else did cooperate. Thus, if cooperation begins there is really
no reason for anyone to end it. Mutual cooperation is equilibrium. If,
however, one of the actors for some reason does not cooperate, it is also
rational for the rest not to do so. The worst outcome is to be the lonely
co-operator, ‘the sucker’, who genuinely tries to establish and sustain a
network while the rest of actors are not. If all actors try to avoid this
position and choose not to cooperate, there will be no governance
network and the actors will end up in the second-worst outcome.11
As an analytic tool, the Assurance Game centers our attention on the
strategic consequences of uncertainty in what seems to be an idyllic
situation. The game highlights the problem of misinterpretations and
lack of information in network formation. In the framework of ‘thin’
rational choice this is not an inconsistency.12 If limited rational actors
lack reliable information about each other’s preferences they have
reason not to cooperate in the Assurance Game. That is, although every
single actor believes that everyone would gain most from network
governance, it is rational for an actor not to cooperate if she doubts that
the other actors make the same interpretation.
That is, while complex interdependencies trigger preferences for
network formation on one level, complexity may prevent the realization
of such networks on another. The Assurance Game seems to grasp the logic
of the ‘structural dilemma’ in network governance (Börzel 1998).
According to Börzel this dilemma evolves from the ‘insecurity caused by
intra-organizational control and the need for intra-organizational imple-
mentation of inter-organizational compromises’ (1998: 261). The problem
is not disagreement on the need for governance institutions among the
networking actors, but the lack of assurances that these network prefer-
ences will be turned into efficient network strategies, a lack due to internal
interactions within participating organizations, professions, coalitions etc.
The uncertainty is a rationale for non-cooperation. (Below I will come back
to the issue of collective actors in network governance formation.)
The generosity problem

Finally, the game known as the Battle of the Sexes tells the story about
the husband and wife who want to spend the night together. However,
she prefers the boxing game, while he prefers the ballet. The two parties
have a common interest in coordinating so that they will reach one of
two superior outcomes: spending the night together at the ballet house
or the boxing stadium, rather than spending the night alone.13
The game illustrates a situation with two different cooperation
equilibriums and a conflict between actors over which is preferable. In this
situation it is not sufficient that the actors are allowed to communicate
and enter into agreements with each other. At least not if the difference
between getting the most and the second most preferred outcome is
important to both actors. And this is the most important insight from
the game: communication and negotiations are not enough for attain-
ing the common good, that is, a governance network. What is needed is
generosity. In order to reach stable cooperation, Actor A has to allow
Actor B to reach B’s most preferred outcome while A accepts a less
preferred one, or vice versa. Hence the Battle of the Sexes captures what
might be called the generosity problem with governance network
formation.14
Is such an analytic game model at all relevant to governance forma-
tion? I believe so. At the heart of the network governance rationale is a
specific idea about the value of informal relations and trust-based
coordination. A rational actor will give up just as much autonomy as is
necessary to establish a coordination institution that will make the
implementation of its own goals more efficient than would no coordi-
nation – and no more. This is not a very demanding assumption.
Nevertheless, it points out a delicate dilemma for interdependent actors:
that of balancing the need for more cooperation in order to improve the
capacity to act and the desire to maintain sovereignty. This dilemma
may trigger tricky games between interdependent actors with a shared
preference for some kind of coordination.
Even though network institutions are informal compared to organiza-
tions, the degree of informality and the arrangements for creating trust
may vary. This variation may be a most strategic concern to actors in
institutional design.15 The problem is no longer that some actors ruin
network governance by being tempted to take the free ride-strategy, nor
that the structural complexity requires an extraordinary amount of
assurance and solidarity between the actors. The actors agree that they
will all benefit from a network-like institution, but they disagree about
its specific form; its modus operandi.16
The problem, then, is to identify one out of many possible institu-
tional solutions when all of them will improve local coordination. The
irony of the game played is that the very multiplicity of more or less
informal arrangements may prevent any of them from being realized

(cf. Elster 1989). Above it was noted that asymmetries in interdepen-
dencies affect what is a rational institutional design. The value of infor-
mality depends on the potential to act alone or together with others.
Despite a common interest in coordinating actions with a minimum
loss of autonomy small variations in the specific institutional prefer-
ences may cause big problems in governance network formation. If real
world games have incentive structures similar to the Battle of the Sexes,
there will always be one partner who sees another potential network-like
solution that would increase her pay off even more. Therefore, we
should expect informal network governance not to attain a stable equi-
librium (Hertting 2003: 342ff.). Consequently, this game model helps us
to understand why there may be continuing cooperation among inter-
dependent actors despite a failure to institutionalize a governance net-
work that works efficiently as a coordination and resource mobilization
device.17
Collective actors and vertical games
So far we have explored the potential of rational choice assuming that the
interdependent actors can be treated as more or less unitary and homog-
enous. It was even argued that institutionalized roles, as complexity
reducing devices among real networking actors, is a methodological
presumption for rational choice analysis. Such an assumption is
sometimes, but not always, reasonable. This is an empirical issue. It
depends on how the actors of the game are conceived.
Often, however, it seems reasonable to expect that network partners
are perceived as multi-level actors. That is, interpretation, strategic
calculus and choices are supposed to be affected by decisions made on
different organizational levels. The networking actors, that is, represent
an organization that is an instrument for political, civil or private
assemblies or boards. Hence actors in network governance are not only
horizontally dependent on each other but also vertically dependent on
other levels. Actors in networks governance formation often have
‘twofold loyalties’ (see Häusler et al. 1993). Trust and agreements that
evolve in horizontal games between networking actors need to be
approved or at least accepted within formal organizations or among a
constituency. If efforts to develop horizontal relations are too successful,
this is the dilemma, they may lead to more suspicion and mistrust in the
vertical game within the interacting organizations. On the other hand,
internal disputes within the organizations or the constituency may
negatively affect the trustworthiness of local agents in the horizontal

games. This is the vertical game of collective actors in governance
network formation. To have honesty simultaneously in both horizontal
and vertical games of network formation, it seems, is more easily said
than done (Hertting 2003: 329).
Summary and conclusion
Where do governance networks come from? In this chapter some ideas

about the mechanisms of governance network formation have been
formulated from a contextual rational choice perspective. The starting
point was simple. Governance networks were supposed to be formed in
purposeful interactions among structurally interdependent actors to
improve their implementation capabilities. Formulated as an interpre-
tive and context-oriented methodology, it was argued that rational
choice seems to help us ask relevant questions and development
relevant ideas about the mechanisms of governance network formation.
This is not the first attempt to do so (O’Toole 1993; Scharpf 1994).
However the arguments, the structure of the arguments and the conclu-
sions presented here differ. Following the well-known macro-micro-macro
schedule of James Coleman (1990), the mechanisms formulated,
examined and elaborated are presented in the Figure 2.1.
To single actors, the experience of being interdependent is equivalent
to experiencing lack of implementation capacity. Two different
dimensions of interdependencies were emphasized: resource dependen-
cies handled through resource exchanges and strategic externalities
handled through control exchange. The distinction is analytical. As such
Structural Governance network

interdependencies
Policy actors Network gaming

experience lack of
implementation
capability
Preference for Network design

coordination strategy
Calculations for
informal institutions
in order to improve
negotiations
Figure 2.1 Perceptions, calculations and games in governance network formation

it emphasizes the fact that mutual dependencies may be more or less

important to actors. The degree of asymmetry is crucial. Furthermore, the
experience of a mutual understanding of dependency relations is not
necessary. A mutual recognition is sufficient.
Second, an internal calculus for informal network modes of coordination
was formulated. Formally independent but functionally interdependent
actors prefer, cetaris paribus, informal and trust-based networks to more
formal modes of coordination. Informal networks make it possible to
manage interdependencies more efficiently without surrendering for-
mal autonomy while remaining able to leave the institution whenever
that becomes desirable. If the exit possibility is why collective actors
prefer network-like institutions, however, we should expect stable
governance networks to evolve among actors that have at least some
capacity to implement their goals on their own, or have other potential
network partners to cooperate with. Hence there seems to be a limit to
the degree of dependency as an incentive for network formation. Actors
that perceive a highly asymmetrical interdependence will probably
prefer more formalized forms of coordination.
Compared to hierarchies and markets, governance networks need
continuous cooperation to work as an institutional arrangement for pol-
icy production. Hence there is no distinct formation-phase of networks.
They need to be established and confirmed through cooperative actions
and attitudes all the way. From this observation, three different cooper-
ation problems where formulated through game theory: the free-rider
problem, the assurance problem and the generosity problem. Rational
choices of many individual actors may very well end up as irrational
outcomes for the collective. Therefore, governance network formation
may fail among a set of rational actors that would all gain from it. This
is not necessarily due to irrationality. Furthermore, game theory teaches
us that there is a threat to governance network formation even in idyl-
lic situations like the Assurance Game. Furthermore, the analysis of the
‘generosity problem’ defined by the Battle of the Sexes suggests that small
details may have major consequences for the formation and maintenance
of governance networks. Even small differences with regard to the pre-
ferred institutional arrangements, reflecting slightly different percep-
tions of interdependencies, may turn governance network formation
into a frustrating endeavor, and make governance network formation
failure a regular experience.
In the final ‘game’ analysis the prospects for straightforward or
even successful network formation may seem disappointing. At least it
should be clear that, from a rationalistic perspective, shared recognition
of interdependencies as well as a shared preference for governance net-

works is not sufficient for the formation of governance networks.
Somewhat paradoxically, to analytically treat actors as rational is perhaps
most productive when governance network formation fails although
interdependent actors agree about the need for it (cf. Miller 2000).
This is not to say that governance networks are not here to stay, or
should not be taken seriously. However, in order to understand meta-
governance and the prospects of meta-governance strategies, we need
analytic tools that reveal rather than obscure the opportunities and
incentives to form governance networks among real world actors.
Notes
* An earlier version of this chapter was presented at the workshop on ‘Democratic
Network Governance: Theoretical Puzzles’ in Roskilde, Denmark, 28–29 April
2005. I would like to thank the workshop participants, including Allan Dreyer
Hansen, Laurence O’Toole, Hans-Erik Klijn and the editors Eva Sørensen and
Jacob Torfing, for helpful comments. I would also like to thank Bo Bengtsson
and Evert Vedung for earlier discussions on the arguments of this chapter.
1. The definition of O’Toole et al. (1997) makes this very explicit: ‘An imple-
mentation network, like any other policy network, is the pattern of linkages
traced between organizational actors who are in some way interdependent.
It is also a socially constructed vehicle for purposive action. Like organizations
themselves, implementation networks are intended to be used as instruments
for mobilizing the energies and efforts of individual actors to deal with the
problem at hand’ (139, my italics). See also Pierson (2000b) for a discussion on
functionalistic reasoning on institutional design within political science.
2. This seems also to be true for the literature on ‘network management’ (Kickert
et al. 1997). These theorists seem more concerned with identifying what
instruments might be used, rather than what strategic incitements may guide
managers and meta-governors in different network situations.
3. On this issue I am particularly indebted to influences from and discussions
with Bo Bengtsson. For a more elaborated argument, see Hertting 2003, ch. 4
(in Swedish). See also O’Toole, Chapter 12 in this volume. O’Toole argues for
a ‘heuristic use’ of game theory.
4. It should be noted that the principle of methodological individualism has a
more pragmatic and policy relevant dimension. If the principle is adapted the
analysis will focus on the level where possible interventions need to be imple-
mented, that is, by actors. This is also to say that scientific interpretations
about phenomena at the system level which refer to the incitements at the
actor level will become more useful and policy relevant than a statistical
explanation where only aggregated data is used.
5. See Elster (1990), Bengtsson (1999) and Hertting (2003) for more elaborated
discussions on the mechanisms and rationality.
6. Interdependency relations in homelessness policy, for instance, are interpreted
very differently if we think that the heart of the problem is the homeless
individuals and their way of life, or the existing housing stock and the func-
tioning of the housing market. From the former perspective (or narrative),
actors dealing with social services, mental illness, drug abuse, crime and
health care are seen as crucial. From a more structural oriented perspective on
the causes of homelessness, housing companies, planning authorities,
housing investment banks etc become most relevant.
7. It should be obvious that there is no need to talk about interdependencies if
they are completely asymmetrical (cf. Scharpf 1978). The point should rather
be understood from the opposite perspective: in order to understand
processes of governance network formation it might be crucial to understand
how important the relevant issue at stake is.
8. This idea is also similar to that of ‘repeated games’ as explanation for the
emergence of cooperative norms (Axelrod 1986).
9. In terms of the institutional grammar of Crawford & Ostrom (1995), gover-
nance networks are neither rules nor norms, but ‘strategies in equilibrium’.
10. With pay off-values of the game matrix this is to say that the actors do not
achieve what is best for the collective, i.e. mutual cooperation (3,3), or what
is best for the individual, i.e. everybody but oneself cooperates (1,4 or 4,1).
11. That is, the preference ordering for every actor is: (1) mutual cooperation
(governance network), (2) everybody but oneself cooperates (a governance
network is established but ego is not a core participant), (3) no-one cooper-
ates (no network), (4) oneself cooperates but no-one else does (no network).
12. Quite the opposite: it exemplifies the value of game theory in analysing
interactions among limited rational actors.
13. See Hermansson 2003:140f., and Scharpf 1997:74f. for presentations and
analysis of the Battle of the Sexes game.
14. The Battle of the Sexes logic as a generosity problem is formulated by
Hermansson (1990, 2003).
15. Cf. Miller’s analysis of dilemmas in political hierarchies (2000:539). For a
more general discussion on the politics of institutional change, see Tsebelis
1990, chapter 4.
16. See also Schneider 1993:263ff. for a similar point.
17. In Hertting (2003) I have interpreted the phenomenon of ‘repeated coopera-
tion in frustration’ over 15 years or more in Swedish urban governance as the
reflection of a Battle of Sexes game.
3
Virtuous and Viscous
Circles in Democratic
Network Governance
B. Guy Peters
Much of extensive literature on network governance centers on

established patterns of relationships among actors (see Bogason 2000;
Sørenson and Torfing 2003) in networks, although the relationships
found at any one time are, of course, the products of development over
time. The nature of the networks at that single time reflects the political
and organizational processes that have created them, the continuing
challenges of maintaining effective networks, and also comprise the foun-
dation for strengthening or weakening the relationships in the future.
Like most important social phenomena there is a dynamic that supports
networks, but also like most other social phenomena, those dynamics
are less well-understood than are the static conditions.
As we study networks and their relationship to the public sector and to
public policy, there is a tendency to focus on those relationships that
have been successful. This tendency is in part because those relationships
that are not so successful are generally dissolved rather quickly and
hence are not so visible as are the more successful ones. Further, we rarely
ask questions about the networks that never even formed at all, and what
conditions led to their absence. To the extent that the theory on net-
works develops further we should be able to specify conditions that are
conducive to the formation of these structures and then examine likely
circumstances in which the networks did not form – the dogs that do not
bark. The focus on the surviving networks, therefore, appears to consti-
tute a selection bias (Geddes 2003), if at times an unwilling or uncon-
scious one, in the study of these structures, and therefore may lead us to
believe theoretically that networks are more significant actors, and more
readily formed structures, than they may actually be in the real world.
61
A more conscious bias in favor of functioning networks may creep

into the work because most scholars who do research on networks are
not only interested in these patterns of governing intellectually, but also
tend to be committed to this form of governing as a more democratic
alternative to the traditional patterns of parliamentary democracy (see
Denters et al. 2003). Therefore, there has been some tendency (probably
again largely unwitting) to report positive results and to dismiss any
failures as aberrations.1 While this commitment to the approach to
governance is understandable, and in some ways is laudable, it can make
gauging the true size of the population of networks, and therefore their
relative success and failure in organizing social actors for involvement in
the policy process, more difficult than it might otherwise be.
Even in the cases of positive network formation and operation, however,
there has been little discussion of the dynamics of developing effective
patterns of interaction within the networks. To the extent that there
have been explanations of successful development, these explanations
have assumed that the individuals involved, and especially the skills of
their leadership, were the principal factor in the observed success (see
Kickert et al. 1997). Those personal factors are certainly important, but
there are other structural factors that also should be considered when
conceptualizing these processes of developing networks. In particular,
the nature of the organizations as collective actors (rather than the indi-
viduals within them), the tasks being undertaken by the networks, and
the political environment in which the network operates will all have
some influence on the success of these structures.
This paper will utilize institutional theory to help understand the
process of creating and maintaining networks, and also for understand-
ing the range of relationships among public and private sector actors
involved in the policy process. In particular, I will be discussing process
of institutionalization and deinstitutionalization within structures that
link state and society. These processes have been discussed primarily in
reference to more formalized structures, but much of the same logic can
apply to these less formal social structures composed of a number of
individual organizations (see Peters 2005). The various conceptualiza-
tions of institutionalization in the literature all provide some insight
into processes of creating stable patterns of interaction among the actors
within an organization, or in our case a collection of organizations.
Institutionalization and deinstitutionalization
The fundamental process of concern here is the process of creating and

strengthening institutional relationships among the actors within a
Virtuous and Viscous Circles in Democratic Network Governance 63
network, and conversely the processes by which those relationships can

be weakened. The concept of institutionalization is a familiar one in
political analysis (Polsby 1968; Huntington 1968), but tends to take on
different meanings dependent upon the particular conception of
institutions that is being used (see Torfing this volume). Institutions are
such a central phenomenon in social and political life that a number of
scholars have addressed these structures, and naturally have done so in
his or her distinctive manners. Thus, there is no single conception of
institutionalization (or indeed institutions) that can be used to discuss
virtuous and vicious cycles in networks, but rather there are several
alternatives.
For normative and sociological conceptions of institutions institution-
alization involves creation of a normative structure that can influence
the behavior of individuals who are members of the structure. Philip
Selznick (1954), a leading scholar in the founding days of this tradition,
argued that institutionalization was ‘infusing a structure with meaning
greater than necessary for the mechanical achievement of their tasks’. In
other words, creating an institution (as opposed to a simple organization
that can be effective even if it remains merely a mechanical entity)
involves creating within the individuals who participate an attachment
that extends beyond simply getting a job done – the structure has to
mean something to the individuals.2
In this sociological conception deinstitutionalization involves the
members of the structure losing that personal commitment and begin to
conceptualize their involvement with the structure in purely mechanis-
tic and utilitarian manners (see Oliver 1992). Alternatively, individuals
may begin to attach a different meaning to their involvement from the
one held by the organizers of the structure, and therefore create
different internal cultures within the one structure. Those alternative
interpretations of the reality within which the institution functions
then makes it substantially more difficult for the institution (in this case
a network) to fulfill its stated purposes. This pattern of multiple realities
may be particularly likely in networks, given that many different
conceptions of the policy area can be held by different organizations
which are involved in the network. For example, much of the network
literature in the Netherlands focuses on networks working in environ-
mental policy, and environmental groups, farmers, industrialists and
ordinary citizens may attach different meanings to their involvement,
even if it is meaningful for all of them.
Rational choice conceptions of institutionalization focuses on rules
and compliance with rules, or on the manipulation of incentives in
order to produce the involvement of individuals with the institution
(Ostrom 1990; Shepsle 1989). Given their informal nature and minimal
enforcement mechanisms networks are not ordinarily well-suited for the
imposition of rules, although informal rules may begin to have the same
force as more formalized norms. Another means of considering the for-
mation of institutions from a rational choice perspective is that they are
a mechanism for reducing transaction costs among the actors involved
in the network (Williamson 1996). Reducing those costs can be
done through rules, through creating routinized patterns of interaction,
or through moving decisions with collective consequences into some
larger structures.3 Thus, networks may want to maintain internal
solidarity by avoiding some decisions.
Rules are not likely to be as viable a source of compliance in policy
networks as they are in other settings, but the networks do still provide
powerful incentives for their members to participate. The ability to
participate in making and implementing policies relevant to their
members provides organizations an important reason to remain mem-
bers of the network, and hence they are likely to remain members so
long as they capable of exerting that influence. Indeed, the danger is
that organizations become coopted by their membership in networks
and begin to consider membership in the network more important
than pursuing the particular goals of their members.4 Further, institu-
tionalization in this utilitarian conception of the dynamics of net-
works would be considered too calculating by the normative
institutionalists, who would require some deeper commitment to the
structure in order to say that institutionalized stability had been
achieved.
Finally, from the perspective of historical institutionalism, the nature
of networks and the interaction among their members may be struc-
tured by their initial formative moments. That perspective, however, to
a significant extent begs the question of when that moment has
occurred for a network, and how networks can respond to changing pol-
icy demands and changing membership if their formation is so crucial.
Paul Pierson (2000a), for example, has emphasized the importance of
positive feedback in reinforcing behaviors in the historical institutional-
ism, with a virtuous cycle thereby being institutionalized and being able
to carry the initial policy ideas forward to even greater success. This
argument is close to that of the rational choice institutionalists, but
focuses more on the maintenance of the policies of the network more
than on the involvement of actors, something that appears assumed in
much of the historical institutionalist literature.
Factors associated with virtuous and

viscous spirals
While the above conceptual definitions and schemes for analysis

provide the means for understanding changing relationships among the
actors in networks, we should go further, and begin to identify variables
that will contribute to the success or failure of policy networks. Unlike
the numerous schemes for identifying institutions and institutionaliza-
tion, there has been somewhat less work identifying the factors associ-
ated with the formation and maintenance of those structures. Christine
Oliver (1992) has, however, developed an analytic scheme for examin-
ing the factors associated with the deinstitutionalization of existing
institutions, and this scheme can simply be run backwards to consider
factors that may be associated with the initial development of institu-
tions, and their successful maintenance. Although her analysis was
directed primarily toward private sector organizations of various types,
the same logic can be used to understand the creation, or dissolution, of
institutionalized network structures.
Oliver’s framework contains five factors that affect the possibility and
rate of change in institutions. Two of these factors are characteristic of
any organization or institution, not just network structures. On the one
hand, all social systems are entropic. Unless ‘energy’ of some sort is
imported (see Katz and Kahn 1978) into the social system that system
will tend to break down, or in terms relevant for this paper, they will
deinstitutionalize. So, for networks, the energy that is needed from the
environment is the continued active support of the relevant social
actors, as well as perhaps the capacity to attract new members whenever
new groups are formed and become significant actors in the policy field.
Also, if the problem with which the network is concerned develops new
dimensions, it may have to adjust and begin to import new actors if it
wishes to be successful.5 Thus, if we assume that the environment of
most policy networks is turbulent, then there will be a need for contin-
uous adjustment and involvement of actors as they emerge to fit into
niches created in that environment.
In contrast, institutions and other social structures have some inertial
pressures so that once they have been created they tend to persist unless
there are significant reasons to dismantle the structure (see Zucker
1988). For networks in the public sector inertial pressures tend to be
especially strong for those institutions that have been in place for longer
periods of time. As in the conventional arguments from historical
institutionalists these networks and their patterns of doing business

have been able to survive initial tests and have institutionalized an
effective pattern. Although they may face future challenges their inter-
nal processes are assumed to have become sufficiently routinized to
weather those challenges, and to overcome the types of pressures for
deinsitutionalization that are outlined below.
Inertia may be even more relevant for social systems than other types
of systems because the individuals involved in the institutions derive
some benefits simply from their participation in the organization.6 For
networks the intrinsic values of participation may not be so great for
individuals, given that they are already members of an organization that
should be providing some of those benefits. Still, the opportunity to
interact with a wider range of people concerned with the same issues,
and perhaps also to be involved in a larger pattern of social action may
enhance the intrinsic appeal of participation and build on the existing
inertia.
The other three factors in Olivier’s framework for understanding
deinstitutionalization (or by inference institutionalization) are identified
as being political, functional and social. These pressures are certainly
relevant for the networks with which we are concerned in this analysis,
although their nature may be substantially different than for private
sector organizations. The framework is relevant even though networks are
composed of individual organizations that themselves may be subject to
the pressures for deinstitutionalization. Although the individual organi-
zations would be a worthy subject for an inquiry of their own, in this
Chapter I will consider changes at that level to be a second order conse-
quence (or perhaps cause) of institutionalization or deinstitutionalization
at the network level.
Political factors
Much of the theorizing about institutionalization and the breakdown of

institutions has been done in the context of the private sector so that
‘political’ in this framework means something rather different than we
might assume at first glance. In this model ‘political’ refers to the extent
to which the established values and myths within the institution are
accepted, and the success of the institution in producing the desired
outcomes for its members. In other words, institutions will be engaged
in virtuous cycles when the institutionalized practices produce both
greater acceptance among the members, and therefore also greater
involvement of those members in the functioning of the network.
Saying that growing consensus or dissensus on values and symbols

within a network is crucial for its maintenance borders upon being a
tautology. The more important question is the source of that internal
consensus, or lack thereof. To some extent that internal conformity of
network members may be a function of the consistency of ideas and real
economic interests among the membership. Part of the logic of networks
is to include the range of interested parties surrounding a policy area,
but that inclusiveness may increase problems in maintaining internal con-
sensus. The crucial factor for the internal effectiveness of the institution
(network) may be the capacity to create common commitment toward
the content of the network – a product or a process or a clientele – even
if the actual approaches to that content may be substantially different
(see Scott 1987).
Some network scholars have discussed this dimension of network
development in cognitive terms.7 In particular, scholars have argued
that effective management of networks involves eliminating the
‘dialogue of the deaf’ that may result from alternative conceptions of
the tasks of the network and the creation of common languages and
understandings. This process then involves altering to some extent the
cognitions of the members of the network. Part of the task for leadership
is in framing the issues and the responses for the network.8 The term
framing generally is used to describe the process for developing
common conceptions of policy issues among a set of actors, and Schön
and Rein (1996) and others (see Bevir and Rhodes, Chapter 4 in this
volume) have used the concept of ‘reframing’ to describe how changing
cognitive maps in a policy are can enable seeming intractable policy
problems to be resolved. That cognitive level of action may be crucial for
networks that are truly inclusive of a range of actors and interests in a
policy area. In a network with those diverse values developing a
common frame may be the only means of actually achieving institu-
tionalization, a premise that would fall clearly in the sociological insti-
tutionalism described above. Similarly, Eugene Bardach’s concept (1999)
of ‘collaboration’ as means of coping with the classic administrative
problem of coordination is in essence a call for reframing issues so that
multiple organizations can find them acceptable to their own policy
commitments while at the same time cooperating with others.
In addition to the necessary cognitive mobilization, the political
capacity to generate consensus in the network may also be a function of
the degree of mutual dependence of the members of the network.
Although almost by definition the members of a network have some
level of interdependence, that mutual dependence does vary. For
example, if the network for delivering a service involves both funders

and the groups that can actually deliver the services, and representatives
of the recipients of the services then this structure is more likely to be
successful in cooperation than are structures that have members that are
less dependent upon one another. Resource dependency approaches to
politics (Rhodes 1988) and to organizational theory have pointed to the
importance of that interdependency in defining interactions and in
institutionalizing the networks.
Further, the internal consensus or dissensus within a network may be
affected by its connection with other institutions and networks. On the
one hand institutions may want to be connected broadly to their task
environments in order to be able to influence those environments and
to be more successful (see functional pressures, below) in achieving their
policy goals. On the other and, however, to the extent that the network
is connected to the external task environment its internal consistency
and the integration among the members may be lessened. In particular,
to the extent that individuals or in the case of networks organizations
play boundary spanning roles those structures may not be able to main-
tain their institutional integrity to the extent that others structures are.
That boundary-spanning role may create resource dependence outside
the individual network and hence potentially make it more effective in
achieving its tasks in that external environment.
Politics is a factor in shaping networks therefore both within the
networks and between the networks and the external environment. To
some extent the arguments presented here (and in the following section)
present networks with a dilemma. On the one hand, if the emphasis is
placed on creating external links that could be functional for the achieve-
ment of goals within the task environment then the capacity to create a
common frame within the network itself may be reduced. On the other
hand, however, too much of an emphasis on the external dimension may
reduce internal cohesion and lead eventually to deinstitutionalization.
While I am attempting to minimize the role of individuals in this analysis,
this is clearly a problem for which effective leadership within the network
will be crucial.
Framing and reframing are appealing analytically, but practically may
not be as easy to achieve as it might appear. We should expect the
interest groups and other actors involved in the typical network to have
commitments to particular ways of viewing and policy area, and also to
have genuine interests. Likewise, ideologies may intervene to limit the
capacity of members of a network to cooperate and to develop a com-
mon conceptualization of their tasks and their environment. Therefore,
simply saying one wants to reframe may well be insufficient to over-

come the resistance. At this point the political capacity of the leadership
of the network becomes crucial for institutionalization.
Functional factors
In addition to the political factors, based largely on the self-interests of

the members of the networks, what Oliver calls functional factors influ-
ence the capacity of networks (or other structures) to institutionalize.
Practices within networks may be more or less conducive to their achiev-
ing policy goals for the members, or even in the internal management of
the affairs of the network itself. In other words, while some practices may
not be favored by the members of an institution those practices may still
be effective, and hence may be maintained even in the face of the disfa-
vor of some members. Conversely, even if internal practices continue to
be accepted internally by the members they may be abandoned if they
are externally dysfunctional (see Covaleski and Dinsmith 1988).
Given the above definitions of functional factors, I will also be
concerned with the macro-level relationships between the policy-making
processes (including networks involved in the process) and the public as
a whole. In particular, there are two alternative outcomes possible in the
relationships between the success of building networks internally, and
their success of making and implementing policy. We can think of these
two outcomes as the internal and the external success of networks.
The Oliver framework tends to treat these two forms of success as comple-
mentary processes, but in reality the two goals may be almost conflict-
ing with one another for the attention of members of networks. Milward
and Provan’s work (1998), on the other hand, tends to demonstrate the
close linkage of internal and external factors in the effectiveness of
network governance.
On the one hand, we could argue that there are virtuous circles in
which success in making and implementing public policies are associated
with more positive perceptions, and greater legitimacy, of the policy-
making apparatus. In terms of the normative approach to institutions
the myths, symbols and routines of the network will become more
widely accepted. That greater legitimacy, in turn, will enhance the
capacity of these policy-making structures to be effective in subse-
quent rounds of making policy, and success will beget success for the
future of the network. In other words, success may breed success at this
locus for analysis, as well as in the relationships among the actors
themselves.
The alternative hypothesis would be that higher levels of integration

within a policy network may separate that network from the rest of the
policy system. That separation, in turn, may make the network less
successful in achieving its collective purposes in the context of a
complex policy process involving a number of actors and a variety of
connections with the private sector. Effective policy making in many
policy fields, and almost certainly a growing number of policy areas.
may require the capacity to coordinate and integrate across policy areas.
Therefore, creating higher levels of integration in a network within the
policy area may involve some judgment about the extent to which the
network should focus on internal consistency and how much it should
focus on the capacity to work effectively with other networks within
their own, and other, policy areas.
Social pressures
Just as networks themselves may change, so too can the environment

within which they function. At the most proximate level these changes
in the social setting of the network can reflect changes in other networks
with which any one network interacts. At a more remote, but perhaps at
a more significant level for networks concerned with public policy, there
can be changes in the receptivity of government to networks pressures
or changes in the policy priorities of government. For example, if there
are high levels of political change in the society and frequent turnovers
in government then networks may not be as effective in institutionaliz-
ing themselves and institutionalizing their political influence as they
would be in more stable political environments. Those problems would
be exacerbated if the changes in partisan control of government also
involved significant ideological changes.
During the past several decades the rapid changes introduced in
government practices will have had a major influence on networks
and on their capacity to exert influence on policy. For example, the NP
Management (NPM) has tended to influence government in the direc-
tion of utilizing markets more than political influences such as networks
to make and implement policy. To some extent the market-based
reforms usually associated with NPM and participatory reforms reflected
in the use of networks have been implemented simultaneously, sometimes
in the same countries and policy areas. Therefore, there are changes in
the policy-making environment that may make the effectiveness of
networked governance formats less effective than they might otherwise
be, and perhaps producing some deinstitutionalization.
Just as governments have been altering their patterns of delivering

policy, the society may also change its assessment of the efficacy of
political action versus market action, or other forms of private action, in
obtaining goals. The well-documented loss of faith on public action
(Pharr & Putnam 2000) may also affect the involvement of actors in
policy networks, even if those networks are perceived as an alternative
to more conventional forms of public sector action. If actors begin to
believe that public action is fruitless, their involvement with the
network may actually intensify, believing that these more informal
structures may be the only effective means of achieving their goals
(McLaverty 2002).
Other factors in explaining success
The Oliver model provides a useful framework for analyzing processes of

institutionalization and deinstitutionalization, but several other factors
should also be mentioned that do not fit neatly into this analytic scheme.
Although one purpose of the institutionalization framework is to remove
the emphasis on individuals, some impact of individuals can not, or
should not, be ignored.
Characteristics of the members
Individuals do play an important role in explaining the success or failure

of networks in creating a virtuous cycle of development. It is rather
obvious that organizations that are better resourced and populated
with more capable individuals will have a higher probability of being
effective both internally and externally.9 That having been said,
however, there are other, somewhat less obvious, characteristics of the
members of networks that will also influence the upward or downward
trajectories of the relationship. The consistency of the membership of
the network will have a significant impact on the development of the
network. In the first instance a membership of an network that is more
consistent in terms of their interests (meaning economic or social inter-
ests) is more likely to be successful in maintaining internal cohesion and
institutionalizing, especially in the sense defined by Selznick. Further,
these networks would be capable of making decisions readily (given the
relative lack of internal dissent) and hence might be more effective
political actors in dealing with their external environment.
On the other hand, more consistent networks may have a more
difficult times being as effective – defined in terms of having their
decisions actually implemented – as desired by their members than

would more heterogenous networks. The more heterogenous networks
may have difficulty in making decisions and in creating social integra-
tion, but once they have decided something it is more likely to be
acceptable to other relevant interests in the policy area. The danger of
success in implementing programs for the network is that the decisions
that are possible with a wide range of interests involved are suboptimal.
Indeed, they may not be much different from the decisions that might
be reached by the interaction of multiple networks, each of which was
more internally homogeneous.
The discussion of ‘epistemic communities’ in the literature on
networks (Adler 1992) demonstrates one of the most important sources
of consistency among a network, and hence a major source of successful
institutionalization. These policy communities founded on the basis of
a common body of knowledge tend to have very strong integration,
given that the members have already been socialized into a profession or
into a scientific discipline that tells then what the important questions
are, and what the right answers to those questions probably are. Again,
however, the greater likelihood of successful integration of these struc-
tures may make it less probable that they will be able to deal effectively
with other groups who lack the common perspective on the issues.
That difficulty in interaction with other networks may be overcome for
purposes of external effectiveness by the legitimacy conferred by
scientific expertise, but in the end the technical solutions may not
always triumph in a complex political environment.
As networks develop over time they face the need to replace member
or simply add new members. The capacity of the network to control its
own patterns of socialization, and perhaps even to control its own
membership, will be important for enhancing and maintaining integra-
tion and institutionalization. In her discussion of institutions Elinor
Ostrom (1990) discusses membership rules as one of the crucial deci-
sions that must be made for the development, and the maintenance of
institutions, and for our purposes the development of networks.
Although, as noted, networks have perhaps more difficulty in enforcing
rules than do other types of structures, membership may be one area
that is easier for them to control.
Finally, the relative equality among the constituent members of the
network will influence the degree of institutionalization and integration
within the network. One can, however, hypothesize contradictory
patterns of relationship. On the one hand, a network composed or rela-
tively equal actors, in terms of size, resources, etc. may find integration
easier. The absence of a dominant player may facilitate the cooperation

of the members who would all believe that they are able to pursue their
interests and shape the programs of the network. On the other
hand, however, having a strong and even dominant actor involved in
the network may provide a focus for its activity and may enhance
institutionalization.
Operating environment
As well as the characteristics of the members affecting the capacity of

networks to integrate themselves and to be successful in pressing their
demands on the political systems, the nature of the operating environ-
ment also influences those networks. Although I will now consider these
issues separately, they may also be related to the characteristics of the
networks themselves. The networks can not be considered in isolation
from the content of the programs that they are involved with, not
the overall political context within which they function. Most of the
discussion of the virtues, or even possibilities, of network governance
has been conducted within Northern European countries characterized
by high levels of individual and social trust. This precondition is not
available in all European societies, much less in most of the rest of the
world. Therefore, to the extent that these conceptions of governance are
to be generalizable, they may have to be considered in very broad
comparative context. That having been said, the development of net-
work type structures may be a means of fostering greater social trust and
cooperation. Individuals may learn how to cooperate within a network
of common interests and that learning may contribute to developing
more extensive social capital in the society.
At one point in the development of the literature on policy implemen-
tation scholars of public administration were criticized for focusing on the
‘single lonely organization’ (Hjern & Porter 1980). While that problem has
been overcome (see Winter 2004) much of the contemporary literature
might be criticized for focusing on the ‘single lonely network’. Just as insti-
tutional theory has as a basic premise that the environment of institutions
is other institutions, so too should the network literature concentrate some
attention of the environment of any one network, composed at least in
part by other networks. The complexity of that environment will influence
the capacity of each network to integrate programs, just as more complex
environments affected the capacity of individual organizations.
One of the most relevant concerns for a network is the range of issues
with which it must be involved. I will hypothesize that the greater the
range of issues that any network is forced to cope with politically the
greater will be the internal strains and hence the less will be the likeli-
hood of successful institutionalization of the network. While we can
define a network as those organizations interacting over a single policy
area or issue that may merely define away the problem, rather than
attempting to understand the consequences of the almost inevitable
connection of any network to a range of concerns and the need to work
in ways that produce more effective integration.
The nature of the operating environment of networks, as composed of
other policy networks and other programs makes coordination one of
the important policy demands for networks. While individual programs
must be made to work well, so too must the assembly of programs in
government as a whole. At a minimum the programs within a particular
area of policy, e.g. economic development or social policy, should work
together effectively. Coordination has been one of the continuing
concerns of government, and given the connections of networks to
governance the need for coordination is no less. However, as I argued
above with respect to the general effectiveness of networks, I would
expect that networks that are more effective internally would be less
effective in dealing with other networks in their environment.
Tasks
Yet another factor affecting the capacity of networks to integrate effec-

tively is the nature of the tasks being undertaken by the network.
Networks are in some settings responsible for both policy formulation
and for policy implementation, or they may responsible for only one or
another of these activities. Everything else being equal, we would expect
networks responsible for implementation to find it easier to institution-
alize and to integrate than are those responsible for both activities, and
also that networks responsible for implementation only would have the
easiest time developing stable working patterns.
Formulating public policies is potentially highly contentious, especially
if the network is inclusive and involves a range of social interests. As a con-
sequence having to decide on a common approach may make institution-
alization difficult, especially when considered from the sociological
perspective of creating common value commitments among the members.
Summary and conclusions
Networks are an increasingly important component of governing in a

range of countries, especially those in Northern Europe. Even if one is
not as committed to this concept as a means of enhancing governance

and legitimacy as some scholars are, it is difficult to deny the role that
these aggregations of organizations and individuals often do play in
governing. The role played by networks varies substantially by country
and by policy area within countries, but in order to be effective these
institutional structures must develop and then maintain some degrees
of internal commitment from their members. They must also have some
capacity to make decisions that can satisfy their members and achieve
their own collective policy goals. Even if networks are successful in
forming initially, they are still under threat of deinstitutionalizing and
succumbing to the entropic pressures present in any social system.
Although the individuals involved in networks are important for
shaping those networks, there are also a number of structural factors
that can affect the success of these structures. The types of organizations
that are involved in networks, the size of those networks, and the range
of activities in which the networks are involved all affect the ability of
networks to institutionalize effective internal interactions. Those same
factors also affect the capacity of the networks to represent the interests
of their members to official and unofficial actors outside the network.
Given that to be successful in getting what it (or more precisely its mem-
bers) want from their involvement in the policy process any one net-
work is likely to have to work with others, the institutional approach
also must be extended to examine interactions between that network
and its environment of other networks and institutions.
Notes
1. In fairness, most areas of the social sciences tend to report positive results
rather than results that might bring into question our favorite theories.
2. In terms of Etzioni-Halevy’s (1964) theory of organizations the participants
must have a moral form of compliance with the dictates of the institution,
rather than the remunerative relationship that would be adequate for most
organizations.
3. Williamson used transaction cost analysis as a means of understanding the
virtues of the firm within the more market as a means of organizing activity
and bringing together.
4. These cooptive arguments have been made about several types of European
policy making arenas, and have been in place for some years. See, for example,
Heisler (1974) and Duran (2000).
5. For example, as immigration becomes a significant dimension for employ-
ment policy in many countries networks established to deal with labor market
issues must expand to include those new actors.
6. One classic approach to organizations focused on the ‘solidary incentives’
offered by memberships in the organization that is, some people would
participate in organizations simply for the social interactions, rather than to

get anything material or to achieve any particular purposes.
7. The institutional literature also has utilized cognitive terms to describe the
basis for forming institutions. One of W. Richard Scott’s (1995) three dimen-
sions of institutional analysis is the cognitive, along with the regulatory and
the normative.
8. In institutional theory the sociological/normative approach has argued this
role for leadership very clearly in the context of institutional change. See
Brunsson and Olsen (1993).
9. Obvious, perhaps, but not necessarily always true. For example, some
organizations of the disadvantaged have been successful in registering moral
claims that have overcome the apparent powers of opposing interests.
4
Decentred Theory, Change and
Network Governance
Mark Bevir and R. A. W. Rhodes
Introduction
Governance is often defined as self-organizing, inter-organizational

networks. Of course, people define governance in all kinds of ways.
Nonetheless, social scientists typically appeal to inexorable, impersonal
forces such as the functional differentiation of the modern state or the
marketization of the public sector to explain the shift from hierarchy to
markets to network governance. We will invoke the Anglo-governance
school as an illustration of such a postivist approach to network
governance. We also will offer a decentred alternative to such positivist
approaches. To decentre is to focus on the social construction of a
practice through the ability of individuals to create, and act on, mean-
ings; it is to unpack a practice in terms of the disparate and contingent
beliefs and actions of individuals (Bevir & Rhodes 2003: ch. 4). When we
decentre governance, we challenge the idea that inexorable, impersonal
forces are driving a shift from bureaucratic government to networks. We
argue, instead, that governance and networks are constructed differently
by many actors against the background of diverse traditions.
After decentring governance theory, we turn our attention to the
analysis of network dynamics. We contrast the mainstream accounts of
the conditions under which networks change, succeed and fail with a
decentred account. We also provide a brief illustration of how ethnogra-
phy helps us to recover meanings among network actors. Finally, we
criticise comprehensive accounts of governance, suggesting that a
decentred theory prompts us to study the everyday practices of situated
agents whose beliefs and actions are informed by traditions and
expressed in stories.1
77
Positivist approaches to network governance
Behind the idea of network governance, there usually lurks the idea that
its emergence reflects something akin to a logic of modernization – a
logic of functional specialisation and institutional differentiation.
Entrenched institutional patterns purportedly ensured that neo-liberal
reforms lead not to markets but to the further differentiation of policy
networks in an increasingly hollow state. Social scientists typically use a
concept of differentiation here to evoke differences based on function.
Because they use differentiation in this way, they offer broadly positivist
accounts of governance. They treat governance as a complex set of
institutions and institutional linkages defined by their social role or
function. They make any appeal to the contingent beliefs and preferences
of agents largely irrelevant.
In Britain, positivist accounts of network governance challenge the
Westminster model (Rhodes 1997a, 2000c; Richards & Smith 2002,
Stoker 1999, 2004; and for discussion Marinetto 2003). They capture
recent changes in British government in a way the Westminster model
cannot. They start with the notion of policy networks or sets of organ-
isations clustered around a major government function or depart-
ment. These networks commonly include the professions, trade
unions and big business. So, the story continues, central departments
need such networks to cooperate in delivering services. They allegedly
need their co-operation because British government rarely delivers
services itself; it uses other bodies to do so. Also, there are said to be
too many groups to consult so government must aggregate interests;
it needs the legitimated spokespersons for that policy area. The groups
in turn need the money and legislative authority that only govern-
ment can provide.
Policy networks are a long-standing feature of British government;
they are its silos or velvet drainpipes. The Conservative government of
Margaret Thatcher sought to reduce their power by using markets to
deliver public services, bypassing existing networks and curtailing the
‘privileges’ of professions, commonly by subjecting them to rigorous
financial and management controls. But these corporate management
and marketization reforms had unintended consequences. They frag-
mented the systems for delivering public services, creating pressures for
organisations to co-operate with one another to deliver services. In
other words, marketization multiplied the networks it aimed to replace.
Commonly, packages of organisations now deliver welfare state services.
Positivist accounts of governance thus concentrate on the spread of
Decentred Theory, Change and Network Governance 79
networks in British government. They tell us not only that fragmentation

created new networks but it also increased the membership of existing
networks, incorporating both the private and voluntary sectors. They
also tell us that the government swapped direct for indirect controls, so
central departments are no longer either necessarily or invariably the
fulcrum of a network. The government can set the limits to network
actions: after all, it still funds the services. But it has also increased its
dependence on multifarious networks.
The Anglo-governance school conceives of networks as a distinctive
co-ordinating mechanism notably different from markets and
hierarchies and not a hybrid of them. They associate networks with
characteristics such as trust. In their view, trust is essential because it is
the basis of network co-ordination in the same way that commands
and price competition are the key mechanisms for bureaucracies and
markets respectively (see Frances 1991: 15, and Powell 1991a). Shared
values and norms are the glue that holds the complex set of relation-
ships in a network together. Trust is essential for co-operative behav-
iour and, therefore, the existence of the network. With the spread of
networks there has been a recurrent tension between contracts on the
one hand with their stress on competition to get the best price and net-
works on the other with their stress on co-operative behaviour. Other
key characteristics are said to be reciprocity and interdependence
(Rhodes 1997c).
In short, the Anglo-governance school tells us a story of fragmenta-
tion confounding centralisation as a segmented executive seeks to
improve horizontal co-ordination among departments and agencies and
vertical co-ordination between departments and their networks of
organisations. An unintended consequence of this search for central
control has been a hollowing out of the core executive. The hollowing
out of the state suggests the growth of governance has further under-
mined the ability of the core executive to act effectively, making it
increasingly reliant on diplomacy. The state has been hollowed out from
above by for example international interdependence, and from below
by for example marketization and networks, and sideways by agencies.
Internally the British core executive was already characterised by bar-
onies, policy networks and intermittent and selective co-ordination. It
has been further hollowed out internally by the unintended conse-
quences of marketization, which fragmented service delivery, multiplied
networks and diversified the membership of those networks. Externally
the state is also being hollowed out by membership of the EU and other
international commitments.
Decentring network governance
A decentred theory of network governance explores the institutions

of governance by studying the contingent meanings that inform the
actions of the individuals involved in all kinds of practices of rule.
Positivist accounts of network governance focus on issues such as
the objective characteristics of policy networks and the oligopoly of the
political market place; they stress power-dependence, the relationship of
the size of networks to policy outcomes, and the strategies by which the
centre might steer networks. To decentre governance is, in contrast, to
focus on the social construction of patterns of rule through the ability of
individuals to create meanings in action. A decentred approach changes
our conception of governance. It encourages us to examine the ways in
which patterns of rule, including institutions and policies, are created,
sustained, and modified by individuals. It encourages us to recognise
that the actions of these individuals are not fixed by institutional norms
or a logic of modernization, but arise from the beliefs individuals adopt
against the background of traditions and in response to dilemmas.
A decentred theory highlights the importance of beliefs, traditions,
and dilemmas for the study of governance (see Bevir & Rhodes 2003,
2006). Any existing pattern of rule will have some failings. Different
people will have different views about these failings since they are not
simply given by experience but rather constructed from interpretations
of experience infused with traditions. When the perceived failings of
governance are in conflict with people’s existing beliefs, such failures
pose dilemmas. Such dilemmas push people to reconsider their beliefs
and the traditions informing those beliefs. Because people confront
these dilemmas against the background of diverse traditions, there arises
a political contest over what constitutes the nature of the failings and
what should be done about them. Exponents of rival positions seek to
promote their particular sets of theories and policies. This contest then
leads to a reform of governance. So, any reform can be understood as a
contingent product of a contest of meanings in action.
The reformed pattern of rule established by this complex process will
display new failings, pose new dilemmas, and be the subject of compet-
ing proposals for reform. There will be a further contest over meanings,
a contest in which the dilemmas are often significantly different, and
the traditions have been modified from accommodating previous
dilemmas.
All such contests take place in the context of laws and norms that
prescribe how they should be conducted. Sometimes the relevant laws
and norms have changed because of simultaneous contests over their

content and relevance. Yet while we can distinguish analytically between
a pattern of rule and a contest over its reform, we rarely can do so
temporally. Rather, the activity of governing continues during most
contests, and most contests occur partly within local practices of
governing. What we have, therefore, is a complex and continuous process
of interpretation, conflict, and activity that produces ever-changing
patterns of rule.
A decentred theory of network governance entails a shift of topos
from institutions to meanings in action. It suggests positivist approaches
to governance tend to restrain the centrifugal impulse of narratives of
network governance. Positivist approaches reduce the diversity of
network governance to a logic of modernization, institutional norms,
or a set of classifications or correlations across policy networks. Their
proponents tame an otherwise chaotic picture of multiple actors creating
a contingent pattern of rule through their conflicting actions.
A decentred theory implies that network governance arises from the
bottom-up. Any pattern of governance is a product of diverse practices
made up of multiple individuals acting on all sorts of conflicting
beliefs which they have reached against the background of several tradi-
tions and in response to varied dilemmas. A decentred theory leads us,
then, to replace aggregate concepts that refer to objectified social laws or
institutions with one’s that we craft to explain the particular beliefs and
actions of interest to us. It inspires narratives of traditions and dilem-
mas. So, what does this decentred theory tell us about the analysis and
management of change in the networks.
The analysis of change in networks
Current explanations of how networks change appeal to inexorable,

impersonal forces such as the functional differentiation of the modern
state or path dependency. They rely on exogenous, not endogenous,
causes, arguing, for example, that networks create routines for policy
making and change is incremental (Marsh & Rhodes 1992: 261). They
identify four broad categories of change: economic, ideological, knowl-
edge and institutional, and all are external to the network. Indeed, the
most common and recurrent criticism of policy network analysis is
that it does not, and cannot, explain change (for a summary of the
argument and citations see Richardson 2000). So, policy network analy-
sis stresses how networks limit participation in the policy process;
decide which issues will be included and excluded from the policy
agenda; shape the behaviour of actors through the rules of the game;
privilege certain interests; and substitute private government for public
accountability. It is about stability, privilege and continuity.
There have been several efforts to build the analysis of change into
governance networks (see for example Sabatier & Jenkins-Smith 1993;
Marsh & Smith 2000; Richardson 2000). We focus on decentred analysis
because it offers both an explanation of change in networks and the
appropriate tools of analysis.
A decentred theory prompts a shift of topos from institution to
individual and a focus on the social construction of policy networks
through the ability of individuals to create meaning (Bevir & Rhodes
2003, ch. 4). It thereby encourages us to look for the origins of change
in the contingent responses of individuals to dilemmas. As noted earlier,
a dilemma arises for an individual or institution when a new idea stands
in opposition to an existing idea and so forces reconsideration. Because
we cannot read-off the ideas and actions of individuals from objective
social facts about them, we can understand how their beliefs and
actions, and social practices change only by exploring the ways in which
they think about, and respond to, dilemmas. An analysis of change and
developments in government must take place through a study of relevant
dilemmas. We build change into the heart of our account of networks by
exploring how individual actors respond to dilemmas and reinterpret
and reconstruct traditions.
Ethnography studies individual behaviour in everyday contexts;
gathers data from many sources; adopts an ‘unstructured’ approach; focus
on one group or locale; and, in analysing the data, stresses the ‘interpre-
tation of the meanings and functions of human action’ (paraphrased
from Hammersley 1990, 1–2). The task is to write thick descriptions or
our ‘constructions of other people’s constructions of what they are up
to’ (Geertz 1973: 9, 20f.; see also: Heclo & Wildavsky 1974; Bevir &
Rhodes 2003, 2006; Richards & Smith 2004).
We cannot provide a detailed exploration of change in networks here.
Any such account would need to recognise that individuals have several
antidotes to, and coping mechanisms for, challenges to their belief sys-
tems. Such challenges can take the form of responding to different
beliefs or to the actions of others and any response will be affected by
the salience of those beliefs and actions for the several parties. Also,
analyses of network dynamics require an understanding of how beliefs
are constructed both in the complex patterns of social interaction and
the handed-down traditions. However, we can illustrate both our theory
and methods ‘in action’ by a brief analysis of managing networks.
Managing change in networks
Under what conditions do networks succeed, and under what condi-

tions do they fail? How do we manage networks so they are successful
and do not fail?
The mainstream literature on networks can muster a long list of con-
ditions fostering networks (for citations see Rhodes 2006a). Networks
are said to thrive where markets and hierarchies fail, where trust and
reciprocity characterise the relationships between organisations, where
management is by negotiation, not command, and where there is a
substantial measurement of agreement on at least the means and prob-
ably the ends of policy. Also, as with any other form of public sector
management, success depends on the relevant information, skills and
resources. When actors husband information and resources, when in
effect they refuse to share, then the cooperation that defines networks
is unlikely to be forthcoming. The mainstream literature also identifies
several other more specific conditions under which networks are said
to arise.
● Actors need reliable, ‘thicker’ information, or local knowledge.

● Quality cannot be specified or is difficult to define and measure.
● Commodities are difficult to price.
● Professional discretion and expertise are core values.
● Service delivery is localised.
● Cross-sectoral, multi-agency cooperation is required.
● Monitoring and evaluation incur high political and administrative
costs.
● Implementation involves haggling (or disputes over who owns the
policy).
Equally the mainstream literature suggests that networks, like all other
resource allocation mechanisms, are not cost free and identifies the
conditions under which they will fall.
● When closed to outsiders and unaccountable for their actions.

● When they generate conflicts: between individual and organisational
commitments, local and national public expectations, flexibility and
rules, and network goals and national regulators.
● When they serve private interests, not the public interest, and are
hard to hold to account.
● When they cannot be to steered.

● When they conflict with other governing structures.
So, the task confronting governments is to create the conditions that

foster networks (or bureaucracies, or markets) and to steer them so that,
for example, they remain accountable and do not flounder through
internal conflicts. How do you manage in inter-organizational contexts
where, in sharp contrast to intra-organisational contents, no manager
can impose objectives on the other participating organizations? The
normal answer is by hands-off management and through persuasion
where independent actors agree objectives; senior management does
not set them. There has been an upsurge of advice from both academics
and consultants on what to do to run a network (and for a comprehen-
sive review of tools for network governance, see Salamon 2002). We give
two examples (see Boxes 4.1 and 4.2). Box 4.1 draws on American
experience.
Box 4.1 Ten lessons on how to manage your network
Be representative of your agency and network.

Take a share of the administrative burden.
Operate by agenda orchestration.
Recognise shared expertise.
Stay within the decision boundaries of your network.
Accommodate and adjust while maintaining purpose.
Be as creative as possible.
Be patient and use interpersonal skills.
Recruit constantly.
Emphasise incentives.
Source: Agranoff (2003: 29).
Box 4.2 Local practical solutions
Explore and agree the objectives of cross-boundary working.

Develop a shared understanding of what the network is for.
Develop an appropriate shared strategy.
Clarify roles, expectations and responsibilities for all players.
Create a culture in which cross-boundary working is likely to succeed.
Create appropriate shared service delivery systems.
Have a clear idea of what success would look like.
Source: Goss (2001: 97–100).
Or, in the same vein, but based on UK experience, Box 4.2 provides
the following pearls.
As Perri et al. (2002: 130) point out network management ‘is not
rocket science’, and it is hard to disagree. But a decentred theory raises
more fundamental objections. Typically, policy-oriented work on gover-
nance treats hierarchies, markets, and networks as fixed structures that
governments can manipulate using the right tools. A decentred theory
undercuts this idea of a set of tools we can use to manage networks. If
networks are constructed differently, contingently, and continuously,
we cannot have a tool kit for managing them. In short, there is no essen-
tialist account of networks which can be used to provide a tool kit for
managing networks.
A decentred theory encourages us to foreswear management tech-
niques and strategies but, and the point is crucial, to replace such tools
with learning by telling stories and listening to them. While statistics,
models, and claims to expertise all have a place in such stories,
we should not become too preoccupied with them. On the contrary, we
should recognize that they too are narratives about how people have
acted or will react given their beliefs and desires. No matter what rigour
or expertise we bring to bear, all we can do is tell a story and judge what
the future might bring. In a decentred approach, to generalise means to
diagnose and make informed conjectures. We cast conjecture in the
form of narratives or stories. Policy analysis is a form of storytelling. So,
for a decentred theory to produce policy advice, we must tell stories that
answer the following questions (and for a more detailed discussion see
Bevir et al. 2003). What stories do we tell? What is the plot of our story?
Who are the leading characters? What are the informing metaphors?
What proverbs do we use?
The road to understanding change network governance lies in
decentred accounts focusing on the political ethnography of networks
to generate narratives that give due recognition to the creative individ-
ual. Networks are constructed by individual actors and not created by
governments or imposed by the researcher. As researchers, we write
constructions about how other people construct the world.
So, how do we write these constructions? We use the example of
management reforms in the police service to provide an illustrative
example (Fleming & Rhodes 2005). The research draws on interviews
with 27 senior and middle-level officers and managers, and on focus
group discussions. The reforms were understood by the respondents as a
shift from command and control bureaucracy through markets to
networks, and the shift posed some acute dilemmas. Their key problem
was not the limitations of working with contracts or any other govern-
ing structure but rather the attempt to balance apparently contradictory
demands. They know how to rewrite the rulebook, manage a contract,
or work with neighbourhood watch but they struggle to reconcile these
ways of working, believing they conflict and undermine one another.
We illustrate briefly both their views on the limits networks and on the
conflicts between networks and other ways of working.
There is commitment from those who see community networking as
the future.
I think the community policing thing is a good idea – I think it works –

the problem of course is that it is hard to keep people in the same
place for significant periods, but I think it’s good, I think it’s good for
the community. We come up with lots of initiatives – we are good at
that – but we are poor finishers – too many goals really. I think we
should hit on three things and do them.
Even the traditional copper sees some virtue in a more integrated

approach.
A whole of government approach might consider bringing all services

under the police umbrella – ambulance, fire, security, so for example
if there was a major football game. The events planner could ring one
number and organise police officers, St John’s Ambulance, private
security; traffic coordination. A policy like this would give us a better
response to things too. The others might not have the powers but
they would have the powers to detain until we arrived or at least
provide a liaison point with the police on the ground. It would give
us much better surge capacity.
But there is also a clear stereotype that the police focus on crime and see
networking as soft.
Police don’t want to get into the crime prevention stuff though. No
one wants to do these jobs – they want to leave it to the warm and
fuzzies. Police want to wear their underpants on the outside and save
the world – they want to make the person pay. Culture has changed
to some extent but it is still influenced by older people. People who
are attracted to the policing role often have that mindset.
They also see the conflict of ideas in reform and it manifests itself in an
aversion to change and criticisms of the leadership but several managers
are all too well aware of the contradictions. One officer makes the point
with brutal simplicity: ‘Terrorism is a problem – it doesn’t go with the
ideology of community policing and crime management.’ Some officers
appreciate the dilemmas they confront and recognise the need to fit
their managerial strategies to the context.
Command and control is situational. In my team, I don’t have

subordinates. I have team members. Years ago a constable wouldn’t
speak to a superintendent – this is not the case now. I invite their
ideas and input and encourage them to talk to me. If they are happy
I have a productive working team. However, as I said, it’s situational.
Fighting fires is a good example. As a commander, when I want some-
thing done, it isn’t up for negotiation, I tell the troops. We have to
rely on command and control in these situations.2
Mainstream accounts of police reform use ‘police culture’ as an explana-

tory variable (see Chan 1997). As Davies and Thomas (2003: 682) point
out, police organisations are resistant to change because of ‘a co-existence
of formalized bureaucratic and standardised working practices, with a
deeply entrenched and pervasive occupational culture’ of hierarchical
subordination. However, these interviews with officers suggest that con-
flict between incompatible ideas simply makes it too easy for dynamic
conservatism to win out. Reform is impeded not by recalcitrant actors –
although it often is – or by police culture – which can act as a brake – but
more importantly by the dilemmas created by the irreconcilable ideas of
bureaucracy, markets and networks. The central dilemma of police
reform is this unholy trinity (and for a more detailed discussion see
Jessop 2000a and Rhodes 2006b).
In sum, a decentred theory of networks provides: a critique of main-
stream accounts of how networks change; an alternative conception of
networks as constructed differently, contingently, and continuously; a
storytelling alternative to the mainstream tool kit for managing net-
works; and an alternative to the techniques of positivist social science by
using ethnographic methods focus on the creative individual as the
agent of change.
Conclusions
Positivist accounts of network governance, including those of the

Anglo-governance school, aspire to be comprehensive. Social scientists
typically aim to provide a general account of what network governance

looks like and why. For example, governance is often characterised by
multiplying networks replacing the bureaucratic hierarchies of the
welfare state. Such comprehensive accounts of network governance
identify one or more defining features. This defining feature then acts as
a central focus that allegedly explains other pertinent features of net-
work governance. For example, the spread of networks allegedly
explains the greater reliance of states on ‘trust’ and ‘diplomatic’ styles of
management, or it embraces the search for co-ordination through joint
ventures, partnerships, and holistic governance. Similarly, networks
succeed where there is trust, reciprocity, cooperation and agreement.
What do such comprehensive accounts imply about the nature of
network governance? They imply, first, that we can define ‘governance’
by reference to one or more of its essential properties, such as multiply-
ing networks, where we can define a network by the interdependence of
its members (whether people or organizations). They imply, second,
that these essential properties are general ones that characterise all cases
of governance: so, we will find governance in its new guise if and only if
we find a spread of networks, or we will find networks if and only if we
find interactions based on trust across organisations. They imply, finally,
that these essential properties can explain at least the most significant
other features of network governance. A comprehensive account of
network governance makes sense, even as a mere aspiration, only if
these implications are valid. But why would we assume that network
governance has one or more essential features?
A decentred theory implies there is no comprehensive account of net-
work governance. There is no necessary logical or structural process
determining the form governance or the success or failure of networks.
Rather, an adequate theory of change and networks should accept that
networks both arise from and change as a result of diverse actions and
practices inspired by varied beliefs and traditions. Patterns of rule arise as
the contingent products of diverse actions and political struggles
informed by the beliefs of agents as they arise against the backcloth of
traditions. This conclusion applies whether we are talking about the civil
service, public sector reform, or the rise and fall of specific networks.
Once we reject the idea of a comprehensive account of network gov-
ernance, we can no longer define it by any allegedly essential properties.
Rather, we understand general concepts such as network governance by
using them in cases. Their meaning derives from the ways in which we
use them in various contexts. What is more, the absence of a compre-
hensive account of network governance suggests there are no set ways in
which we must – or must not – use the term. There need be no single
feature shared by all those cases or narratives to which we would apply
the general term ‘network governance’.
We understand ‘network governance’ as a set of family resemblances.
Wittgenstein (1972) famously suggested that general concepts such as
‘game’ should be defined by various traits that over-lapped and
criss-crossed in much the same way as do the resemblances between
members of a family – their builds, eye colour, gait, personalities. He
considered various examples of games to challenge the idea that they all
possessed a given property or set of properties – skill, enjoyment, victory
and defeat – by which we could define the concept. Instead, he sug-
gested that the examples exhibited a network of similarities, at various
levels of detail, so that they coalesced even though no one feature was
common to them all.
We do not master such family resemblances by discovering a theory or
rule that tells us precisely when we should and should not apply it. Our
grasp of the concept consists in our ability to provide reasons why it
should be applied in one case but not another, our ability to draw
analogies with other cases, and perhaps our ability to point to the
criss-crossing similarities. Our knowledge of ‘network governance’
is analogous to our knowledge of ‘game’ as described by Wittgenstein: it is
‘completely expressed’ by our describing various cases of governance in
and through networks, showing how some cases can be considered as
analogous to these and others cannot.
No doubt some of the family resemblances that characterise network
governance derive from a focus on meaning in action and apply to all pat-
terns of rule. A decentred theory highlights, first, a more diverse view of
state authority and its exercise. All patterns of rule arise as the contingent
products of diverse actions and political struggles informed by the varied
beliefs of situated agents. So, the notion of a monolithic state in control
of itself and civil society was always a myth. The myth obscured the real-
ity of diverse state practices that escaped the control of the centre because
they arose from the contingent beliefs and actions of diverse actors at the
boundary of state and civil society. The state is never monolithic and it
always negotiates with others. Policy always arises from interactions
within networks of organisations and individuals. Patterns of rule always
traverse the public, private, and voluntary sectors. The boundaries
between state and civil society are always blurred. Trans-national and
international links and flows always disrupt national borders. In short,
state authority is constantly being remade, negotiated, and contested in
widely different ways within widely varying everyday practices.
Second, these everyday practices arise from situated agents whose

beliefs and actions are informed by traditions and expressed in stories.
In every network, we can identify traditions, often embodied in rituals
and routines. Actors pass on these traditions in large part by telling one
another stories about how things are done, and about what does and
does not work. For example, British civil servants are socialized into the
broad notions of the Westminster model, such as ministerial responsi-
bility, as well as the specific ways of doing things around here; they are
‘socialized into the idea of a profession’, and learn ‘the framework of the
acceptable’ (Bevir & Rhodes 2006, ch. 7). Network governance is not any
given set of characteristics. It is the stories people use to construct,
convey, and explain traditions, dilemmas, and practices.
A third family resemblance is that the central state has adopted a less
hands-on role. Its actors are less commonly found within various local
and sectoral bodies, and more commonly found in quangos concerned
to steer, co-ordinate, and regulate such bodies. A decentred theory sug-
gests, crucially, that such steering, co-ordination, and regulation take
many diverse forms. In Britain, the pre-eminent example is ‘joined-up’
government as the Blair government seeks to devise policy instruments
that integrate both horizontally across central government departments
and vertically between central and local government and the voluntary
sector (Bevir 2005: 83–105).
A decentred theory provides: a grasp of the family resemblances that
contribute to a general characterisation of network governance; a dis-
tinctive analysis of change in networks; and an emphasis on the role of
storytelling in network management. Nonetheless, it denies any logic to
the specific forms that network governance takes in particular circum-
stances. It resolves the theoretical difficulties that beset the positivist
alternatives. It avoids the unacceptable suggestion that institutions fix
the actions of individuals in them rather than being products of
those actions. It replaces unhelpful phrases such as path-dependency
with an analysis of change rooted in the beliefs and practices of situated
agents. And yet it allows political scientists to offer aggregate studies by
using the concept of tradition to explain how they come to hold those
beliefs and perform those practices.
Notes
1. On the limits of both rational choice theory (as in Hertting, ch. 2) and histor-
ical institutionalism (as in Peters, ch. 3), see Bevir 2005: ch. 1. For fuller discus-
sion of a normative or ideational strand of institutionalism (as in Peters, ch. 3),
see Adcock et al. 2006. On the differences between post-structuralism (as in
Triantafillou, ch. 10) and our approach see Bevir 2004. On traditions and the
analysis of institutions see Rhodes 2006c. For a symposium in which we dis-
cuss our approach with scholars inspired by positions akin to rational choice,
new institutionalism, and post-structuralism, see Finlayson et al. 2004.
2. All quotes from Fleming and Rhodes 2005: 199–200. See also Fleming 2006.
Part II
Governance Network Failure
5
Governance Network Failure
The increasing and more systematic use of governance networks in

public policy and governance can be seen as a political response to the
many reports of government failure and market failure that emerged in
the 1970s and onwards. The attempts of central and local governments
to govern society through comprehensive planning programmes,
macro-economic fine-tuning and elaborate systems of bureaucratic
control were accused of being too costly and of failing to provide flexible
responses to emerging policy problems and new challenges. The neolib-
eral marketization strategy aimed to solve these problems by means of
privatizations, deregulation and the introduction of quasi-markets that
supposedly stimulated an efficient and flexible allocation of goods and
services. However, market imperfections, negative externalities and the
failure to provide proactive solutions to societal problems have revealed
the limits of the neoliberal revolution. In this situation governance net-
works have been seen as an attractive alternative, or complement, to
governance through hierarchies and markets. Although it is still subject
to political contestation, governance through negotiated interaction
between public and private actors is by many political decision makers
supposed to provide flexible and proactive solutions to problems and
challenges characterized by a high degree of complexity, uncertainty
and interdependency. The ever expanding discourse on partnership and
joint up government bears witness to the warm political embrace of new
forms of network governance.
Many people have great expectations that governance networks will
facilitate an informed, consensual and legitimate decision making that
will lead to responsive, just and tailor-made solutions. Nonetheless,
there are good reasons to expect that governance networks will not
deliver all this and that they are just as prone to crisis and failure as state
95
and markets. Indeed, there are many discouraging examples of gover-

nance networks that are either weighted down by political apathy, irre-
concilable conflicts and an unproductive group think. Neither is it
difficult to find examples of the ‘worst case scenario’ where governance
networks are initiated by politicians, controlled by public administra-
tors, and finally given no real influence on public governance. Surely,
there are also encouraging examples of flexible and proactive network
governance with a real impact on public policy making. But the problem
is that network governance relies on precarious social and political
processes and takes place in an uncontrollable political and economic
context. Hence, many things can go wrong and prevent effective net-
work governance. As such, it might be difficult to motivate the relevant
actors to participate due to the presence of high transaction costs and
small immediate chances for a single actor to gain real political
influence. It might be impossible to resolve the internal conflicts in the
network and build a sufficient amount of trust to allow the network
actors to reach common solutions that go beyond the least common
denominator. There might be an inadequate attempt to metagovern the
self-regulating governance networks due to the absence of a resourceful
and legitimate metagovernor, the failure of the metagovernor to under-
standing how governance networks can be governed, or political resist-
ance from the network actors to external interference. Finally, there
might be a discrepancy between the temporal and spatial horizons of
the government and the various governance networks. Hence, the pro-
tracted negotiation process in governance networks might clash with
the deadlines prompted by the election cycle that drives the actions of
elected politicians. Likewise, particular governance networks might not
operate at the same scale as the responsible political authority, or might
even have difficulties identifying the relevant political authority with
whom to negotiate its policy proposals ( Jessop 2002: 239).
The euphoric praise of the merits of governance networks has
prevented an understanding of the reasons for governance network fail-
ure and the conditions for success (Klijn & Koppenjan 2000: 117). There
have been many attempts to demonstrate the advantages of governance
networks, but the causes of failure have received little attention. In this
second part of the book we shall aim to compensate this neglect by
devoting three whole chapters to a theoretically informed discussion of
some of the key aspects of governance network failure and the political
and institutional conditions for success. Before looking further into the
various conditions for success and failure, we shall in this chapter
provide an overview of different theoretical approaches to analysing
Theoretical Approaches to Governance Network Failure 97
governance network failure. We shall see what the four theories of

network governance presented in the introduction have to say about
governance network failure in order to briefly assess the differences and
similarities. However, first we need to define more clearly what we mean
by failure.
Clearly, when we talk about governance network failure, we are
referring to an inability to provide effective governance through negoti-
ated interaction between a plurality of public and private actors. So
everything depends on what we mean by effective governance. We
should be careful not to define and measure effective network governance
in the same way as we measure the effectiveness of state and market
governance (Jessop 2002: 236). Effective government is defined as the
ability to transform substantial values and majoritarian decisions into
standardized policy outputs and problem-solving policy outcomes. This
will not provide a fair standard for measuring the effectiveness of gover-
nance networks since one the main ideas of network governance is to
formulate and implement public policy through negotiated interaction
that seeks to transform the resources, competences and preferences of a
plurality of actors who aim to provide tailor-made solutions to specific
cases. When it comes to market governance, effectiveness is defined as
the ability to provide low cost Pareto-optimal solutions at the basis of
procedures ensuring perfect competition. This also provides an entirely
unfair yardstick for measuring the effectiveness of governance networks
since in such networks economic competition is replaced by political
negotiations that at best can be expected to provide satisfying rather
than optimizing solutions. Nevertheless, if the effectiveness standards
pertaining to state and markets seem to be unfair, and slightly off the
point, when it comes to measuring the efficiency of governance
networks, they are not completely irrelevant. Hence, it is clearly a
problem if governance networks fail to define any common objectives
and reach some collective decisions that take us beyond the least
common denominator, or if they produce joint solutions that are either
overly costly or tend to shift the costs of the benefits offered to the
network actors to external groups and actors. While this is true, we
should nonetheless continue to look for relevant effectiveness standards
elsewhere in order to make sure that we measure the effectiveness of
governance networks in relation to their own particular strengths rather
than in relation to the inherent strengths of state and market.
It is consistently argued that the main virtue of governance networks
is their ability to formulate and implement policy solutions that are
both flexible and proactive. Flexibility refers to the ability to adjust
public policy and governance to the ever changing context, preferences

and experiences of the actors within a particular policy field. That a partic-
ular policy is proactive means that it aims to take the future opportunities
and constraints into account. Flexible and proactive governance can be
obtained through ongoing contacts and negotiations between informed
and empowered network actors, who might confront each other in
intense power struggles, but who are urged to listen to and engage with
each other, because of their mutual resource dependencies, the existence
of institutional rules, norms and incentives, and the proliferation of
public discourses that are hailing the social and political actors as
responsible partners in processes of joint governance.
Governance networks provide favourable conditions for flexible and
proactive governance, but there is no guarantee that network governance
will result in effective governance (Jessop 2002: 230). Flexible and proac-
tive network governance is first and foremost conditioned by the ability
of the relevant and affected actors to establish and maintain a relatively
well-functioning governance network based on active engagement and
trustful exchanges. In addition, the network actors must be capable of
identifying and defining relevant policy problems and determine their
interrelations and causes. Finally, they must be able to create a regulative,
normative and cognitive framework that facilitates problem-oriented
negotiations based on agonistic respect and supports the production of
well-informed and learning-based solutions in the face of persistent
conflicts and power struggles.
This broad definition of effective network governance, and the under-
lying conditions of possibility, provides a possible starting point for the
discussion of governance network failure. However, as we shall see, the
different theoretical approaches to network governance have different
criteria for assessing the success and failure of governance networks; they
offer different explanations of why governance networks sometimes fail;
and they provide different accounts of the conditions for success. Let us
take a closer look at some of the main theoretical strands in the literature
in order to provide an overview of the different accounts of governance
network failure.
Interdependency theory
Interdependency theory (Rhodes 1997a) tends to view governance

networks as a mechanism for interest mediation between a number of
autonomous, strategic actors who are mutually dependent on each others
resources (knowledge, innovative ideas, funding, formal authority, etc.)
in order to govern and regulate a particular policy area. The strategic

policy actors cannot solve the pressing problems that confront them on
their own because they do not have the necessary resources and capacities
and because other actors may block the solutions that they prefer.
Hence, the strategic actors are forced to engage in negotiated interaction
with other actors in order to exchange resources and capacities and find
jointly acceptable policy solutions.
It follows that the ultimate criteria for assessing the success, or failure,
of governance networks is their capacity for joint problem solving
through negotiated exchange of resources. Since governance networks
are often dealing with complex and vaguely defined problems that arise
in a conflict-ridden and uncertain terrain, problem solving is not a
question of doing the right thing that solves the problem once and for
all. Rather, it is a question of defining and solving a problem in a way
that is acceptable to the key stakeholders and do not harm those who
are not participating in the network and who might veto a particular
solution (Klijn & Koppenjan 2000: 117).
Formulating and implementing a satisfactory solution to unclear,
complex, and basically ‘wicked’ policy problems requires: 1. that all the
most relevant and intensely affected actors are included in the network-
based negotiations; 2. that the actors are willing to pool their resources;
and 3. that they can somehow agree on a common conception of the
nature of the problem, the range of feasible options and the most impor-
tant decision-making premises. There is no guarantee that these require-
ments are fulfilled. In fact, there is whole set of hindrances that must be
overcome in order to facilitate network-based problem solving. Let us
see how this can be done by looking at the three requirements in turn.
The inclusion of the relevant and affected actors might be hampered
by the closedness of governance networks. Schaap and van Twist (1997)
identify four different forms of closedness whereby particular network
actors, or the governance network as a whole, exclude other relevant
and affected actors. First, there is unconscious social closedness if the
rules, norms and procedures that regulate the access to a governance
network exclude certain actors from the network. Second, social closed-
ness is conscious if the network actors realize that some relevant actors
are excluded, but are unwilling to remove the political or socio-cultural
barriers for access to the network. Third, there is an unconscious
cognitive closedness to the extent that, owing to the discursive frame of
reference of the network actors, other actors are excluded from the
negotiations in the governance network. Finally, cognitive closedness
becomes conscious when the network actors reflect on the exclusionary
effects of their frame of reference, but are unwilling to change it in order

to facilitate a broader participation in the governance network. Whereas
unconscious social or cognitive closedness might be counteracted by
stimulating the reflexivity of the network actors and empowering the
weak and marginalized actors, it is more difficult to counteract conscious
forms of social and cognitive closedness. Social and cognitive fixation
must be broken up by means of reframing strategies that question the
legitimacy of interest-based exclusions, problematize the network
actors’ frame of reference, and provide incentives for decisions based on
a broader inclusion of stakeholders. However, it is not enough to open
up the closed and exclusive governance networks as some policy actors
will choose not to participate, either out of fear for losing their auton-
omy and identity through participation in public policy making, or
because they are so big and powerful that they will gain very little from
mutual exchange with other actors.
Resource pooling basically means that the network actors are willing
to share all their different resources in terms of formal authority, legal
rights, legitimacy, knowledge, experiences, innovative ideas, political
competences, organizational capacities, and financial means. However,
the problem is that the network actors tend to operate in an uncertain
social and political terrain, where they cannot be sure how the other
actors will respond to their actions. Uncertainty as to whether other
actors will act opportunistically when the occasion emerges will prevent
resource sharing due to the development of defensive, non-cooperative
strategies. This fatal problem is overcome by the development of trust,
which is defined as the stable perception of an actor A that another
actor B will abstain from opportunistic behaviour (Nooteboom 2002).
Although trust does not logically imply reciprocal action, it is unlikely
that trust based on non-reciprocal action can be sustained in the long
run (Klijn & Koppenjan 2004b: 83). A high degree of interdependence is
conducive for the development of trust since trust permits mutual
exchange of resources. Trust is developed on the basis of concrete
experiences with social and political interaction with particular actors,
but it might over time develop into a general norm-based expectation
that all actors will act in a non-opportunistic, or even reciprocal, man-
ner. If this happens in a particular governance network we will say that
it based on a high level of trust, which will tend to facilitate the pooling
of resources between the interdependent actors.
The last requirement that must be fulfilled in order to safeguard the
problem-solving capacity of governance networks is the provision of
a common framework for negotiation and compromise formation.
A common framework is not given in advance. In fact, governance

networks are characterized by the absence of a clear and formal consti-
tution that forms and guides the problem-solving process (Hajer &
Versteeg 2005). The network actors must themselves construct a com-
monly accepted framework for identifying and defining policy problems
and for crafting and selecting satisfactory solutions. However, there are
many hindrances for the creation of such a framework. As such, Klijn
and Koppenjan (2004b: 115ff ) draw our attention to five important
problems. First, the network actors might be caught in a ‘dialogue of the
deaf’ due to the presence of persistent conflicts between different frames
of reference that are perpetuated by selective fact-finding and research.
The cure for this problem is frame reflection, cross-frame debate and
joint fact-finding (Rein & Schön 1986, 1992).
Second, there might be a limited coordination between the network
actors either because they fail to recognize their mutual dependency or
because uncertainty increases the perceived risks of joint action. This
problem can be mitigated by increased interaction that might help to
reveal resources complementarities between the actors, and by develop-
ment of procedural rules that ensure compensation of those actors who
carry the costs emanating from the unintended outcomes of joint
action.
Third, the lack, weakness, or incompatibility of institutional rules and
norms might increase the transaction costs of cooperation and lead to
social and political fragmentation. The solution to this problem is
institution building and institutional reforms which are something that
often take a long time and require a strong political leadership.
Fourth, the lack of systematic metagovernance, due to persistent
conflicts between different would-be metagovernors, incompetence on
the part of the predominant metagovernor, or resistance from the
network actors, might lead to crisis, stagnation and fragmentation. This
problem can be solved by an authoritative decision that places
the responsibility for metagoverning a particular governance network
with a qualified and experienced political agency.
Last but not least, external events might in one stroke undermine
sedimented perceptions, procedures and power relations and leave the
network actors in a state of chock and bewilderment. A flexible redefin-
ition of the tasks of the governance network that aims to exploit the
new opportunities opened by the new events is crucial for reducing their
negative and disruptive effects.
As such, there are many things that can go wrong and lead to gover-
nance network failure and the only cure seems to be a conscious and
careful network management that aims to regulate the internal

processes and institutional framework of governance.
Governability theory
Governability theorists define governance networks as a relative institu-

tionalized arena for horizontal coordination between relatively
autonomous actors (Mayntz 1991; Kooiman 1993; Scharpf 1994, 1997).
The functional differentiation of modern society results in an increasing
proliferation and autonomization of social systems and subsystem and a
growing fragmentation of social and political actors who tend to think
of themselves as free and autonomous actors (Mayntz 1991). This makes
it more and more difficult for central government agencies to govern
society. However, the governability of society can be restored by the
formation of crosscutting governance networks that facilitate horizontal
coordination between a range of operationally autonomous actors.
Coordination between actors with different preferences, resources
and codes is made possible by the presence of interdependency (Mayntz
1991). However, mutual dependence between a group of actors does not
necessarily lead to the formation of governance networks. The develop-
ment of institutional rules, norms and procedures that help the
autonomous actors to recognize their interdependency, and act upon it,
is crucial for the formation of governance networks. A common set of
rules, norms and procedures is also decisive for the development of neg-
ative coordination, where the network actors aim to avoid stepping on
each others toes, or even positive coordination, where the network
actors modify their preferences and begin to develop common percep-
tions and preferences (Scharpf 1994: 45). Institutions reduce the
transaction costs and help to create and sustain trust among the net-
work actors, both of which will enhance the capacity for coordination
within governance networks.
This is important since ultimately successful network governance is a
function of the ability of governance networks to produce negative
and/or positive coordination through negotiated interaction (the only
difference between the emphasis on coordination in governability
theory and the emphasis on problem solving in interdependency theory
is that the former adopts a system perspective on societal governance
whereas the latter adopts an actor perspective). According to Mayntz
(1991: 16), the problem is not so much the presence of conflicts and
antagonisms between the actors, but rather their myopic indifference
to the negative externalities they produce in the course of action.
The production of institutional rules and norms helps to minimize

negative externalities and facilitates the formation of mutually acceptable
compromises that take the negative externalities into account.
However, in order for the horizontal coordination within governance
networks to solve the fundamental problem of the mounting ungovern-
ability of society, they must not only create a balance between political
demands and the available capacities and resources, but also provide a
balance between the need for flexible policy adjustments and the need
for order and stability (Kooiman 1993).
Governability is defined as a ‘permanent balancing process between
governing needs on the one hand and governing capacities on the other’
(Kooiman 1993: 43). Governance networks are successful to the extent
that they contribute to an effective and legitimate adjustment of
governing needs to the available capacities and of governing capacities
to the articulated needs. Political problems and opportunities are trans-
formed into governing needs when the social and political actors call for
action in relation to perceived policy problems. In principle, all
problems and opportunities can be turned into governing needs.
However, governance networks might help to reduce the articulation of
governing needs by mean of involving those who articulate governing
needs in the hard and troublesome problem solving process. At the same
time, governance networks might contribute to the enhancement of the
governing capacities by integrating the relevant and affected actors as
responsible co-producers of solutions and by pooling their resources.
The attempt to govern our increasingly complex, dynamic and diverse
society must take all this complexity, dynamism and diversity into
account, rather than simply ignoring it or declaring it unwanted
(Kooiman 1993: 48). Governance must be flexible in order to cope with
complex patterns of social and economic interaction, dynamic forces that
destabilize status quo, and a growing diversity of meanings and life-styles.
Flexibility is ensured by the integration of a plurality of autonomous
actors whose conflicting conceptions and strategies tend to make gover-
nance networks just as complex as the social and political reality they
attempt to govern. However, the complex interaction of autonomous,
strategic actors within relatively self-regulating governance networks
might conflict with the need for stable planning, steering and coordina-
tion (Kooiman 1993: 44f ). But governability theorists do not give up the
ambition of a smooth societal coordination on the basis of common
goals and ideas that ensure political stability and democratic legitimacy.
As such, they insist that governance networks must be metagoverned in
order to make sure that the flexible policy making, which aims to mirror
the complex and dynamic society, goes hand in hand with the need for
stable coordination in accordance with the democratic goals of society.
Integration theory
Whereas interdependency theory views interest–based conflicts as the

stumbling bloc of governance networks, governability theory perceives
the lack of coordination and the failure to balance needs with capacities
and flexibility with stability as the main sources of governance network
failure. However, the two theoretical approaches are in perfect agree-
ment when it comes to the question of what drives social and political
action. As mentioned above, both theories are based on what March and
Olsen (1995: 7 ff) define as a ‘logic of consequentiality’ as opposed to
their own favoured ‘logic of appropriateness’. March and Olsen claim
that social and political actors do not act on the basis of rational calcu-
lations of costs and benefits, but rather act on the basis of a situational
interpretation of the matching rules, norms and procedures and a
contextual conception of their identity which is shaped and reshaped by
the institutional framework.
March and Olsen’s definition of governance networks is based on
Mayntz’ definition that tend to emphasize the negotiated coordination
of interdependent, but relatively autonomous actors (Mayntz 1989;
March & Olsen 1995: 70). Surprisingly, March and Olsen fail to address
Mayntz’ implicit understanding of governance networks as a framework
for rational, though institutionally mediated, games between calculat-
ing actors and to emphasize the constitutive role of the institutionally
embedded logic of appropriateness that shapes the identities and inter-
actions of the network actors. The institutional aspect of governance
networks is highlighted by Powell and DiMaggio (1983: 148) who define
an ‘organizational field’ as the totality of public and private organiza-
tions and groups that operate within a certain area, but rush to empha-
size that organizational fields owe their existence to their institutional
structuration that involves the formation of patterns of interorganiza-
tional interaction, relations of dominance, shared knowledge and infor-
mation, and common perceptions of goals and tasks. Organizational
fields that are governed by negotiated self-regulation, rather than by
hierarchical command or market competition, are exactly what we in
this context will denote governance networks.
Although March and Olsen do not emphasize the institutional aspect
in their definition of governance networks, they clearly emphasize the
institutional basis of the political interaction that takes place within
governance networks (March & Olsen, 1995: 27ff). Governance

networks are deemed successful to the extent that they solve relevant
policy problems and exploit new opportunities, and their capacity to
produce such positive outcomes depends on their institutional
adaptiveness. The pace of societal transformations is increasing and the
formulation of adequate responses to new problems and challenges
requires a constant institutional adaptation of public policy and gover-
nance to the ever changing environment. Institutional adaptation is not
ensured by the ‘efficiency of history’ (March & Olsen 1995: 184 ff), but
is a result of institutional conditioned perceptions of the need for
changes and the desired and feasible courses of action.
Intelligent adaptation is based on learning, but learning is by March
and Olsen (1995: 197) conceived in anti-rationalistic terms as an insti-
tutionally conditioned experience of the past, contemplation of the
future, and imitation of others. Adaptation on the basis of experience-
based learning does not lead to the formation of ‘one best fit’ since it is
limited by the ignorance of the interplay between causal factors in the
present and in the future; hampered by conflicts between the actors
identities and the institutionalized logics of appropriateness; and prob-
lematized by the ambiguity of the preferences, norms and rules that are
defined within the different logics of appropriateness (March & Olsen,
1995: 197f). As such, experienced-based learning is a process with many
obstacles and pitfalls. However, ideally, the learning process runs
through three integrated phases: 1. experimentation based on variation
and risk taking; 2. selection and inference from experiments based on
socially constructed evaluations; 3. retention of learning in institutional
rules and procedures that ensure a tacit, collective memory (March &
Olsen 1995: 199 ff).
The presence of a plurality of autonomous network actors facilitates
the collection of a wide range of information about the problems and
opportunities facing the governance network. In addition, the interaction
between relevant, affected and interdependent actors might also ensure
a high degree of reflexivity in governance networks. As such, the prospect
of developing a high degree of adaptiveness in governance networks
through experience-based learning appears to be good. However, the
learning-based adaptiveness of governance networks might be impeded
by the lack of capacities for experimentation, the conservative identi-
ties of actors who want to preserve status quo, and the failure to resolve
the internal conflicts between the actors that struggle over the assess-
ment of experiments and the formulation of strategies for institutional
reform.
The lack of adaptiveness decisively reduces the chances that governance

networks will produce effective problem solving and exploit new oppor-
tunities. The cure for this problem is conscious attempts to develop the
adaptiveness of governance networks through conscious metagover-
nance (March & Olsen 1995: 223 ff). One possibility is to facilitate
experimentation by means of constructing the right incentives, aligning
motivations, and weakening strict accountability regimes and high
governance aspirations that both tend to penalize failure much more
than they reward success. Another possibility is to facilitate knowledge
production by means of improving patience, the validity and reliability
of accounts, memory and learning from others.
Governmentality theory
The governmentality theorists tend to view governance networks as a part

of a new governmentality. Governmentality is defined as a collectively
shared mentality, or rationality, that condition how we organize and pro-
duce concrete acts of government which aim to enhance the general
well-being of the population by regulating the conduct of individual and
collective actors (Dean 1999: 2; Foucault 1991a: 103). It consists of con-
tingently articulated forms of knowledge, calculations and technologies
that are drawn upon and combined in the actual governing of society.
The liberal governmentality has been transformed through the
problematization of the historically contingent governmentality pro-
grammes. Hence, the programme of laissez faire liberalism was prob-
lematized due to its failure to deal with the mounting problems of social
inequality, economic instability and market externalities. In response to
this problematization a new welfare state programme was developed.
The social welfare of the population should be enhanced through an
elaborate system of bio-political interventions based on statistical infor-
mation and expert knowledge and carried out through a combination of
comprehensive planning, bureaucratic rule and a host of disciplinary
and normalizing technologies. The neoliberal revolution in the 1980s
claimed that the welfare state was too big and inefficient, and called for
the expansion of the unfettered rule of the market forces. However, the
market had to governed and regulated, not in order to mitigate its detri-
mental effects on social equality and cohesion, but in order to reduce
the competition hampering effects of society and construct the individ-
uals as rational, risk-taking entrepreneurs (Dean 1999: 159 ff; Gordon
1981: 41f). Now, the failure of the neo-liberal governmentality program
and its tidal wave of privatizations, quasi-markets and free consumer
choice to reduce public expenditure and alleviate the burden of the state
has prompted the development of a new form of advanced liberal gov-
ernment (Dean 1999: 165f ). Advanced liberal government introduces a
new kind of reflexive governance that carefully considers how different
tasks can best be solved. Is it through the imperative command of pub-
lic authorities, through the free reign of the market forces, or through
the construction and regulation of the responsible actions of self-
regulating citizens, firms, communities, partnerships, and networks?
The latter option involves the mobilization of the knowledge, resources
and energies of a plurality of actors whose identities and free actions are
shaped by rules, norms and narratives that tend ensure conformity with
the overall objectives of governance (Rose 1999). The mobilization of a
plurality of public and private actors within relatively self-regulating
governance networks helps to realize the key ambition of advanced
liberal government which is to ‘govern at distance’ through the shaping
of the freedom of autonomous actors (Rose 1996: 43).
Governmentality theory has no intention to improve societal gover-
nance in general or network governance in particular. It is merely
concerned with analysing and problematizing how we govern and are
governed. However, the different historical forms of governmentality
may contain normative ideas of proper governance that can help distin-
guish between success and failure. Hence, echoing the classical liberal
concerns, the program of advanced liberal government is concerned
with the scope and economy of governance (Dean 1999: 19). The first
demand is that governance should reach and include ‘each and all’. The
population is divided into a number of target groups and all the indi-
viduals belonging to these groups must be targeted and subjected to
governance. As such, governance should be totalizing in the sense that
it must include every individual and the population as a whole.
However, governance must not be totalitarian in the sense of suppressing
the freedom that the acts of government aim to mobilize and guide in a
particular direction. On the contrary, governance should aim to increase
the individual capacity for free and responsible action. In other words,
totalization must be combined with individualization.
The second demand is that governance should be economic in the sense
that the objectives of governance should be realized through the
deployment of the least possible amount of resources. Governance must
economize with its resources and minimize the costs of governance.
Resources and costs are primarily defined in fiscal terms, but also include
the deployment of force and repression, which should be kept at a
minimum.
Network governance is successful to the extent that two classical

liberal demands are fulfilled. Ideally, governance network should be able
to meet the two criteria for success quite easily. Not only are governance
networks recruiting and mobilizing local actors who either organize, or
have a close connection or access to the individuals belonging to various
target groups. The local network actors also deploy their own energy,
knowledge and resources in order to adjust the overall objectives to the
local needs and to target the subjects that are constructed by, and
subjected to, a particular form of governance.
However, governance networks do not always produce a well-targeted
and economic governance. Although this is not something which is
discussed by the governmentality theorists, it is clear that there are
many obstacles to meeting the two demands. As such, the ability to
govern ‘each and all’ depends on the recruitment and mobilization of all
the relevant stakeholders. However, conflicts and lack of incentives
might prevent this. In addition, the problem is that it is far from all
target groups that are organized, or presented by particular organiza-
tions. Weak and diffuse groups with little resources are often difficult to
organize, or to reach through local organizations.
As for the demand for economic governance, there are two considerable
risks. First, it might take a lot of time, energy and resources to mobilize,
organize and manage governance networks due to the presence of
political apathy, conflicts and political resistance. Second, they might
become so strong in terms of their political power and influence that
they are capable of attracting much more resources than it actually
requires to solve a particular governance problem.
The attempt to counteract these problems and ensure a well-targeted
and economic governance in and through self-regulating networks and
partnerships involves a combination of political empowerment and
disempowerment of the social actors. It also involves the discursive
construction of solidaristic values and orientations that can counteract
the pursuit of egoistic network interests.
Similarities and differences
The four theoretical approaches all recognize the possibility that

governance networks might fail. They provide valuable input to the
discussion of the criteria for assessing the success and failure of gover-
nance networks and they help us to see how problems and obstacles
might be overcome. Interdependency theory and governability theory
both view governance networks as a means to produce a certain
outcome. As such, interdependency theory emphasizes the problem-

solving capacity of governance networks, whereas governability theory
emphasizes their contribution to horizontal coordination across institu-
tions, levels and actors. In both cases the achievement of the desired
outcome is, ultimately, conditioned by the ability of the actors exploit
their mutual dependence to overcome the externalities emanating from
their operational autonomy. The main difference between the two
approaches is that interdependency theory tends to see interest-based con-
flicts as the main threat, whereas governability theory is more concerned
with the problem of indifference and also emphasize the problems of
maintaining a balance between governance needs and governance
capacities as well as between flexibility and stability. The cure is the
development of generalized trust, a common set of rules and norms, and
a careful metagovernance.
Integration theory tends to be more concerned with the internal
processes of governance networks than with their capacity for problem-
solving. The network actors tend to develop a common and relatively
institutionalized logic of appropriateness that shape their identities and
govern their interaction. However, in order to provide good and sound
solutions to relevant policy problems, the institutional framework of
policy making within governance networks must adapt to changes in
the environment. Intelligent adaptation requires experience-based
learning that might be hampered by the lack of learning capacities,
cognitive closedness, and internal conflicts between the network actors.
Again conflicts seem to be a key obstacle to successful network gover-
nance, but conflicts are here thought of as something to be overcome
through processes of institutional socialization and integration.
Governmentality theory is in many respects the odd man out.
Successful network governance is neither assessed in terms of their
problem-solving capacity nor in terms the internal processes facilitating
problem-solving. Instead, governance networks are too be evaluated on
the basis of criteria articulated by liberalism that is conceived as a par-
ticular governmentality rather than an abstract philosophical doctrine.
Like all governance, the governance produced by the networked inter-
action of active, responsible and self-regulating actors should reach it
target without being too costly and without deploying unnecessary
force and repression. However, governmentality theory is not really
interested in evaluating the performance of governance networks, but is
merely concerned with revealing the contingent, and ultimately politi-
cal, character of the way we govern ourselves and others. Nevertheless,
it seems clear that the participation of all relevant and necessary actors,
which is also emphasized by interdependency theorists, and conflicts

and lack of incentives, which is also emphasized by governability
theorists, might prevent governance networks from governing ‘each and
all’. Internal conflicts and lack of incentives to self-restraint might also
prevent economic, low-cost governance. However, in line with the
governmentality approach it would be more appropriate to conceive
strategies of empowerment and disempowerment and discursive story-
telling as the proper cure for governance network failure.
The four theoretical approaches might point out different sources of
governance network failure and propose different ways of solving the
problems. However, the various sources of governance failure seem to
supplement rather than contradict each other, and the different cures all
seem to involve the exercise of some kind of metagovernance (see the
third part of this book).
The structure of Part II
Although the discussion above might give the impression, successful net-
work governance is not merely a question of facilitating the construction
of governance networks that are open, consensual and efficient. Indeed,
it can be shown that successful network governance requires a certain
degree of closure, conflict and legitimacy. As such, governance network
failure can more generally be seen as a result of the failure to balance
openness and closure, consensus and conflict, and efficiency and legiti-
macy. The need to counteract the development of fatal imbalances
between these key aspects of network governance, and the great difficul-
ties involved in attempt to strike the right balance, is discussed in
Chapters 6–8 that deal with each of the three dilemmas in turn.
6
Closure and Governance
Linze Schaap
Introduction
Governance and closure: an uncomfortably linked pair. For do

governance networks not guarantee openness? If so, then when would
networks tend to be closed? And if closure occurs, can it help to explain
governance network failure, that is, the ‘inability to provide efficient
governance through negotiated interaction between a plurality of
actors’ (see Chapter 5)? Or does the cause of governance network failure
lie within the definition itself? For a significant presupposition is pres-
ent within this definition; that is, governmental actors intend to steer,
but how can one steer through negotiations?
Insight into the closed nature of governance networks could help us
understand governance failure and how the primacy of governmental
decision-making itself can cause governance failure. Governance failure
may be the result of denying the underestimated tendency of social
systems such as governance networks to be closed, the idea that closure
is a negative feature of networks, or the assumption that governance
networks can be considered as instruments for steering intentions. This
chapter aims to address these questions. How can we understand the
openness and closure of governance networks, and how can we deal
with them? What kinds of closure exist? To what extent are openness
and closure necessary conditions for governance? And to what extent do
they contribute to governance failure? The insights applied in order to
answer these questions are drawn from theories on autopoietic social
systems (for example Luhmann 1984a; Schaap 1997). The analysis pre-
sented here therefore fits the so-called ‘socio-cybernetic systems
approach to governance’ (Rhodes 1997a: 50f).1 In that approach, society
is centre-less and the political system is one among many systems.
111
Matters of closure and openness are occasionally discussed in the

literature; however, there is little consensus on how to perceive closure
and openness. The two concepts are generally treated as mere features of
governance networks. Their exact value and weight can only be deter-
mined via empirical research. As a result, it remains unclear what exactly
closure and openness are and what factors cause or influence them. The
next sections will deal with these topics. I will present two different
stands on the openness and closure of governance networks: the first
focusing on the closure of networks and the second emphasizing the
open character of governance networks.
Governance networks: open, closed, or both?
In a large part of the governance literature, ‘closure’ does not appear to

be an issue. For example, Pierre and Peters (2000) pay no attention to it
at all, and neither does John in his book on local governance (John
2001). On the contrary, John only links ‘governance’ to notions of open-
ness and ‘government’ to closure. When ‘government’ was the dominant
governing pattern, ‘closed institutions of local politics permitted rule by
small, elite cabals’ (John 2001: 3). When ‘governance’ became a custom,
networks became more open (John 2001: 9). Bogason appears to be of a
similar opinion in his discussion of fragmentation, a typical governance
feature: ‘… one may see the state as intertwined with society and vice
versa, not as a closed entity directing social actors’ (Bogason 2000: 37).
Closure and governance would seem to represent two separate worlds.
But are such claims not overly optimistic? The shift to governance may
certainly lead to an increase in social and cognitive openness; that is,
improved access to policy-making for actors as well as interests and
ideas. However, one may wonder whether there are not also other,
allegedly less favourable, results. For example, under certain conditions,
could governance lead to the exclusion of groups and/or interests and
ideas? This might occur, since governance increases cognitive and social
complexity, thereby allowing more participants and aspects to be taken
into account, but also implying an increased number of normative
dilemmas and more uncertainty regarding policy outcomes.
There are authors who deal with the potentially closed nature of gov-
ernance networks. Rhodes and Marsh (1992: 247ff; see also Rhodes
1997a: 43f), for instance, distinguish between policy communities and
issue networks (using policy networks as the generic term). Policy com-
munities have several characteristics, including potentially limited num-
bers of participants and a conscious exclusion of some groups. This refers
Closure and Governance 113
to the social closure of policy communities. Closure at a cognitive – be it

normative – level appears to be another feature of policy communities, as
participants share the same basic values. Rhodes and Marsh further argue
that networks can never be wholly closed (1992: 259). For example,
changes in the environment affect the network. Such changes can be eco-
nomic, ideological, knowledge/technical, and institutional in nature
(ibid. 257). They assume that networks ‘contain, constrain, redirect, and
ride out such change, thereby materially affecting its speed and direction’
(ibid. 260). One could say that networks themselves influence or even
determine the kind of change actually taking place. If that is the case,
what is then decisive? What determines the containing, constraining,
redirecting, and riding out the change? If the conservatism of networks is
dynamic (ibid. 260), what makes it dynamic? What leads to the conser-
vatism? We seem to require an additional concept in this regard.
Most explanations of such difficulties in governance are sought in
actor-related features. Hanf and Scharpf (1978), for instance, point to
the distribution of power and resources in the network. Rhodes and
Marsh (1992) argue that actors’ self-interests may cause closure. The rel-
atively closed nature of policy communities, resulting in incremental
changes, favours ‘the status quo (it. Rhodes & Marsh) or the existing bal-
ance of interests in the network’ (ibid. 263). Mayntz (1987) and Mayntz
and Scharpf (1995, as cited in Kooiman 2003: 81–3) argue that self-
governing tendencies, increasing autonomy of networks, increasing
abilities to resist political governance actors, and the lacking political
will and administrative competencies to intervene and govern are the
main causes of steering difficulties, that is, governance failure.
However, we still require a more encompassing explanation for closure
and the openness of governance networks (see Schaap & Van Twist 1997:
62). The balance of power and resources in the network or actors’ self-
interest may very well be of importance, but we must determine where
interests arise; whether there are perceptual differences with regard to the
balance of power and resources; what consequences this will have; and
whether closure can be surmounted by interventions. Similarly, the
question of the relationship between the openness and closure of
separate actors and the network as such has hardly been dealt with.
A systems theoretical contribution?
As stated in the introduction, this chapter aims to shed light on the rela-
tionship between both openness and closure, on the one hand, and
governance on the other. Decision was also made to apply insights from
autopoietic systems theory. Although the closure of social systems can

be considered a more or less natural effect of increasing complexity, and
autopoiesis helps to understand the closure of social systems (Luhmann
1984a, 1990a, b; in‘t Veld et al. 1991), applying Luhmann’s theory on
autopoietic social systems can nevertheless represent a hazardous
operation.
A considerable number of authors mentioned autopoiesis in their
analysis of governance but ultimately decided not to go through with it.2
Nevertheless, the theory of autopoietic social systems is potentially use-
ful in understanding openness and closure, together with possibilities
and impossibilities of network governance (Schaap 1997; Schaap & Van
Twist 1997; Esmark 2003). I will test its usefulness in this chapter.
Luhmann’s theory on autopoietic social systems stresses the closed
nature of social systems. For understanding reality and gaining knowl-
edge of it, we must make distinctions (see the logic of distinction,
Spencer Brown 1972). By applying distinctions, we assign meaning to
our environment, but this is a paradoxical kind of openness: the
openness depends on the use of a closed distinction. When we use the
distinction ‘inside–outside’, we can divide everything we see in terms of
either inside or outside. We are open to everything, but, and this is the
paradox, only by using one distinction at a time. Every observation we
make, every meaning we assign, refers to our own distinctions. Self-
referentiality is therefore a feature of every observer; using distinctions
at the same time means being selective. We cannot perceive something
as ‘red’ when applying the inside–outside distinction. Openness and clo-
sure therefore go hand-in-hand. Systems are no longer either closed or
open, as they are both. The main question for systems theory, according
to Luhmann (1984a: 25), is how self-referential closure can generate
openness.
Society does not consist of individuals, but of communications
(Luhmann 1984a). Individual persons interact and communicate. They
select information, they select a specific form for sending the message,
and the receiver of the message selects a meaning, since he applies his
own distinctions for understanding the message. Communication is
necessarily a threefold selective process. In ongoing interaction, mean-
ing uncouples itself from the individual psychic systems. Social systems
do not consist of individual thoughts of psychic systems, but of com-
munications. Communication has an ‘autonomizing’ effect (Luhmann
1984a: 200; Kiss 1990: 97). Social meaning acquires a certain amount of
autonomy; social systems emerge, losing a one-to-one link to the mean-
ings assigned by individual participants. This will even occur in simple
conversations between a limited number of persons. Not every issue

‘fits’ the interaction, even if one of the participants wants it to. The
social system needs the environment for energy and information, but it
is not the same.
This is even more the case at the societal level, as such. Social systems
have emerged, which in Luhmann’s understanding (1984a) are com-
munication systems. They also create their own distinctions, and they
create boundaries in doing so, deciding what the system is and what it
is not. Which aspects of reality are meaningful to the system depends
on internally determined distinctions. The system, therefore, deter-
mines itself, its knowledge, its past, its future, and its goals. All social
systems are self-referential and self-reproductive; hence, they are
referred to as autopoietic (Luhmann 1984a). Functional differentiation
is a characteristic of modern society (Luhmann 1982). Functionally dif-
ferentiated sub-systems have emerged, such as the juridical, political,
economic, scientific and religious sub-systems. Autopoietic social
systems have a central code to which every observation refers. All
communication in which reference is made to the distinction ‘to have
or to not have’ is economic communication (Luhmann 1989), while
‘power or no power’ is the central code of the political sub-system
(Luhmann 1984b), and ‘truth or not’ the one of the scientific sub-system
(Luhmann 1990c).
Social systems continuously apply their own distinctions and create
and reproduce their own identity, which consist of values, a self-steering
programme and roles. To analyse system identities, the central questions
are the following:
a. Which social systems ( communication systems) can be distin-

guished and which distinctions do they use to assign meaning to
empirical phenomena?
b. What are the structures, that is, values, underlying a systems commu-
nication, self-steering programme and the roles being recognized in the
systems communication?
c. In what way are these systems closed?
The self-referentiality has a fundamental impact on the possibilities for

influencing social systems from the outside, as through political steer-
ing. Effective political intervention is difficult to achieve due to the
closed nature of the systems (Schaap & Van Twist 1997).
However, systems do influence or at least incite or irritate – one
another – since they are each other’s environments. Several intersystem
relations are at stake:
● Interpenetration (Luhmann 1984a: 289ff). Individuals (that is, psychic

systems) bring their own backgrounds, education, habits and so on to
the system. They can influence the interaction, while both systems
keep their autonomy. The results of interpenetration can be
temporary, or may last a little longer when new expectations arise; in
systems terms: new structures.
● Interference (or: structural coupling between systems: Luhmann 1984a:
441 ff; Teubner 1989: 110; Teubner 1991: 133). In this intersystem
relation, one system relies on other systems for information and
‘Sinn’. To put it differently: they expect parts of their environment to
remain unaltered. In other words, they make themselves dependent
on other systems for their development (Kneer & Nassehi 1991: 346).
Interference can take on various shapes (Teubner 1989: 111), for
instance persons playing roles in different systems or events leading
to disturbances in different systems.
● Linkages. Whereas interference is still a rather temporary contact,
linkages are more or less permanent. Social systems co-develop
via linkages. For instance, a contract links the economic and juridical
sub-systems of society. The same holds for formal organizations and
networks. Several sub-systems come together, their information and
sense-making must function alongside for a long time, dealing with
the same events.
Intersystem relations may develop, but they do not lead to less closed
characteristics of social systems. The interpretation and reconstruction of
the relation remains determined within each and every system. Systems’
autonomy remains unchanged. This is the case even with linkages. A con-
tract or a governance network between, for instance, a local authority,
social housing corporations and private investors (all of them formal
organizations, hence social systems) may be very relevant for the develop-
ment of a deprived area, each of them maintaining its autonomy. A formal
organization may (as it does) combine economic and juridical insights, but
such linkages will not influence the economic and juridical sub-systems in
society. A network (if considered an autopoietic system) contains elements
and insights from several sub-systems, be they juridical, economic, scien-
tific, religious, environmental, or political, the sub-systems they come
from will continue to be autonomous (see Koppen 1991).
A system’s self-steering program will repeatedly stand in the way of an
outsider wanting to govern the system. Government may want to
influence the internal policies of private organizations, but the organi-

zations’ self-steering programs will determine what kind of effects
government measures will have. The counter side of this situation is that
the same program may also cause a highly effective government policy.
The approach adhered to in this chapter may be linked to interde-
pendency and governability theories, as discussed in Chapter 5, provided
that resources and interdependencies are not understood as objectively
existing objects, but rather as meanings assigned in communication.
This means that there are no resources and interdependencies beyond
those perceived by social systems; ignored resources and interdependen-
cies do not exist. Coordination, for instance, as in governability theory,
does not mean making use of existing interdependencies when the lat-
ter are not perceived by the involved systems.
Governance networks and types of social systems
Luhmann distinguishes between three types of social (that is, communica-

tion) systems. First, there is society as such, consisting of all possible ‘com-
munications’. Society has been differentiated in functional sub-systems,
such as the juridical, economic and political sub-systems. The second types
of social systems are formal organizations in which a strong emphasis
exists on membership as a means of inclusion and exclusion. The third
type is (human) interaction, in which the communication is face-to-face
and very much ad-hoc. The latter type of system has been considered a
systems’ equivalent to networks. Esmark (2003: 7) states: ‘networks, i.e.
interaction systems?’ and, ‘the temporarily and spatially limited coupling
achieved by an interaction system can in fact be achieved without destroy-
ing the operational autonomy of the implied organizations.’ This is an
interesting attempt at linking ideas on network governance to systems the-
ory; however, it is not without difficulties. Interaction systems, as stated
previously, consist of immediate, face-to-face communications. In net-
works, however, there is much more communication. Networks may exist
for many years and communication may assume numerous forms and not
only face-to-face. Furthermore, even if we were to treat networks as
interaction systems, the potential closure problem would not vanish, as
interaction systems themselves are also closed systems.
In empirical research (see Schaap 1997, 2003), it therefore proved
useful to conceptualize a fourth type of social system: policy communi-
cation systems. This kind of system is based on policy distinctions (for
instance, strong-weak regional authority) that assign meaning to all
kinds of policy proposals and information. Such policy communication
systems strongly resemble policy communities (see Rhodes & Marsh

1992). As all autopoietic social systems, they do have a central code and
distinctions based on that, as well as values, a self-steering programme
and roles. Interaction is a feature of policy communication systems (as is
the case in autopoietic interaction systems), but not just face-to-face
interaction. They are based on membership (as are formal organiza-
tions), but membership is voluntary, that is, it may change overnight
and it is not formally regulated.
In the study of governance networks, it is an empirical matter to deter-
mine whether policy communication systems exist, what are their code
and distinctions, their values, self-steering programme and roles (see
Schaap 2003).
When it comes to understanding closure, as stated previously, the usual
suspects are actor-related features, particularly actors’ self-interest.
Important to note at this stage is that self-interest is not an objective and
predictable phenomenon that can be attributed to some actors. Instead, it
is to be treated as a role (that is, an expectation) some actors may play.
Studying governance networks means analysing the values, the self-steering
and the roles; that is, finding out how the system perceives individual
persons and their actions. They may be understood as individual human
beings, as the representatives of an organization, or as the delegates of it.
The policy communication system can attribute various meanings to
actors, varying from ‘provider of information’, or ‘having legal powers’, or
‘possessing vital financial resources’ to ‘final decision-maker’.
In the third section above, I asked the questions, ‘What determines
the containing, constraining, redirecting, and riding out the change? If
networks’ conservatism is dynamic (ibid. 260), what makes it dynamic?
What leads to the conservatism?’ The answer now is: the policy
communication system does.
Two types of closure
Governing closed social systems is an important matter; however, we must

first distinguish between different kinds of network closure. Considering
the literature on networks, distinction can be made between two kinds of
closure: social and cognitive closure. Another distinction runs between the
closure of separate actors and the closure of an entire network: closure
within networks or closure of networks (Schaap & Van Twist 1997).
Closure in the social dimension occurs when certain actors are
excluded from the interaction, for example because other actors fail to
appreciate their contribution or dislike their presence and attendance.
The number of potential links between actors in the network is deliber-

ately restricted, for instance in policy communities, as discussed by
Rhodes and Marsh (1992: 247ff ).
The entry of actors into the network (and the concomitant relative
closure in the social dimension) can be informally arranged, and it can
be made formally verifiable by, for example, introducing institutional-
ized membership as a condition of participation in interaction
(Hayward 1986: 16; as cited by Jordan 1990: 327). Governance network
closure in the social dimension, that is, social exclusion, can also result
from a conscious strategy or unconsciously applied rules.
In addition to closure in the social dimension, closure in the cognitive
dimension can also occur. Two forms of cognitive closure can be identi-
fied, that is, closure in the sense of an inability to perceive and closure in
the sense of an unwillingness to perceive. The latter is a conscious
strategy, whereas the former is not.
In the first form of cognitive closure, none of the actors in a network
have direct access to the reality ‘outside’. They must first perceive that
reality and assign meaning to it before it can play a role in their decisions.
For this purpose, actors have their own frame of reference (Rein & Schön
1986): a frame of reference that organizes their perceptions and thus
enables the interpretation of a complex reality. Frames of reference have
selective and regulatory effects that facilitate the ascription of meaning,
and this effect is a requirement for interpretation, since it is impossible to
fathom reality the way it ‘really is’ for actors in all its complexity. The fact
that actors by definition ascribe selective meaning (supported by their
own frame of reference) to reality results in cognitive closure. Actors are
then cognitively closed to those aspects of reality to which they do not
ascribe any meaning or to which they ascribe a different meaning.
Cognitive closure stemming from an unwillingness to perceive can
also occur when actors declare a particular line of approach to be out of
order. Phrases such as, ‘We’re not discussing that now’ or ‘You can’t con-
sider every angle’ are symptomatic of this. In contrast to the first type of
cognitive closure, this constitutes a (goal) conscious strategy to reduce
complexity.
Social and cognitive closure interfere with one another. Excluding
certain actors (social closure) opens the door for the promotion of an
exclusion of particular points of view (cognitive closure). Conversely,
downplaying (consciously or unconsciously) certain aspects of reality as
they are perceived (cognitive closure) can result in certain actors not
being involved in the interaction (social closure). Thus, the two types of
closure can reinforce one another.
This overview offers a number of distinctions: between the social and

cognitive dimensions of closure and between conscious and uncon-
scious exclusion. Social closure or exclusion means that actors are
excluded from the interaction, excluded from membership of the gover-
nance network. A network is closed in the cognitive dimension if and
when knowledge, information, ideas, or proposals are ignored and
denied access to the agenda. Both social and cognitive closure can be
conscious and unconscious in nature. Actors and networks can make
deliberate choices for denying access to either other actors or informa-
tion. Closure can also result from informal rules, habits, standard oper-
ating procedures – of which the actors or network are hardly aware.
One distinction that has only been implicit up to this point is
between actor and governance network. As stated previously, the net-
work literature often leaves open as to whether it is the closure of actors
or the closure of networks that is at stake. The various types of closure
are as likely to occur at the network level as a whole as at the level of the
actors within it. Actors are defined here as the individuals, groups or
organizations active in a network. The distinction between them is of an
analytical nature and can be defined empirically in a number of ways. It
is merely intended as an aid to analysing (steering problems resulting
from) conscious or unconscious, social or cognitive closure. When I use
the term ‘actors’, here it is understood that any further interaction on
their part, for instance, within their own organization, is not the subject
of research. The term ‘network’ indicates the opposite: here, it is pre-
cisely the interaction between actors and the impact that this interac-
tion has on governance that is to be highlighted.
Figure 6.1 can serve as a basis for diagnosing the problems that occur
in the management of and within governance networks.
Dimension Social Social Cognitive Cognitive

Actors’ Network’s Actor’s Closure Network’s
Closure Closure Closure
Unconscious 1 3 5 7
Unconscious Unconscious Actors’ inability to Networks’ inability
exclusion by exclusion by perceive to perceive
actors networks
Conscious 2 4 6 8
Conscious Conscious Actors’ Networks’
exclusion by exclusion by unwillingness to unwillingness to
actors networks perceive perceive
Figure 6.1 Types of closure

Figure 6.1 shows that we are dealing with eight forms of closure: two
types of closure (conscious and unconscious) in two dimensions (social
and cognitive) and at two levels (actor and network). In the following
sections, the causes of these forms of closure will be cursorily examined
(for a more extensive analysis, see Schaap & van Twist 1997).
Three explanations for closure
Distinguishing types of closure is one thing, but explanations are

required if we want to cope with closure.
Explanation is initially to be found in the supposed interdependent
nature of governance networks. Actors depend on each other’s resources
for the achievement of their objectives and the fulfilment of their
policies (Aldrich 1979; Benson 1982; Kickert et al. 1997). Despite their
interdependence, actors in a network are always relatively autonomous
and thus to some extent able to cut themselves off from the interven-
tion of other actors. Interdependence implies that each of the actors
possesses veto power. The governance of policy networks thus
constantly means operating in a field of tensions between dependence
and autonomy. The actors in the network will not automatically accom-
modate the objectives of one actor, even when that one actor is
‘government-related’ (Kiser & Ostrom 1982). One may expect veto
power to result in a number of conscious forms of closure, since it takes
a deliberative decision to use veto power (see Figure 6.1, forms 2, 6, and 4):
conscious social exclusion at the actor level, actors’ conscious cognitive
closure, and conscious social exclusion at the network level.
A second explanation of the closure of networks factor is the existence
of actors’ ‘frames of reference’ (Rein & Schön 1986), which function as
filters. Actors only perceive ‘the’ facts if and insofar as their frame
allows. This offers an explanation for the relative closure occurring
when there is a confrontation between different points of view, and
actors attempt to reach consensus by means of a ‘closing’ argument
based on ‘the’ facts. The filtering effect of the frames determines the
receptivity to new developments, as well as the actor’s explanation for
them. Information from other actors is not perceived in the way it is
intended but in the way it is interpreted by an actor, modified by the fil-
tering effect of the frame. Sometimes actors can only ascribe meaning to
totally new developments and facts by having a different frame. To
achieve such a ‘reframing’, however, the actor must at the very least
clearly understand why his present frame of reference is inadequate for
perception. This is not as easy as it seems, because in order to
understand this, the actor must see what he has been unable to perceive.
In order to be able to do so, he must first undergo ‘reframing’. Closure of
actors’ frames of reference will surely cause the following forms of
closure: actors’ conscious cognitive closure (form 5, ‘unwillingness to
perceive’), when actors willingly declare issues, opinions and arguments
out of the question; actors’ unconscious cognitive closure (form 5, an
‘inability to perceive’), since actors may not be able to understand the
specific issues, opinions and arguments, due to their frame of reference.
Finally, closed frames of reference may result in unconscious social
closure (form 1), when interaction with specific actors is thought to be
irrelevant, unnecessary or rejected without any reflection regarding the
reasons why.
There is a relation between those two explanations. Each actor
ascribes (selective) meaning through his own frame of reference and
bases his actions on this interpretation and perception. It thus stands
clear that an actor’s opinions about his own negotiating options or
those of someone else; of his own position of power or that of someone
else; every idea about his own interests or the interests and motives of
the other actors in the network; are inevitably the result of a process of
ascribing meaning; a process that cannot be isolated from that actor’s
frame of reference. The use of veto power is therefore a result of an
actor’s frame of reference.
Communication between actors cannot compensate for the closed
nature of frames of reference. Communication is not a process in which
one actor broadcasts a message received by another actor. Instead, three
selections of meaning are being made as a result of which social meaning
attains autonomy (see the third section above).
This leads us to the third factor causing closure. Due to the role they play
in the policy communication system (network), they must acknowledge
the existence of restrictive values and norms, customs and rules. They have
a social and symbolic dimension. The symbolic side concerns external
phenomena, such as traditions, ceremonies, rites, myths, heroic figures
and so on and so forth. Then there is the social side, the (underlying) rules
of interaction, which determine what is (not) correct in interaction
between actors; what is good and what is not; and what is appropriate and
what is not. In applying those phenomena, the network continuously
reproduces them. We may therefore conceive a governance network as an
autopoietic social system, more specifically as an autopoietic policy
communication system (see previous discussion in this chapter).
Applying this theoretical approach, one can understand the closure of
a network, apart from closure of individual actors. The closed network
policy communication system, then, will most probably lead to a number

of specific kinds of closure. First of all, it will result in unconscious
cognitive network closure (form 7, network’s inability to perceive), since
the network’s structure, self-steering programme and perceived roles will
exclude issues, opinions and arguments. It will also lead to conscious
cognitive network closure (form 8, unwillingness to perceive), when a
deliberate choice is being made in the network’s communication process
to deny excess to specific issues, opinion and arguments. Finally, it may
lead to unconscious social network closure (form 3, unconscious exclu-
sion by networks) and conscious social network closure (form 4, con-
scious exclusion by networks), when actors are denied access to the
interaction.
The relations between explanations for closure
It was earlier stated that the possession of veto power is not an objective
fact independent of actors’ perception, but on the contrary, merely
attains meaning for actors through the frames of reference via which
they identify that veto power in order to subsequently be able to attach
consequences to them. Actors’ positions of power and the way in
which they ‘frame’ reality influence each other, but the frame level is
the most fundamental of all.
Extending this argument, I now state that actors’ frames of reference
develop in communication processes, processes embedded in the
network policy communication system. The frames and the policy
communication system influence each other, but the system must be
considered to be the most comprehensive. Actors’ interpretations can,
in any case, only become socially relevant the moment when commu-
nication about that perception becomes possible, thus at the moment
when the interpretation between actors becomes nameable. In this
sense, the network policy communication system forms a condition for
framing within networks. Furthermore, the system not only describes
what actors perceive but also has a structuring effect: it also constructs a
reality. By ‘articulating’ reality, certain aspects of reality are highlighted
while other aspects are glossed over and thus fall outside of the
(intellectual) field of vision.
We can now schematically represent the cohesiveness between the
three (interfering) clarifications for closure (see Figure 6.2).
Considering the network policy communication system as the main
and encompassing source for closure means that cognitive kinds of
closure are the more influential ones, not so much the social ones.
Closure
through the possession of veto
power
Embedded in the closure of actors' frames of reference
In interaction with the closure of the network policy communication system
Figure 6.2 Relationships between clarifications for closure
In other words, cognitive closure represents a natural phenomenon in

social systems and may possibly lead to social closure. In the remainder
of this chapter, slight emphasis will be placed on cognitive closure.
Governing closed networks?
Because of the closed nature of network policy communication sys-

tems, effective intervention is difficult to achieve. This holds for all
systems wanting to govern, including the political subsystem.
Political power may exist, but is no more than the language of the
political subsystem itself (see Luhmann 1981, 1990b). Power is noth-
ing but communication, based on the communication code ‘power or
no power’; an actor is powerful only when this code is being used.
When communication is based on different codes, for example justice
or no justice, governmental activities assume different meanings.
Policy necessitates reflection; politics should take into account the
closure of social systems and their autopoietic identity, forcing them
to make an image of themselves as they are observed by other systems.
Policies may then stand a chance to become effective. Consequently,
making an analysis of the environment is a necessary condition for
governance.
It then becomes obvious that talks of effective policy-making, effec-
tive governance and policy failures tend to be hazardous. Perceptions of
what is effective or what is a failure differ. Actors may co-operate for dif-
ferent reasons, meaning that not all outcomes will be acceptable to all
participants. If an agreement in the negotiations between those partici-
pating in the network is reached, then the question becomes whether
interested but not involved outsiders may consider that result to be
acceptable. In other words, the self-steering of a closed network policy
communication system can be effective as far as the network itself is
concerned, but it can result in something another system will definitely

assign a negative meaning. For pragmatic reasons, however, we can decide
to take one system’s point of view, for instance the political subsystem
wanting to intervene in developments in other subsystems.
In the remainder of this chapter, I will explore tools for the
governance of networks and thus for dealing with the closed nature of
network policy communication systems. The three explanations for
closure discussed previously (actors’ veto power, their frame of refer-
ence, and network policy communication system) will be of use. The
emphasis will be on finding ways for dealing with the three types of
explanations; not so much the eight types of closure. Being able to deal
with the specific forms of closure will be the logical result of being able
to cope with their causes.
It must be kept in mind that the first and second causes of closure are
embedded in the third. Governing a network policy communication
system may take the shape of targeting veto power and it may even be
effective. However, this effectiveness will depend on the actors’ frames
of reference. Reframing those frames may be another goal of governance,
but then again, the frames depend on the network policy communica-
tion system’s self-steering program. In the remainder of this chapter,
I will leave it at that and distinctively discuss how we can deal with the
three causes of closure.
Governing veto power?
The veto power of actors, so it was concluded, is one of the causes of

conscious forms of social closure among actors and in networks and
of actors’ cognitive closure. Attempt at breaking through these forms of
closure can only be successful if this consciously-used veto power is
accounted for. If this is not done (or done incompletely), then any
attempt at forcing openness will prove to be no more than a superficial,
symptoms-related treatment.
In the literature, several strategies and instruments can be found for
handling veto power. Kickert and Koppenjan (1997: 46ff) provide an
overview and distinguish between ‘game management’ and ‘network
structuring’. The first is meant as the management of the interactions
within an already-existing governance network, the latter to influence
the structures and/or self-steering program of a network or to create a
new network. A governance network manager might decide to apply
game management strategies as well as network structuring for breaking
through the closure resulting from the use of veto power. Kickert and
Koppenjan (ibid.) mention the following.
The first strategy is (selective) network activation, initiating or stimu-
lating interaction between some of the participants in the network; it is
aimed at solving specific problems. This way, if a certain actor is
expected to use the veto power and thus create a barrier in the negotia-
tions, the other involved actors can disregard the barriers. Of course, this
can only be done as long as the resources of that specific actor are not
exclusive. If the actor possesses necessary and unique resources,
the other actors must find a way to deal with this difficult actor.
Selective network activation can still be of use. It may be necessary to
restructure and expand the network slightly further and to include more
issues and actors. That way, a situation may eventually result in which
the difficult actor himself becomes dependent on resources possessed by
other actors. Network expansion is not without risks, or course; it may
lead to increased complexity and additional kinds of veto power.
The second strategy involves arranging interaction, creating commit-
ment amongst some of the participants, sometimes in a formalized
manner. This is a useful strategy once actors already reach some kind of
agreement, some kind of mutual understanding, or even a shared inter-
est goal. Commitment then means that the actors promise to refrain
from using the veto power, or to use it only under certain specified
circumstances. The commitment can have a formalized nature, but that
is not always a necessity. In some situations, especially when actors deal
with one another in multiple processes and/or networks, actors cannot
possibly withdraw at the price of being unreliable.
When selectively activating the network or arranging interaction does
not lead to agreement and actions, a third strategy might be of use.
Brokerage is an option – that is, bringing problems, solutions and actors
together – since they develop independently, as policy-making and
decision-making are fragmented and non-linear processes. A broker may
find solutions to problems that the involved actors themselves had not
thought of. In certain circumstances, an actor considering the use of
veto power may discover the existence of alternative options. Brokerage
is no guarantee, but it may prevent the emergence of interaction
barriers. Brokerage may also stimulate the manifestation of conflicting
proposals and thus the creation of advocacy coalitions (Sabatier 1988,
in: Kickert & Koppenjan 1997: 49).
A fourth strategy is to facilitate interaction and create procedures for
ongoing interaction in such a manner that consensus is within reach.
Facilitating then means creating procedures for ongoing interaction,
discussions, negotiations, and decision-making. Actors must bind them-

selves to those procedures; if not, facilitating will not prevent the use of
veto power. Whether agreed-upon procedures will actually do so depends
on the kind of procedures. As long as creating a blockade or stepping out
of the interaction represents acceptable options, closure of the first kind –
by the use of veto power – remains a possibility.
The last game management strategy is mediation and arbitration, to
be applied in cases where interaction is difficult or even blocked. This
strategy is an option in those cases where an actor has already used veto
power or at least threatened to do so. An independent actor from within
or outside the network may undertake several mediating activities in
order to maintain interaction. If insufficiently successful, binding or
non-binding arbitration may represent the last resorts. Here again, there
is no guarantee for success. Even when actors commit themselves to
arbitration and to accept the verdict, it may still appear difficult to
enforce the decision (Kickert & Koppenjan 1997: 51). Public actors in
particular cannot easily be forced to act at will, potentially their public
duties stand in the way. An arbitration decision may not be legally
binding per se.
In addition to game management strategies, Kickert and Koppenjan
(1997: 51ff) describe ‘network structuring’ as an option. I have already
mentioned the option of expanding the network and introducing new
actors to the interaction in order to decrease the uniqueness of the
resources of an actor threatening to use veto power. Network structuring
is a more general option than just that. A network manager may con-
sider changing the network as such by the introduction of new actors,
by pulling out involved actors, or by changing the dependencies
between actors. Those measures can only be undertaken if the network
manager has the required capacities. As far as public actors are con-
cerned, this may help. Central government, for instance, can influence
the statutory position of local and regional authorities and their legal
capacities. It can also attempt to modify the legal and financial capaci-
ties of private actors; however, it cannot be certain that the results will
be as expected.
A further option is what might be referred to as ‘management by
chaos’, that is, to disturb the network (Kickert 1993), since ‘(B)y drasti-
cally disrupting the environment of the system, it may be brought out
of balance to such an extent that the leap to a new internal balance
becomes possible’ (Kickert & Koppenjan 1997: 53). Fundamental
changes in legal and financial provisions, for instance, could be decided
upon. They allegedly disrupt the environment in the systems. However,
it is questionable whether the new balance would lead to more preferred

outcomes. The opposite may just as likely be the case.
Dealing with veto power, we may conclude, is no simple task. A
number of strategies are available to prevent the use of veto power;
however, none of them can guarantee success. In addition, mediation
and arbitration are options when (the threat of) veto power is already at
stake. Even then, there is no certainty of bringing the veto-wielding
actor to her knees.
Actors within the network ultimately have two options: the first is to
simply accept that veto power is a continuous threat and to accept the
use of it. This will most likely result in fewer preferred situations for
many actors. An alternative option is to simply give it a try; that is, to
apply as many strategies as possible in order to prevent the use of veto
power and apply mediation and arbitration as soon as blockades are
actually created. Trial and error will demonstrate whether this works.
Governing closed frames of reference?
Dealing with closure because of veto power proved to be difficult. For

governing a governance network, it may be necessary to delve a little
deeper and deal with individual frames of reference. Frames of reference,
as stated previously, are the sources for the use of veto power.
Again, the strategies for game management and network structuring
may be of use (see Termeer 1993; Termeer & Koppenjan 1997). Whatever
the strategy may be, the keyword for influencing the frames of reference
is reflection. The network manager can selectively activate the network
and initiate or stimulate interaction between some participants in the
network. That way, actors with different frames of reference will be con-
fronted with different perceptions and they may feel stimulated to
reflect on their own perceptions. Actors can arrange a certain amount of
shared views and perceptions, or at least know each other’s frames of
reference, to some extent. Brokerage, facilitating interaction, mediation
and arbitration may support this process. The same holds for the
network structuring strategy. All of these strategies can be applied in
such a manner that they influence the frames of reference. Brokerage,
for instance, may be aimed at furthering a common language (Termeer
& Koppenjan 1997: 92), as actors with contrasting professional back-
grounds do not automatically understand one another. A mediating
actor can strive for mutual understanding between actors.
One of the intervention strategies occasionally suggested in this con-
text is that of ‘reframing’. In this strategy, the intention is to have actors
reflect on their own frame of reference by making them conscious of the

existence of alternative frames of reference and other possibilities for
assigning meaning to reality (Levy & Merry 1986: 99). There is little
point in such an intervention if actors consciously declare those alter-
native frames of reference to be out of order and refuse to recognize that
other perceptions are possible regarding the nature of the problem or
what is to be done; an opinion they might support by the use of veto
power. Calling in external advisors, recruiting new personnel, appoint-
ing a committee, introducing new ideas, stimulating reflection or
arranging reframing in order to break through the closure of frames
of reference can be successful. The problem is that the potential success of
reframing depends on the same closed frames of reference. If a frame of
reference wilfully closes itself from alternative perceptions, no remedy is
possible. The actor apparently treats the perceptions as norms and pre-
vents learning (see Luhmann 1984a: 436ff). If the frame of reference
leaves open the possibility of learning, reframing is an option.
Again, a prediction of the results of the application of one strategy or
the other, or of the application of strategies as such, cannot be made.
Actors already interacting for a long time may well enough know each
others’ views and perceptions and, thus, each others’ frames of reference.
They will know what can be done or not. Things will be more compli-
cated in newly established or issue networks. Actors have limited mutual
experiences and will simply have to wait and see. Increasing interaction
and discussion will most probably be useful. The greater the discussion
and interaction, the more likely actors get to know the frames of refer-
ence of other actors and their own. Simulations may assist the reflection
process.
Governing closed policy communication systems?
Finally, we must reflect on the closure of the policy network system, as

this is the most encompassing level in the governance network. The
policy network system determines whether the use of veto power is
allowed and whether frames of reference can be formed and altered. The
policy communication system of a network contributes to the inclusion
and exclusion of actors and views in the network. Actors can distance
themselves from the dominant policy communication system, which
may lead to reflection at that level, but such distancing does have its
limits. If the actors entirely fail to comply with the game rules of the
network, they are literally sidelined. In this sense, the closed policy
communication system functions in a structuring manner: through the
policy communication system of the network, actors are assigned a

meaning in the communication that they are not entirely able to
determine or control.
Steering closed social systems, hence governing governance networks
with closed policy communication systems, requires reflection. Actors,
including governments, must take into account the closure of social
systems and their autopoietic identity. If they do so, their awareness of
the closed nature of the social environment of government might
increase. Governments may become aware of the functional differentia-
tion of society and thus of their non-hierarchical position. Other social
systems are equal to the political sub-system that governments may
want to govern, but governmental will as such is an insufficient tool for
steering society. It is the era of the ‘Entzauberung des Staates’ (Willke
1987a); governments can no longer act as if they were fairy tale kings
and queens. Other social sub-systems have their own perception of
government and its actions. Any governments would therefore be well
advised to make an image of itself as it is possibly observed by other
systems. Governmental steering and policy measures may then stand a
chance of becoming effective. Consequently, analyzing the environment
is a necessary condition for governance.
A first, rather simple, possibility is to make use of the self-referentiality
of the juridical sub-system. Government may for instance change con-
tract law and will thus stimulate changes in other systems (Teubner
1989); or it may create provisions for interest groups to participate in
judicial procedures or it may create the duty for authorities to consult cit-
izens or interest groups. In doing so, the self-steering program, the policy
communication system or other social systems will have their doing.
This option could be interpreted as network structuring at the meta-level.
A second, more complicated, method for effective governance is
governing in such a manner that the changes of policy failure are
decreasing, necessarily takes the shape of context steering (Teubner &
Willke 1984; Willke 1987b; 1991) or inciting governance (Schaap 1997).
A number of options are available, keeping in mind that applying these
options means inciting other closed systems, not getting what you want.
Political intervention aims at coupling the autonomous self-steering of
societal systems with political wishes, using diverse steering instruments.
Moreover, political intervention no longer aspires to regulate society at
large, but instead offers policy options; the effectiveness of those options
increases if a choice between a limited set of options is mandatory.
Providing options means that the autonomy of the sub-systems to be
governed remains intact; it also means providing stimuli for choosing.
Options may concern internal regulations for private companies, for

contracts, and so on. Options may also concern governmental structures,
as was the case in the UK when the Blair Government obliged munici-
palities to decide on system reform and choose among a number of govern-
ment models (Pratchett 2002: 333ff; Stoker 2002: 35; Sullivan 2003: 49ff).
An actor providing options leaves the choice to others; however,
options may be accompanied by requirements or by legal and financial
provisions. A closed policy communication system in a governance
network may thus be triggered or seduced to open up and allow for inter-
vention. To be offered a choice among a limited number of options can be
attractive, as it diminishes the noise in the environment of the system.
Some concluding remarks
Dealing with closed governance networks, understood as closed sys-

tems, is not easy; that is for sure. We may be able to analyse and catego-
rize types of closure; even find explanations. To actually govern closed
systems is yet another thing.
In this chapter, I explored the potential contribution of Luhmann’s
theory of autopoietic social systems to the discussion on governing
governance networks. This enabled me to distinguish between several
kinds of closure, providing a heuristic tool for analyzing closure. A
second contribution was the analysis of explanations for closure. And
thirdly, the autopoiesis concept provided a basis for finding alternative
strategies for the governance of closed systems.
The core of the governance model discussed in this chapter is the
fundamental acceptance of the autonomy and self-steering of closed
social systems, governance networks with closed policy communication
systems among them. Directly influencing such kinds of systems is
impossible at the cost of destroying them by legal or financial means. In
this model, the autonomy – and yes indeed, the closure – of systems is to
be cherished. They are the better source for dynamics, for progress, for
development. Systems may appear immobile or conservative, but that is
in the eye of the beholder. The conservatism will most surely have a
dynamic character, depending on the internal self-steering program of the
system. Closed social systems continuously change, as long as the envi-
ronment provides stimuli for assigning meaning and learning. That very
circumstance, external stimuli leading to internal changes, is the better
option for governmental actors to influence governance networks.
Actors with steering wishes and intentions had better wait a little,
however. If governing means making decisions regarding policy
content, executing the policies and forcing society and its networks to
oblige, then government runs a risk. It may diminish or even destroy the
self-steering potential of society, which will result in total dependence
on government. In some sectors of society, this may already be the case.
This does not mean that government will have to retreat, to the con-
trary. That would lead to fewer stimuli for systems, hence fewer oppor-
tunities for development and learning.
Instead of a retreat or traditional governing, an alternative is at hand.
Governments may increasingly define their tasks as network game man-
agers and the actors responsible for network structuring. As such, they
should leave the choices regarding the content of decisions to others. Even
then, they would be entitled to suggest solutions and possible policies.
In other words, government may become the actor steering the con-
text of networks, offering stimuli – partly procedural, partly consisting
of choices between varying content. The better option for politics is a
policy of options.
Notes
1. Rhodes apparently interprets the socio-cybernetic approach as an actor-
oriented approach, discussing the multiplicity of actors, interdependencies
among them, and so on. This interpretation can be challenged, as Luhmann
(who might be considered the founder of the approach) hardly ever spoke of
actors, but rather of social systems, especially in his later works on autopoiesis
(see Luhmann 1984).
2. Mayntz (1987: 101ff ) did recognize Luhmann’s basic analysis of societal dif-
ferentiation and the operational closure of societal subsystems; however, she
denied that the characteristics of the features of modern societies would lead
to governance problems (as discussed previously), since she considered self-
interest to be the main source of closure. Scharpf (1988) limited the
autopoiesis analysis to a certain kind of social systems (those that heavily
depended on government for their financial means). His solution to systems’
closure was to adapt the governance instruments to the specific kind of system
that was to be governed. Pierre and Peters (2000: 39) considered autopoiesis to
be a potential approach to the analysis of governance, but refrained from
exploring the potentialities. Bogason (2000: 74), referring to Luhmann’s work,
acknowledged the fact that society consisted of ‘subsystems which basically
operate on their own; there is a whole language structure linked to each
subsystem making it operate perfectly on its own, but on the other hand
preventing it from communicating perfectly with other sub-systems’.
Nevertheless, he did not discuss this theoretical approach any further, because
‘it operates at a level of abstraction that is too high for the analysis of collec-
tive public action in the locality’. Kooiman also refrained from an analysis of
autopoiesis, sometimes without saying why (for example 2003: 124).
7
Consensus and Conflict in
Policy Networks: Too Much
or Too Little?
Joop F. M. Koppenjan
Introduction
One of the criticisms of theories on policy networks and governance is

the lack of attention devoted to power and conflict (Barns 1997).
Indeed, reading through the literature on policy networks – and
especially on policy network management – one is struck by the
approach to the network processes and management challenges as a
politically neutral and predominantly managerial activity (for example,
Agranoff 1986, 2003; O’Toole 1988; Mandel 1990, 2001; Glasbergen
1995; Kickert et al. 1997). Key concepts are interdependencies and coor-
dination. Due to interdependencies between a diverse set of public and
private actors, public decision making, implementation and service deliv-
ery is about mobilizing, facilitating and guiding collaboration processes.
These processes can be characterized as problem solving: putting pieces
of a puzzle together, realizing package deals that provide win-win situa-
tions: solutions that benefit all of the involved actors and society as a
whole (Scharpf 1978; Dery 1984).
While attention to conflicts and power relations is almost absent, the
opposite is true for the concept of consensus. Network governance pri-
marily appears to be a consensus-building method: mobilizing resources
and creating support to solve a serious societal problem or deliver a
highly appreciated public service. The implication being that within
network settings, there is a serious lack of consensus, though consensus
itself is something good. Consensus is worth striving for, and perhaps
even something we cannot have enough of.
133
The emphasis on consensus building has rendered policy networks

and network theory the target of criticism from both academics and prac-
titioners (Tatenhove & Leroy 1995; Barns 1997). Striving for consensus
and the accompanying preference for ‘softer’ steering instruments, such
as covenants and consultation, lead – so runs the criticism – to protracted
deliberation processes that consume excessive energy and money but
ultimately produce weak compromises, deadlocked decision making or
non-implementation, in addition to hindering administrators and
politicians from doing what they have been hired for: shouldering their
responsibilities. Instead of network processes, there is a need for positive
action, leadership and a clear and transparent division of responsibilities.
This criticism of the policy network approach as a normative model
runs somewhat parallel to the characterization of policy networks found
in the slightly more dated literature on policy networks (Ripley &
Franklin 1987; Jordan 1990; Marin & Mayntz 1991; Marsh & Rhodes
1992). According to this literature, networks are characterized by an
overdose of consensus, thus excluding important interests and consid-
erations and producing beneficial advantages for established oligarchies.
Networks do not call for consensus building; rather, there is a need for
confrontation and conflict capable of breaking through existing network
structures and reducing their perverted outcomes. In this contribution,
we aspire to shed more light on these contradictory views on the role of
consensus and conflict in policy networks and network theory. How is it
possible for so many different perspectives to exist regarding the same
phenomenon? In addition to this question, we raise a normative issue:
how should we value the occurrence or absence of consensus and con-
flict? Ultimately, we are interested in the implication of the answers to
these questions for network governance.
In order to address these questions, we will first explore the concepts
of consensus and conflict (see next section). We will then proceed to
take a closer look at the two contrasting approaches to policy networks
in network-related literature mentioned above. In the following section,
an overview of the literature presenting policy networks as forms of
intensive cooperation with a surplus of consensus is given. Next the
literature that presents networks as multi-actor settings with a lack of
consensus is described. In the penultimate section, the two perspectives
are compared and contrasted, and we examine how it is possible that
they can present such different images. A meta-position is explored,
consisting of a contingency approach to the management of the
consensus-conflict dimension. In the final section, the implications of
these observations for network governance are discussed.
Consensus and Conflict in Policy Networks 135
Consensus and conflict: an exploration of

two ambivalent concepts
Consensus means agreement, often understood as agreement in the

field of more fundamental values. Understood in this way, the concept
of consensus connects to that of deliberation: the process of an
exchange of opinions upon which consensus is established (Butler
1987; Elster 1998); however, this is not the only interpretation of the
concept. Apart from agreements on norms and values, consensus may
also refer to problem formulations and objectives, solutions or policy
measures, working methods, technologies or interaction rules, and so
forth. The most minimal consensus can concern an ‘agreement to dis-
agree’, making it possible for interdependent actors who disagree about
almost everything to coexist without fighting. Also, because actors
may be dependent upon one another as regards more then one issue,
agreement on one topic may be accompanied by agreement or dis-
agreement on another. Furthermore, actors may consider the issue
with regard to which they have developed consensus as vital or less
important. In other words, consensus may vary in scope, level and
intensity.
Conflict is not exactly the opposite of consensus. The absence of agree-
ment does not imply controversy per se. After all, interests, perceptions
and preferences may diverge. Conflict assumes differences of opinion
between parties regarding, for instance, interests, views and preferences.
They are relevant in so far as they prevent actors from realizing their
goals. At worst, a conflict may consist of parties mobilizing their
resources in an all-out struggle aimed at the destruction of the opponent.
Conflict may be about the division of scarce resources but may just as
well be caused by an incompatibility of character as by ideological
differences (Coser 1956; Cobb & Elder 1983). As with consensus,
conflicts may vary in scope, level and intensity.
The positive and negative functions of consensus

The concept of consensus is ambivalent. On the one hand, it has posi-
tive connotations. A certain degree of consensus is a necessary precon-
dition for cooperation between different parties (Scharpf 1978, 1997;
Susskind & Cruikshank 1987). Consensus can also denote the degree to
which parties form part of a community: that they have not merely
developed common ground for action, but also a shared identity.
Instead of being an essential precondition, consensus can also be viewed
as something worth striving for unto itself.
On the other hand, consensus may have drawbacks. For example, the
consensus democracy in the Netherlands has been judged very differently
over the years. Not so long ago, the harmonious cooperation between the
public sector, employers and employees referred to with the term ‘polder
model’ was seen internationally as an ideal, shining example; currently, it
serves as the explanation for lagging economic performance: consensus
standing in the way of innovation and competition. In education in par-
ticular, the Dutch consensus culture is said to nip any form of excellence
in the bud: anything above average gets cut down.
A high degree of consensus may mean that there is no room for
deviating perceptions, preferences and interests, no room for innovation,
no room for competition and excellence.
Janis (1982) has examined the consequences that excessive consensus
can have for group decision making and shown the extent to which the
quality of the decision making suffers as a result. In small groups marked
by extreme consensus, specific problem definitions and their solutions
become accepted and inviolable at an early stage. Information is biased
and searching behaviour is suppressed; important goals, interests, effects
and alternative options are insufficiently examined; critical voices are
quashed. This leads to choices for solutions being made prematurely,
before the effects and risks of those choices have been adequately
examined (see also Hart 1990).
At the level of policy communities and entire nations, consensus
among a majority of the population may lead to the neglect or suppres-
sion of the opinions and interests of minorities. Schattschneider (1960)
talks about the ‘mobilization of bias’. Mobilization of bias occurs
through the working of conscious or unconscious mechanisms of exclu-
sion that Bachrach and Baratz (1962) refer to as ‘non-decision making’.
Consensus can obviously be the effect of historical processes through
which parties have gradually developed a shared identity that is
transferred to new generations through lengthy socialization processes.
However, oppression, violence and state terror may also play an impor-
tant role. Behind the Chinese government’s pursuit of a ‘harmonious
society’ lurks a totalitarian regime that endeavours, both at home and
abroad, to maintain the illusion of unanimity while at the same time
ruthlessly crushing any opposition.
Nevertheless, even if consensus is not obtained under duress, it may
still be a terrible thing, since the content of consensus, just like the goal
of cooperation, may comprise something absolutely reprehensible.
Whereas consensus may work against minorities, consensus-based
decision making works the other way around, providing minorities with
veto-power. This mechanism explains why consensus decision methods

are used in the context of empowerment strategies of weakly organized
or underrepresented groups (Butler 1987). Besides contributing to empow-
erment, which may be judged in a positive light, consensus decision
making may lead to a reduction of policy efficiency due to minorities
blocking the implementation of the decisions preferred by majorities,
prolonged deliberations and suboptimal outcomes. Requiring consensus
makes it difficult to realize effective and innovative policy proposals and
creates unfavourable conditions for bargaining, since actors do not
develop the skills to realize trade-offs.
However, although consensus building is often associated with
consensus decision making – attempting to increase the level of
agreement in certain settings – it does not necessarily imply the use of
consensus as a decision rule. Consequently, consensus building does not
automatically generate the same sort of impacts as consensus decision
making and should not be confused with the latter.
The negative and positive functions of conflict

In the same way as consensus, conflict in the context of an interaction
setting between multiple, interdependent actors, may be valued in dif-
ferent ways. It may hinder cooperation, prevent problem solving or may
even result in parties mobilizing their resources in an all-out destructive
struggle producing lose-lose situations. As with consensus, however,
conflict is not by definition negative. Conflicts may fulfil a number of
positive functions.
● Conflicts are an expression of the importance that parties attach to

certain values, problems or solutions.
● Conflicts demonstrate the vitality of an administrative system and
prove that disagreements are not suppressed and parties are not dis-
couraged or silenced by non-decision making.
● Conflicts generate information about the opinions of stakeholders,
alternative problem definitions, undervalued effects and goals, as
well as alternative options.
● Conflicts can lead to the initiation of research and bring about the
gathering of knowledge.
● Conflicts may result in new problems or solutions being placed on
the agenda, in the rethinking of existing priorities, and in the realiza-
tion of reforms and innovations (Coser 1956; Crozier & Friedberg
1980; Brown 1983; Mastenbroek 1987; Schellenberg 1996).
In the same way as consensus building and consensus decision making,

conflict generation as a decision-making strategy and problem-solving
process may be judged in different ways. It may be considered a means
of mobilizing resources to empower minorities, build coalitions and
guide the attention of the general public, the media and decision mak-
ers to previously neglected interests or issues (Cobb & Elder 1983).
However, seeking conflicts may also result in blocked interaction
processes or the destruction of trust and social capital.
All things considered, the concepts of consensus and conflict are
multi-dimensional and ambivalent. The scope, level and intensity of
consensus and conflict may differ: this is not always recognized and may
give rise to misinterpretation regarding the possibility or impossibility
of achieving joint action. Furthermore, the concepts and strategies
used to accomplish them may be appreciated in different ways. The
contradictory views on consensus and conflicts in network theory are
examined in the next sections.
The first face of policy networks: a surplus

of consensus
Network theory is criticized for its high degree of conceptual ambiguity

about what ought to be understood by the concept of networks
(Dowding 1995; Blom-Hansen 1997; Börzel 1998). Acknowledging the
many nuances in network relations, the network literature – insofar as it
is relevant to the discussion of the consensus-conflict dimension –
appears to fall into two groups (Rhodes 1990; Van Waarden 1992). The
first strand of network literature sees a policy network as a tightly con-
nected, exclusive group of actors. This group includes the slightly older lit-
erature on subsystems, subgovernments, iron triangles, neo-corporatism
and policy communities. Alternatively, the second group emphasizes
the fragmented and dynamic nature of policy networks. In this section,
the first group is discussed.
The debate on subsystems, subgovernments and

iron triangles in the US
Subsystems and subgovernments, terms used by authors such as
Freeman (1965) and Ripley and refer to patterns of interaction between
actors involved in making decisions in a special area of public policy,
especially actors connected with the US Congress: congress committees,
executive agencies and special interest groups. Subgovernments, according
to Ripley and Franklin (1987), are involved in routine decision-making
for extended periods. This set of participants is consequently well informed

about both the content of policy issues and procedures. The concepts of
subsystems and subgovernments are more or less synonymous with the
more contested concept of ‘iron triangles’.
Milward and Wamsley (1985) have attempted to give the concept of
subsystems a somewhat broader meaning: they are heterogeneous, have
variable cohesion and exhibit internal complexity. The horizontal
structure of pressure groups, civil servants and politicians is, according
to these authors, connected with a vertical structure in which ‘govern-
ment programs bind program professionals and their professional
associates through all layers of government into vertical functional
autocracies’ (Milward & Wamsley 1985). In this manner, Milward and
Wamsley link the idea of subsystems with implementation networks of
service delivery (Hjern & Porter 1981; O’Toole 1988). The latter litera-
ture is also inspired by interorganizational theory, with its emphasis on
resource dependency and resource exchange (Levine & White 1961;
Benson 1979; Aldrich & Whetten 1981).
Laumann and Knoke (1987) stress the role of private participants in
these networks: ‘despite their lack of formal decision making authority,
many private participants possess sufficient political clout to secure that
their interests will be taken into account.’
Neo-corporatism and policy networks: the

European experience
In Europe, the concept of neo-corporatism may be regarded as one of
the origins of network theory: the state grants powerful interest groups,
especially employers and trade unions, a special position in the political
process, thus enhancing the steering capacity of government and creating
consensus for the implementation of jointly decided solutions (Schmitter
& Lehmbruch 1979; Cawson 1986). In contrast to pluralism, the policy
process is relatively closed: only well-organized and state-licensed interest
groups play a prominent role in policy formation. In this way the
fundamental conflict between capital (employers) and labour (unions)
was resolved.
Building on both interorganizational theory and neo-corporatism, the
concept of policy communities was elaborated in Europe. Whereas
neo-corporatism primarily refers to economic issues at the macro-level,
policy communities were identified in various policy areas at the
meso-level of government (Heisler 1974; Koppenjan et al. 1987; Jordan
1990). Policy communities, according to Rhodes, are networks ‘charac-
terized by stability of relationships, continuity of a highly restrictive
membership, vertical interdependencies based on shared service deliv-

ery, responsibility and insulation from other networks and invariably
from the general public (including parliament)’ (Rhodes 1988: 78).
Members of policy communities share certain interests that separate
them from others. They have also developed more or less formalized
organizational arrangements that support their interactions and exclude
others (advisory boards, consultation procedures, state-approved
membership). To a greater extent than the US literature on interorgani-
zational relations, subsystems and subgovernments, the concept of
policy communities refers to the opinions and values shared by actors.
Actors within policy communities share a policy paradigm: a view of the
world, their surroundings, the issues at play, actors that are part of
the communities and the main instruments and strategies that can
or must be used (Benson 1982; Wilks & Wright 1987; Klijn 1997). To
address the issue of the role of consensus within networks, which has
been relatively neglected within network theory, we can build upon the
literature on policy learning, policy frames and discourse analysis.
Policy networks and the role of consensus

Actors within policy networks are involved in close interaction over an
extended period as a result of which they develop a joint perception of
their world, the problems they deal with and the technologies they use
in doing so (Veld et al. 1991; Termeer 1993). These joint perceptions
relate to what Sabatier (1988) referred to as the policy core. The policy
core is the set of assumptions underlying existing policies that are fun-
damental to the identity and core interests of the actors in a policy coali-
tion. The policy core cannot be subject to discussion and is difficult to
change. Rein and Schön state that within communities, there is a joint
perception or frame with regard to the nature of issues and the relations
between the parties involved. Hall (1993) would call this a policy
paradigm. This consensus prohibits policy learning: perceptions and
policies will only be adapted as long as claims from the outside are not
fundamental.
Interestingly enough, Hall (1993) connects the static concept of
policy paradigms with the more dynamic idea of policy discourses. The
concept of policy discourses suggests that consensus is not based on a
static, delineated set of ideas and values; instead, it is considered to be
constantly maintained, reshaped and discovered in an ongoing discus-
sion between members of a policy community, producing assumptions
and discussion rules that fulfil the function of meaning creation and
gate-keeping at the same time.
A policy discourse is a storyline that structures the debate, determining
which arguments are valid or not. It is both a method by which actors
attempt to arrive at shared meanings and an instrument of consensus
guarding: a tool for the exercise of power by means of which actors
attempt to impose their own interpretations of reality on others (Hajer
1995; Hoppe 1999).
Not all parties have the same opportunities in this argumentation
game. Experts, policy makers, institutionalized interest groups and the
media will be most influential. Politicians and citizens are floored and
outmanoeuvred by information overload and an array of interpreta-
tions. Weakly organized interests are not heard (Hoppe 1999; Hajer &
Wagenaar 2003; Koppenjan & Klijn 2004).
The consensus on policy content within policy networks is often
strengthened by expert knowledge and scientific research. Knowledge
production is not an autonomous process. Policy network and knowledge
production are often intertwined. Scientific epistemological communi-
ties are occasionally even colonized by the government when it funds
scientific activities and educational programs or establishes and main-
tains authoritative scientific institutes that develop knowledge monop-
olies in their field. Research is frequently prompted by perceptions of
the policy community and by knowledge as developed within the domi-
nant framing of the problem. Knowledge develops in interaction
between knowledge producers, policy makers and interest groups mak-
ing subjective decisions partly inspired by societal influences. Research
and knowledge are thus intertwined with policy networks and frame-
bounded, contributing to and strengthening the consensus on policy
formulation and the set of policy alternatives to be considered within
policy networks (Wynne 1989; Gibbons et al. 1994; Veld 2000; Nowothy
et al. 2001).
Policy networks and the overproduction of consensus

Authors of this older strand of network literature share the perception of
networks as relatively closed settings for interaction in which interde-
pendent yet autonomous actors have gradually developed formal and
informal relationships supported by joint perceptions and values and a
common language and identity. These networks can thus be regarded as
arrangements for settling conflicts between mutually dependent actors
within a specific policy area or sub area. The consensus underlying policy
networks can therefore be understood as institutional capital, resulting

from a unique historical learning process of enduring interaction.
Although this consensus is very functional for the actors involved in
settling their conflicts and renders interaction possible (which is neces-
sary for dealing with their interdependencies and joint interests), the
literature on policy networks generally stresses the dysfunctionality of
this consensus. Authors judge that these networks have developed
excessive consensus.
The consensus that actors have developed on the basis of their joint
interests results in ‘mobilization of bias’ and ‘non-decision making’
(Schattschneider 1960; Bachrach & Baratz 1962; Olson 1965; van Eijk &
Kok 1975; Cobb & Elder 1983). At first sight the policy-making within
the network may have pluralistic features: within a multi-actor setting
actors pull and push with regard to problem definitions and policy
measures while non of them wins or looses all the time. At a closer look
issues that threaten the joint interests of the members of the network or
that do not fit their common reality definitions may systematically be
kept out of the arena.
Due to this ‘consensus overdose’, networks are relatively closed to the
outside world, with the following implications:
● a lack of opportunity to address new problems and problem defini-

tions or unorthodox innovative measures that threaten the interests
of the established parties. Consequently, policy networks prohibit
effective and innovative policies.
● networks are organized around the common interests of their mem-
bers. They are privileged oligarchies, optimizing particularistic inter-
ests at the cost of public interests.
● this is the more true, since authors who are supposed to represent
the public interest are not part of the policy network; or if they are,
are socialized in such a manner that they identify themselves with
the interests of the network.
● processes within networks are not transparent, not open to democratic
accountability mechanisms and difficult to control from the outside.
● other societal actors who do not share the interests and views of the
policy network are excluded. Especially the interests of underrepre-
sented or non-represented actors (which include future generations)
are not articulated or taken care of.
● interests and actors that are part of other policy networks do not have
access to the policy formation and implementation arenas within the
policy network. Policies are therefore designed and decided upon
from a narrow sectoral view, thus limiting the possibilities for more
integral or holistic approaches to problems (Marin & Mayntz 1991;
Marsh & Rhodes 1992).
The second face of policy networks:

insufficient consensus
However, the approach to policy networks as subsystems or policy com-

munities is not the only possible approach, nor is it undisputed. It has
already been indicated in the previous section that Milward and Wamsley
(1985), for example, pointed out that the degree of cohesion and
‘closedness’ of networks can vary significantly. Each network has its own
characteristics. In particular, the more sociologically oriented network
approaches concentrate on mapping out the morphological characteris-
tics of networks and their internal differences (Laumann & Knoke 1987).
The observation of the existence of different types of networks made
Heclo (1978) challenge the concept of iron triangles. In his view, net-
works are virtually the opposite of closed. They are characterized by a
large number of actors involved in administrative processes in varying
configurations. The boundaries of the network are vague: ‘participants
move in and out constantly’. That is why Heclo spoke of issue networks
with a strongly ad hoc character (see also Kenis & Schneider 1991). This
approach fits in with Kingdon’s description of the fragmented nature of
the American political system (1984, 1995). Kingdon argued that in the
USA, agenda-building activities, the development of policy measures
and the making of policy decisions take place in separate processes,
which he calls ‘streams’. Due to this institutional fragmentation, the
natural state for public decision making is one of a lack of consensus.
The opportunities available for realizing policies are therefore very lim-
ited. It is only under exceptional conditions that problems, solutions
and political preferences meet. Under these exceptional conditions of
consensus – policy windows – policy can be brought about. These con-
ditions usually only exist temporarily; consensus is extremely vulnera-
ble and difficult to maintain.
Multi-actor and multi-purpose games in

settings with weakly developed institutions
As a result, processes within networks acquire an unstructured character
within which actors must exert themselves to articulate their problems
and place them on the agenda, linking them to solutions that have
often been devised by others, while at the same time ensuring that they
acquire and retain sufficient support to propel these problem-solution
combinations through the relevant decision-making circuits and get
them implemented within a complicated implementation structure
(compare Pressman & Wildavsky 1973).
The lack of consensus leads to laborious interaction processes, games
in which a changing set of actors with divergent interests, goals and
perceptions pull and push to bring about problem formulations, policy
measures and resources. They are characterized by conflict, impasses,
misunderstandings, unexpected twists and turns, and sudden break-
throughs. A successful outcome is not guaranteed: the interaction may
become deadlocked, stop altogether, or end up in compromises that are
unsatisfying for all concerned (Allison 1971).
Attempts at solving conflicts over problem formulations or solutions
using expert knowledge or scientific research works counterproduc-
tively. Since knowledge production does not develop independent of
societal processes, social conflict and problem frames may enter the
research domain. Experts and scientists often do not agree on even basic
facts, and their contradictory reports may contribute to information
overload and confusion. Expert knowledge and research outcomes
embraced in specific policy arenas often encounter a lack of authorita-
tiveness in wider circles and trigger investments in contra-expertise,
resulting in competing claims of truth, policy advocacy and report wars
(Hoppe 1999; Veld 2000).
Issue networks, arenas and policy networks

The meaning of the concept of ‘issue network’ bears strong resemblance
to that of the concepts of ‘arena’ and ‘game’. Issue networks appear to
be more or less the same as arenas in complex policy processes. They are
occasionally also considered to be an advocacy coalitions or adhocra-
cies: sets of actors negotiating with one another about the way to solve
a specific problem or realize (or block) a particular project (Allison 1971;
Crozier & Friedberg 1980; Sabatier 1988). By-products of these interac-
tions may be game rules or shared perceptions. Depending on the dura-
tion and the success of the interaction, these may assume more solid
shape and become institutionalized. Issue networks, arenas and advo-
cacy coalitions may thus also be regarded as proto-networks; interac-
tions in networks’ early stages.
How is it possible that authors differ so greatly in opinion with regard
to the essential characteristics of the policy network phenomenon?
One explanation is that in reality, different networks co-exist. Due to

the specific issues at stake, the characteristics of the wider policy area or
the particular historical development of interaction processes, policy
networks may have developed differently (Koppenjan et al. 1987;
Rhodes 1988; van Waarden 1992). Lowi (1963), for instance, using the
term arenas in this context, argued that depending on the nature and
intensity of the conflict or the clash of interests between a set of actors,
a specific configuration of actors or ‘arena’ develops. Some arenas have
a more pluralist (open) character; others tend towards a more elitist
(exclusive) structure (see also Olson 1965).
Heclo provided a historical explanation for these differences (Heclo
1978). The nature of networks shifts under the influence of social devel-
opments. Following his line of reasoning, one might state that networks
are no longer able to develop or maintain the closed character that they
had in the heyday of iron triangles and neo-corporatism, due to societal
differentiation, citizens becoming more self-conscious, the rise of pow-
erful interest groups and private organizations, the growing administra-
tive density, the increasing role of the mass media, politicians taking
active stands, and the more stringent requirements imposed on trans-
parency, accountability and democratic legitimacy. New issue networks
will increasingly arise and penetrate existing policy networks.
It may be assumed that these types of horizontal settings, in which
consensus, for all intents and purposes, is lacking, primarily occur in
three current policy situations (Klijn & Koppenjan 2004):
1. when the need for interaction arises for the first time between actors
who were not previously aware of their mutual dependencies;
2. when new problems or actors manage to penetrate existing networks,
thus creating chaos so that new forms of consensus must be devel-
oped in order to tackle previously unknown, politicized problems or
to enable interaction between old and new participants; and
3. when problems cut across networks so that actors from different net-
works must learn to interact with one another.
This latter example in particular reveals that in some situations, the lack
of shared institutions and consensus in a network situation cannot be
equated with an institutional vacuum. On the contrary, interaction is
possibly hampered by the presence of a variety of frames, paradigms or
policy cores firmly anchored in the networks of which the various
representatives are a part. It is precisely in these situations that funda-
mental policy controversies may develop, where parties question each
other’s policy cores (Rein & Schon 1986, 1992; Sabatier 1988; Sabatier &
Jenkins-Smith 1993; Schön & Rein 1994). Rein and Schön use the term
‘frame conflicts’ to typify these controversies: conflicts between parties
about whose frames apply. In addition, a lack of joint values, shared lan-
guage, and common rules may inadvertently result in a ‘dialogue of the
deaf’, which blocks the realization of joint solutions and the tackling of
societal problems (van Eeten 1999; Klijn & Koppenjan 2004).
Governance in issue networks

Since institutional arrangements and shared outlooks, values, languages
and rules are relatively weakly developed or inconsistent in this area,
there are few mechanisms available to mitigate conflicts, and there is a
strong need for consensus building in order to enhance interaction and
make joint results possible.
Much of the literature on interorganizational management, intergov-
ernmental management and network management deals with this
problem of conflict reduction and consensus building (for example
Agranoff 1986, 2003; O’Toole 1988; Mandel 1990, 2001; Kickert et al.
1997).
In general, there are three main strategies for dealing with this
problem:
1. rearranging relations in order to make interaction and consensus

building unnecessary or easier;
2. negative coordination: avoiding interaction or conflict (at the cost of
not benefiting from the advantages of collaboration); and
3. positive coordination: strategies aimed at conflict reduction and
consensus building (Scharpf 1978, 1997).
With regard to the third category of network governance recommenda-

tions, most authors consider attempts to create consensus about values
difficult: as mentioned earlier, the adjustment of policy cores, frames or
paradigms is very difficult. Some authors actually advise against it. By
seeking this type of agreement, the drawbacks of consensus decision
making discussed earlier (policy inefficiency, lack of innovation, absence
of a will to negotiate) will be triggered. Olsen (1972), for instance, states
that ‘… if decision makers clarify their values and beliefs, conflict will
arise’. Striving for shared values ignores the fact that a pluriformity of
values exists in a complex society. Instead of a consensus on values, most
authors suggest that there is a need for a strategic consensus, which at the
same time recognizes and respects the various interests and perceptions
existing within network situations (Brown 1983; Agranoff 1986; Hanf &
O’Toole 1992). According to Glasbergen (1995), this strategic consensus
may be about goals and problem perceptions; however, at this level,
values and norms remain pretty much at stake.
Some authors suggest it is far more practical to focus on the develop-
ment of solutions that are advantageous to all of the involved parties. By
bargaining (trading off costs and benefits that are valued differently in
such a way that the involved parties are better off), investing resources in
developing solutions with added value and exploring the solution space
in search of integrative solutions, win–win-situations can be realized.
Scharpf (1978, 1997) calls this strategy ‘problem solving’ (compare also
Fisher & Ury 1981; Teisman 2000). For the realization of these win–win-
situations, actors need not agree on perceptions, objectives or values.
Strategic consensus may also refer to the development of a certain
agreement on interaction rules as a minimum requirement for actors
entering into interaction processes (Burns & Flam 1987; Ostrom 1990;
Klijn 2001). In order to bring and keep parties together in these circum-
stances entrepreneurs, brokers, facilitators and arbiters are considered
essential (Friend et al. 1974; Forester 1989; Mandell 1990).
In addition to the development of strategic consensus, positive coor-
dination can be realized through conflict resolution. Conflict regulation
mechanisms, mediation and arbitration are indispensable attributes for
conflict resolution. Van Eeten (1999) shows how, in the case of dialogues
of the deaf, ‘crossovers’ can be sought by redefining a problem, creating
a new agenda or developing a joint language. Agreeing to disagree can
provide the first step in bringing about a minimum of consensus that at
least allows for coexistence and perhaps for future rapprochements.
The true face of policy networks and its

implications for network governance
Recognizing the existence of two perspectives on policy networks

renders it easier to understand the role of consensus and conflict within
these networks. It becomes clear that it is not so much the normative
positions taken by authors on networks or consensus and conflict that
differ, but rather, that the authors are actually talking about different
networks. The authors subscribing to the first approach describe closed
networks. In these networks, a high degree of consensus has been devel-
oped which in some respects is very functional. Where consensus has
been carried too far, however, it leads to the exclusion of interests, the
production of negative externalities, a lack of transparency, external

accountability and democratic legitimacy, and innovation fails to occur.
The authors representing the second approach are talking about com-
pletely different types of networks or employ a different perspective in
which representatives of the ‘closed’ policy networks, as described by
the authors of the first approach, meet each other in an arena cutting
across these policy networks. See Figure 7.1.
In short: the empirical findings of both groups of actors are not mutually
exclusive but are, on the contrary, highly compatible with one another.
This also holds true for their normative positions. The opinion regard-
ing the negative consequences of an overdose of consensus in closed
policy communities is quite likely to be shared by the authors of the sec-
ond approach; however, they are addressing different settings or issues
when making their recommendations for network management as a
method of consensus building. The network governance recommenda-
tions they suggest are not intended for consensus dominated networks,
which the authors of the first approach have in mind.
The insight that the two approaches are not conflicting but
complementary offers an opportunity to search for a meta-position.
This might consist of a contingency approach: the degree to which
Policy network 1 Policy network 2
arena 2
policy game 1
policy game 2
policy game 3
arena 1 arena 3
Figure 7.1 Policy games and arenas that cut across various policy networks
Source: Klijn & Koppenjan (2004: p 88).
excessive consensus or conflict exists depends on the characteristics of

the specific situation. Given the different characteristics of network
situations, network governance logically cannot be restricted to
strategies directed towards solving a lack of consensus; instead, they
should also counteract an overdose of consensus and a lack of
functional conflict.
Network governance and the management of

the consensus-conflict dimension
If we agree that interaction between mutually dependent parties is the
central issue in network theory, we can then argue that both consensus
and conflict perform important positive functions in policy networks.
Since consensus plays a role in specific, ad hoc interactions – but may
also be institutionalized in the form of a durable set of joint values,
norms, outlooks, frames, language and rules at the level of the network –
it makes sense to distinguish between two levels within policy networks:
the institutional level and the interaction level.
At the institutional level, consensus is an expression of the degree to
which actors within the network have learned to interact. It can be con-
sidered as institutional capital: the yield of enduring, unique historical
processes of institutionalization and learning that makes earlier generic
experiences, knowledge and wisdom available for current interaction. At
the level of specific games and processes in specific arenas, a certain level
of consensus is a necessary precondition for interaction in situations of
mutual dependency. At this level, consensus contributes to problems
being dealt with promptly; supplies the rules by means of which com-
plex and conflicting-inducing issues are made manageable; and ensures
that interaction costs are reduced.
The presence of conflict at the institutional level may be an indication
that the network is not completely closed nor static, as a result of which
institutionalized values, norms, frames, rules, etcetera are questioned,
reflecting opportunities for development and adaptation to new cir-
cumstances. At the interaction level, conflict may contribute to the
articulation of formerly underrepresented interests, trigger the mobiliza-
tion of new resources, information and research, and promote the trans-
parency of processes. Since a greater variety of views, information, ideas
and options are forwarded through conflict, it contributes to the quality
of interaction processes and the innovativeness of the solutions and
policy measures contemplated (Coser 1956; Brown 1983; Mastenbroek
1987).
Too much or too little

Besides fulfilling positive functions, consensus and conflict within
policy networks may become dysfunctional (Brown 1983).
Excessive consensus at the institutional network level leads to the
systematic suppression and exclusion of problems, interests, parties and
innovations. Too little consensus means that an institutionalized
practice hardly exists, meaning that interaction processes develop with
difficulty and derail easily.
At the process level, excessive consensus manifests itself in a lack of
participation, information, options and variety, which ultimately results
in ineffective and non-innovative solutions. A lack of consensus may
lead to the escalation of conflicts but also, in the absence of conflicting
interests or when parties have limited involvement, to prolonged
stagnation or interaction failing to get off the ground.
Excessive conflict at the institutional level means that a network
disintegrates, or that a struggle for the institutional relations within the
network will make it impossible for interactions between parties con-
cerning other, substantive matters to come about. Too little conflict
leads to insufficient articulation of interests, and inadequate allocation
of resources and use of capacities within the network.
In concrete interaction processes, excessive conflict means that
problem solving stagnates and transaction costs mount or – if one of the
parties manages to get its own way – that ineffective solutions are
brought about and relations between the parties involved are seriously
damaged. This latter observation helps us remember that the interaction
and institutional levels are connected and that events within processes
have an impact on the institutional level.
Too little conflict within interactions may lead to the excluding of
important insights, interests or options, to insufficient energy being
generated to tackle a problem, or to insufficient variety being created so
that a decision lacks the necessary quality.
Table 7.1 summarizes the implications of the level of consensus and
conflict within policy networks.
This reflection on the role of consensus and conflict in networks and
interaction processes within networks yields some important insights
into the requirements for good network governance. Contrary to the
dominant emphasis on consensus building and conflict regulation in
network literature, an important challenge for network management is
to prevent and reverse overinvestment in consensus building and to
pursue a healthy level of confrontation and conflict.
Table 7.1 Implications of the level of consensus and conflict within networks
Institutional level Interaction level
Consensus Too much systematic exclusion of exclusion of problem

interests and parties, perceptions, information,
non-transparency alternatives, innovation
Functional offers certainty in simplifies problem solving;
uncertain setting; reduces transaction costs
moderates conflict
through enduring
relations
Too little no sustainable interaction is not brought
institutional solutions about or reaches deadlock
Conflict Too much hinders process; high transaction costs;
network disintegrates solutions are not brought
about or are ineffective,
with negative
consequences for relations
Functional prevents closedness; has mobilizing and
does justice to accelerating effects;
pluriformity, contributes to information
promotes transparency provision, variety, quality
and innovation
Too little insufficient articulation lack of commitment and
of interests and variety; too little
inadequate allocation mobilization of resources
of resources and use
of capacities
Conclusion: managing the consensus–conflict

dimension in network-settings
Of course, the question remains: to what extent can governance influ-

ence the content and degree of consensus and conflict? A number of
conditions regarding consensus and conflict are difficult to manipulate.
There are legal, social and cultural differences between sectors and coun-
tries that can support or hamper the network governance aimed at
bringing about consensus or influencing conflict. In a country with a
strong civil society and a tradition of ‘wheeling and dealing’, with rela-
tively moderate socio-economic cleavages, all this will be easier to
achieve than in a country torn by internal strife due to fundamental dif-
ferences, inequity and a long tradition of hostility and mutual distrust.
Even in the first situation, however, not every conflict or consensus is
governable. Within these limiting conditions, an important challenge

to network governance is to manage the content and degree of
consensus and conflict in network situations. This analysis shows that
in contrast to the dominant view of network governance as a form of
conflict reduction and consensus building, network governance should
also include strategies to promote a functional level of conflict and
prevent consensus from being taken to excess.
Moreover, this analysis emphasizes that network governance is not a
politically neutral activity, aimed at conflict reduction and consensus
building. Influencing the degree and nature of consensus and conflict in
a network situation is pre-eminently political in nature. This means that
network governance is an exercise of power and will be the subject of
struggle and manipulation. It should therefore be embedded in the
democratic process and surrounded by safeguards.
8
Network Governance: Effective
and Legitimate?
Tanja A. Börzel and Diana Panke
Introduction
Network governance has been present in the domestic affairs of highly

industrialized states for a long time. Corporatist arrangements, for
instance, constitute one such form of network governance. With the
‘governance turn’ in International Relations and European Studies,
(Jachtenfuchs 2003) network governance beyond the nation state has
been increasingly studied (Reinicke 1998; Kohler-Koch & Eising 1999;
Börzel & Risse 2005). Many authors consider network governance as a
significant solution to a whole variety of problems relating to gover-
nance within and beyond the nation-state (Marin & Mayntz 1991;
Kooiman 1993; Le Galès & Thatcher 1995; Reinicke 1998; Cutler, et al.
1999). In this perspective, the success of network governance (NWG) is
defined on the basis of the NWG’s effectiveness. As a consequence, legit-
imacy, as the second yardstick for good (and successful) governance, is
often neglected. This chapter is based on the assumption that the
criterion for successful governance is twofold: it must be effective
and legitimate. Hence, overemphasizing effectiveness as yardstick for
successful governance – as much of the literature on network gover-
nance does – is problematic and might lead to attestations of success,
where there is none. Another strand of the literature takes legitimacy
issues seriously and explicitly criticizes network governance for its lack
of democratic participation and accountability (e.g. Scharpf 1993; Benz
1995; Brühl et al. 2001). Since successful NWG requires effectiveness and
legitimacy, it is elementary to inquire about the possible relationships
between both standards for good governance. While some scholars have
argued that legitimacy problems can be compensated by effectiveness
gains (e.g. Reinicke 1998), others contend that there is a certain trade-off
153
between effectiveness and legitimacy (Scharpf 1999; Börzel & Hösli

2003). The more effective (network) governance is (or in Scharpf’s terms:
the higher the output legitimacy), the more problems of (input) legiti-
macy it suffers, and vice versa.
This chapter argues that network governance is in demand for both in
order to provide a viable alternative to other forms of governance. It
examines the relationships between effectiveness and legitimacy and
inquires whether effectiveness and legitimacy are necessarily competing
yardsticks for good governance in networks. We claim that the two
standards for good governance do not necessarily compete. Under certain
conditions, effective and legitimate governance are complementary or
even reinforcing.
The chapter proceeds in four steps. In the first part, we identify the main
features of networks that distinguish them from other types of governance
arrangements. The second part defines effectiveness and legitimacy as
the two main normative standards for good (network) governance. Their
possible relationships are discussed in the forth part. The chapter con-
cludes with a short summary of the argument and some considerations
for future research.
Networks as governance
The governance literature distinguishes between three major types of

social order: markets, hierarchies, and networks1 (Lindblom 1977;
Williamson 1979; Powell 1990). While hierarchies coordinate social
action by using command and control mechanisms, markets are spon-
taneous orders that emerge form the self-coordination of autonomous
actors. Networks, in turn, function by non-hierarchical coordination
based on the exchange of resources and/or trust. Political scientists have
adopted this typology when studying political structures and processes.
Here, governance refers to the setting and implementation of collec-
tively binding norms and rules for the provision of public goods and
services (Kohler-Koch 1999). While public actors have the power to
impose such norms and rules hierarchically, private actors engage in
voluntary self-regulation. In networks, finally, public and private actors
cooperate on a non-hierarchical basis in the making of public policies.
Some have adopted a more narrow understanding of governance
and reserve the concept for the non-hierarchical coordination between
public and private actors, i.e. networks. Governance involves the
‘[f]ormulation and implementation of collectively binding decisions
through the participation of state and non-state actors in public/private
Network Governance: Effective and Legitimate? 155
networks’ (Mayntz 2002). In this view, network governance is ‘gover-

nance without government’ (Rosenau & Czempiel 1992) to which others
have referred to as ‘cooperative’ (Mayntz 1998), ‘modern’ (Kooiman
1993), ‘interactive’ (Kohler-Koch 1997) or simply ‘new’ (Héritier 2002).
The different understandings of (network) governance can be quite
confusing. While there is no authoritative definition, the literature seems
to agree on two characteristics as being constitutive of networks, which
are causally linked and make them distinct from alternative types of gov-
ernance. The first relates to the equal involvement of public and private
actors (structure), and the second refers to the mode of non-hierarchical
coordination (process).
Governance structures emerge from the actors involved and their
relationships. Markets are structures of spontaneous order, in which
(private) actors stand on equal footing and autonomously coordinate
their actions (no structural coupling). Hierarchies, by contrast, are char-
acterized by relationships of domination and subordination between
(public) actors which substantially constrain actors’ autonomy of action
(tight coupling). Unlike markets, hierarchies are able to act intentionally
by hierarchically coordinating the actions of (private) actors based on
the possibility to use force. In networks, finally, public and private actors
enjoy equal status (although the distribution of financial, material, or
ideational resources can be unequal). Public actors may have the power
to impose decisions on private actors. Yet, their relationship with private
actors is not defined by domination but based on the exchange
of resources. Private actors offer public actors information, expertise,
financial means, or political support, which the latter need to make and
enforce collectively binding norms and rules. In exchange, private
actors receive substantive policy influence since public actors do not
adopt and implement policies against the interest of the private actors
involved.
With regard to governance processes, we can distinguish two modes:
hierarchical and non-hierarchical coordination (steering). Hierarchical
coordination involves authoritative decisions, which can be imposed on
actors against their will. Those decisions can be unilateral, e.g. by admin-
istrative order or judicial ruling, but they also include majority voting,
since the minority has to bend to the will of the majority (Scharpf 1992).
Drawing on theories of social action, non-hierarchical coordination can
take two forms (cf. Checkel 2001; Börzel & Risse 2005). Rationalist theories
point to positive (side payments) and negative (sanctions) incentives or
negotiated compromises on the basis of fixed preferences (bargaining).
Sociological approaches, by contrast, emphasize non-manipulative
processes of learning and persuasion, in which actors are socialized into

new norms and rules and change their preferences accordingly (arguing).
Governance structures and processes are causally linked. While states
can steer hierarchically and non-hierarchically, networks, by definition,
are based on non-hierarchical coordination. Markets, by contrast, are
not able to steer or govern at all since they are unable to act intention-
ally. Therefore, markets are not a mode of governance but can only serve
as steering instruments if market principles (e.g. price mechanisms)
are invoked to influence actors’ behaviour (Mayntz 1997).
In sum, network governance refers to the formulation and implemen-
tation of collectively binding decisions by the systematic involvement
of private actors with whom public actors coordinate their preferences
and resources on a voluntary (non-hierarchical) basis. Many authors cel-
ebrate network governance as a significant solution to a whole variety of
problems of governance within and increasingly beyond the nation-
state (e.g. Marin & Mayntz 1991; Kooiman 1993; Le Galès & Thatcher
1995; Reinicke 1998; Cutler, Haufler & Porter 1999). In an increasingly
complex and dynamic world, where hierarchical co-ordination is ren-
dered difficult if not impossible and the potential for deregulation is
limited due to the problems of market failure, networks seem to remain
the only form of governance that is capable of providing collective
goods and services. Yet, the effectiveness of network governance has
been challenged on both theoretical and empirical grounds. Likewise,
network governance has been criticized for its lack of democratic partic-
ipation and accountability. Finally, some scholars seem to suggest that
there is a zero-sum relationship between the two standards for good
governance. Increasing effectiveness necessarily reduces the legitimacy
of network governance, and vice versa.
The next section will argue that network governance needs to be both
effective and legitimate in order to provide a viable alternative to hier-
archy. The remainder of the chapter will then evaluate the effectiveness
and legitimacy of network governance and explore potential tensions
between the two normative standards for good governance.
The demand for effectiveness and legitimacy
The core of governance is authoritative rule making or the production of

public goods and services for a certain group of people. Since authority
has been organized in political orders (cities, empires, states as opposed
to the ad hoc exercise of power), political philosophers have always been
concerned with an important question: What distinguishes good from
bad governance? Early answers focused on the legitimacy of political

orders as a normative criterion (Aristotel 1993; Plato 1973). Max
Weber’s writings on the protestant ethic, the legality and the rational
political order advanced a less normative standard (Weber 1966, 1976):
effectiveness.
Effectiveness: solving problems and meeting demands

Effective governance encompasses a substantive and a procedural aspect
(cf. Héritier 2003; Neyer 2004). Firstly, effective governance produces
policies that solve the problems and satisfy the demands they were
designed to cope with (goal attainment; problem-solving capacity).
Secondly, effective governance requires the production of sufficient
policy output without delays or deadlocks at reasonable cost (efficiency).
Like legitimacy, effectiveness is a standard for good governance, because
it contributes to the maintenance of political systems. In system-
theoretical perspective, political systems exist in order to produce
collectively binding outcomes, suited to solve problems and satisfy
demands. When it fails to fulfil its problem-solving (Easton 1965), goal-
attainment and adaptation tasks (Parson 1976), there is no functional rea-
son for the future maintenance of the respective political system. In other
words, an ineffective political system that produces either predominantly
inadequate outcomes or too few solutions to cope with the environments
or other sub-system’s problem-pressures or demands cannot maintain
itself in the long run (Easton 1957; Parson 1960).
The emergence of network governance has been closely related to the
declining effectiveness of hierarchy (state failure) in domestic politics
and the absence of hierarchy (anarchy) in international politics, respec-
tively. Under the conditions of environmental uncertainty and increasing
international, sectoral, and functional overlap of societal sub-systems,
network governance has a crucial advantage over hierarchy and market.
While markets (deregulation) are unable to control the production of
negative externalities (problems of market failure), hierarchies (govern-
ments) do no longer possess and control all necessary resources to
produce polices capable of addressing societal problems. Through net-
work governance, governments can mobilize resources in situations
where they are widely dispersed among public and private actors at dif-
ferent levels of government, international, national, regional, and local
(Kenis & Schneider 1991; Marin & Mayntz 1991; Kooiman 1993; Mayntz
1993; Le Galès 1995; Wolf 2000). Private actors offer public actors infor-
mation, expertise, financial means, or political support, which the latter
need to make and enforce collectively binding norms and rules. In
exchange, private actors receive substantive policy influence since

public actors do not adopt and implement policies against the interest
of the private actors involved.
Whether network governance is really more effective than its two
rivals is ultimately an empirical question. The evidence presented by the
literature is at best mixed. One problem is that many studies suffer from
a selection bias. There is more research on successful network gover-
nance arrangements than there is on failures. For instance, is the world
a better place because non-governmental organizations (NGOs) sitting
at the negotiation table were able to kill the prospects for a Multilateral
Agreement on Investment (MAI); or would a MAI that included human
rights and environmental concerns have been preferable? In addition,
no empirical studies comparing government attempts to solve certain
problems by means of hierarchy with network governance arrangements
for similar purposes have been undertaken yet. It would go beyond the
scope of this chapter to systematically evaluate the effectiveness of
network governance. Instead, the next paragraph presents a few empirical
examples form international politics, which challenge the effectiveness
of network governance (cf. Börzel & Risse 2005).
Private actors certainly provide public actors with important resources
to make international policies work. In the issue areas of international
development and of humanitarian aid, for instance, public–private part-
nerships between UN organizations and the private sector (mainly
NGOs) were simply a necessity given the lack of material resources of
international organizations. Similarly, in the human rights area we can
observe that the regular provision of information by the NGOs commu-
nity to various UN human rights committees and national governments
has not only greatly improved our knowledge about human rights
violations, but has also increased compliance with international human
rights norms (Risse et al. 1999). However, it is unclear whether the
mutual resource dependency of public and private actors actually leads to
a net increase in the problem-solving capacity of network governance. If
the international community strips international organizations such as
the UN of material resources, the sharing of authority with non-state
actors can easily result in problem-shifting rather than problem-solving
as international organization come to rely on NGOs. In many cases, net-
work governance arrangements between states, international organiza-
tions, non-governmental organizations and/or companies are simply
neoliberal solutions in disguise; that is, they amount to the privatization
and de-regulation of formerly public services. This seems to be very
much the case concerning public–private partnerships involved in the
humanitarian aid and development sectors. Moreover, international

organizations and states do not have the ability to reassume responsibil-
ity for delegated tasks in cases of private failures as functions were
delegated because they were not capable of delivering them in the first
place.
Finally, the inclusion of private actors as the primary rule-targets in
the process of rule-making can certainly increase the problem-solving
capacity by ensuring compliance with international norms and rules
(Chayes & Chayes 1993; et al. 1998). Yet, including private rule targets
in international treaty-making and other governance arrangements
might simply lead to ‘lowest common denominator’ solutions or even
result in deadlock. If those who have to bear the costs of compliance are
involved in the negotiating process, they may attempt to weaken inter-
national rules and regulations or prevent them altogether (Héritier
2003). Ultimately, then, the purported ‘good news’ about private actor
compliance might in fact not tell us very much about the successful
solution of cooperation and compliance problems through network
governance.2 We might end up with international policies that do very
little to address problem-solving challenges. In sum, the effectiveness of
network governance may be less promising than often assumed by the
literature. This is all the more problematic because effectiveness gains
are often seen as a compensation for the weak legitimacy of network
governance, if such problems are considered at all (cf. Scharpf 1999). We
argue, by contrast, that network governance has to be both effective
and legitimate to serve as a functional equivalent or even substitute for
hierarchy.
Legitimacy: generalized support and voluntary obedience

The stability of any political order does not only depend on its effective-
ness but ultimately rests on the obedience of individuals with collectively
binding norms and rules. Max Weber (Weber 1966, 1976), identifies
three sources of obedience: self-interest, legitimacy, and fear (see also
Hurd 1999). Whereas the three are functional equivalents, only legiti-
macy can secure the stability of political systems in the long run.
Political order cannot be maintained purely on the basis of fear. The
threat and use of coercive force presuppose a wide array of effective
monitoring and sanctioning institutions, the repeated use of which
would quickly deplete a system’s resources and lead to collapse in the
longer run. Arbitrary monitoring and sanctioning economizes resources,
but is not sufficient to deter non-compliance, particularly with costly
norms and rules (Axelrod 1984).
Specific reciprocity is equally prone to non-compliance. Firstly,

political systems are simply not capable of making decisions that serve
everybody’s self-interest. Since obedience imposes costs, incentives
towards non-compliance are especially strong for actors with opposing
substantial interests. Secondly, even actors whose substantial interests
are served by a political decision have incentives for non-compliance.
Similar to fear the absence of centralized monitoring and enforcement
institutions renders the strategy of free-riding increasingly attractive,
since actors can benefit from the others’ compliance without themselves
paying the costs (Hardin 1982, 1986; Axelrod 1984). This strategic
reasoning holds true regardless of the extent to which rules reflect the
self-interest of actors.
Unlike fear and self-interest, legitimacy towards a political system
leads to a ‘generalised preparedness to accept, within a certain margin, a
decision whose content is not yet known’ (Luhmann 1975: 25).3 If
actors regard a political system as legitimate, they acknowledge political
power as rightful (Weber 1966) and voluntarily obey its decisions
even in cases in which political outcomes do not correspond to their
self-interest. Legitimacy is conducive to the maintenance and stability
of political order, since the attribution of rightness to the existence of a
political system institutionalizes compliance as a norm, which guides
individual action even in cases of incompatible interests (Luhmann
1996). Should individuals disobey, a legitimate political order enjoys the
generalized support of the vast majority of its subjects, in exercising
the use of coercive force as the last resort for restoring compliance.
At first sight, legitimacy is not an issue for network governance.
Network governance is based on the voluntary cooperation of public
and private actors. By definition, it precludes the use of force and induces
voluntary compliance by consensus decisions. Since those affected by
the rule have a say in the rule-making, political decisions should not
produce any conflicts with the self-interest of the rule-addressees.4
Likewise, if the very purpose of legitimacy is to induce compliance as a
norm, which is considered appropriate for all procedurally correct
decisions regardless of their content, one could argue that if those who
define the content of rules are also those who bear the costs, there is no
functional demand for legitimate governance anymore.
On normative grounds, legitimacy appears to be irrelevant as a quality
standard for network governance. Yet, in the real world, network
governance hardly ever brings about congruence between those who
define the rules and those who must bear the costs of compliance.
Again, international politics provides some illustrative examples.
On a first glance, the legitimacy implications of NWG differ between

the national and the international context. NWG arrangements with
deficits regarding democratic legitimacy might matter less within states
than on the international level as regards the implications for the
overall legitimacy of governance-systems. While undemocratic NWG
co-exists with democratic institutions within most of the modern states,
such compensatory institutions are extremely rare or even completely
lacking on the international level. This could indicate that demands for
democratic legitimacy are higher for NWGs beyond than within nation-
states. However, governance-systems on the international level usually
cover sectoral or functional tasks rather than the whole variety of regu-
lative, distributive, and re-distributive issues as dealt with within states
(Lowi 1972). NWGs on the international level do not render democratic
institutions within nation-states obsolete. Hence, NWGs on the interna-
tional level do co-exist with democratic institutions in states as do
NWGs within single states.
In the absence of hierarchy in the international system, network
governance has not only been heralded as a more effective alternative to
inter-state cooperation in international organizations and regimes.
Many argue that network governance also helps to improve accounta-
bility and transparency at levels of governance beyond the nation-state.
The participation of representatives from the corporate sectors and
transnational civil society is taken to enhance the participatory and
democratic nature of international institutions, by helping to improve
the correspondence between the ‘rulers’ and the ‘ruled’ through the
creation of a transnational demos (Florini 2000b; Reinicke & Deng
2000).5 Moreover, many INGOs, epistemic communities, and other
transnational advocacy networks command undisputed moral authority
in a given issue-area, as demonstrated by Amnesty International and
Transparency International in the areas of human rights and corruption
respectively (Keck & Sikkink 1998; Galtung 2000). Including these moral
authorities in governance mechanisms is said to increase governance
legitimacy.
It has to be asked, however, whether such hopes can be satisfied. Some
have contended that the inclusion of private actors, both for-profit and
not-for-profit, augments rather than alleviates the problems of democratic
legitimacy in international institutions because private actors contribute
to the ‘de-governmentalization’ and ‘commercialization’ of world poli-
tics (Brühl et al. 2001). The participation of non-state actors does not
necessarily make international governance more democratic and there-
fore more legitimate. This is especially true if inclusion is incomplete
and the participation of NGOs and other non-state actors is selective

and lacks transparency. More than thirty years ago, Karl Kaiser raised
exactly the issue of the democratic accountability of transnational
actors (Kaiser 1971). Why is the Transatlantic Business Dialogue includ-
ing multinational corporations, EU and US officials more democratic
than direct negotiations between the democratically elected US govern-
ment and the EU Commission representing equally democratic member
states? The situation is even more complicated where the participation
of civil society actors in international governance is concerned. While
the representatives of non-profit organizations are not generally elected,
these actors can legitimately claim to represent the public interest from
which they draw their moral authority (Wolf 2000). Moreover, many
NGOs have the seal of government approval through the legal recogni-
tion extended to such organizations (Florini 2000a). However, there is
an increasing gap (and subsequently growing conflict) between those
representatives of transnational civil society who are allowed inside the
governance mechanisms and those who remain outside. While most
transnational civil society groups claim to represent the public interest,
some NGOs tend to be self-selected and elite-driven (Keohane & Nye
2001). To some degree, the ‘participatory gap’ (Reinicke & Deng 2000) is
unavoidable and the resulting conflict within civil society might actu-
ally be helpful, as it keeps both sides honest. But evidently network
governance also raises concerns about transparency and exclusivity:
which INGOS are allowed to sit at the bargaining table and how much
information about international negotiations can they disclose to the
public if they wish to maximize their influence on bargaining out-
comes? Moreover, the global cleavage between the rich and powerful in
the North, on the one hand, and the poor and powerless in the South,
on the other hand, is reproduced in transnational civil society. The
INGO world overwhelmingly represents the civil society of the OECD
world including their cultural values (Boli & Thomas 1999). Southern
NGOs have only limited resources to push their concerns in the INGO
community. In fact, one could even argue that this is part of a broader
pattern of Western global hegemony in the international system. Thus,
including selected non-state sectors in network governance arrangements
does not resolve the democratic deficit of international governance and
helps at best to alleviate democratic legitimacy deficits of NWG.
Whether networks decrease or, on the contrary, even exacerbate the
legitimacy problems of governance, seems to depend on several condi-
tions. The crucial question is how inclusive or exclusive the network
governance arrangements are. The more exclusive the networks, the less
accountable and the less transparent they become. However, ‘all-

inclusive’ governance arrangements might lead to a serious lack of
efficiency and reduced effectiveness. In other words, there appears to be
a trade-off between legitimacy and effectiveness. The final section of this
chapter, however, will argue that this trade-off only holds under certain
conditions.
Effectiveness and legitimacy: a trade-off?
Legitimacy and effectiveness are both standards for good governance

within network arrangements. It has been argued that effectiveness and
legitimacy constitute competing governance requirements (Scharpf 1999;
Börzel & Risse 2005). This becomes truly problematic if we concede that
network governance is in serious demand for legitimacy given the
incongruence of rule-makers and rule-addressees in most existing net-
work governance arrangements. Any attempt to increase the democratic
legitimacy would necessarily compromise the efficiency and, in turn,
also the effectiveness of network governance. However, current public
and scholarly debates overlook that the two standards for good
governance do not necessarily compete with each other but can also be
complementary or even reinforcing.
The perceived trade-off between effectiveness and legitimacy is due to
the exclusive focus on politics. Opening-up the decision-making processes
in networks for a broader number of actors increases the diversity of
interests represented but also renders decision-making more difficult. If
actors within NWGs argue, it will be more difficult to achieve a consensus
when there are many participants. If bargaining prevails, even compro-
mises that reflect the lowest common denominator will become more dif-
ficult to achieve, the higher the number of actors at the negotiation table.
Any reform to solve the legitimacy problem necessarily appears to create
additional problems for effectiveness and vice versa. If, however, the
emphasis of reforms is put on political institutions (polity) and political
outputs (policy) rather than on political processes (politics), a trade-off
between effectiveness and legitimacy may be circumvented.
The concept of legitimacy as defined above primarily relates to the
polity dimension. From the core values of legitimacy, different criteria
for the gestalt of political institutions can be derived. Effectiveness, by
contrast, is mainly a matter of policy requiring a high problem solving-
capacity of policy outputs irrespective of core values of legitimacy. We
argue that polity reforms aimed at increasing legitimacy can also
enhance the effectiveness of policy.
In our modern times, two different models of democracy are

prominent, from which two different polity-related reforms to increase
legitimacy can be derived:6 liberal and deliberative democracy. While
both rest on the core values of equality and liberty, the content and
scope of these two values are interpreted differently leading to alternative
ideal typical institutional designs of a polity. Deliberative democracy
defines liberty in a positive way as the liberty to participate in politics
(Rousseau 1996; Schaal & Strecker 1999; Dryzek 2000). Equality is inso-
far important as equal participation of those who are affected by policies
is mandatory. The equal participation of stake-holders, in turn, requires
agenda-setting to be transparent. A second precondition is that the
decision-making process is accessible for all individuals, irrespective of
their resources. Liberal democracy, by contrast, is based on a negative
definition of liberty as individual freedom from the state (Holden 1988).
Hence, active participation in the policy-making process is not a value
in itself. Rather, liberal democracy highlights the principle of equal
representation. By delegating decision-making competencies, represen-
tation allows individuals to enact their liberty without active engagement
in the political sphere. At the same time, representation must be restricted
to protect the public sphere from abuses of individual liberty and equality
by those who exercise delegated power. Elections provide individuals
with an effective tool to control their representatives and hold them
accountable if decision-making is transparent.
Most existing network governance arrangements suffer from legiti-
macy problems since they produce political decisions which are binding
for actors that have not had the chance to participate in the policy
process. In order to remedy the problem, the two alternative models of
democracy suggest different institutional solutions. From the perspec-
tive of deliberative democracy, reforms of network governance should
aim at increasing the access of affected actors. Broader access is a step
towards ideal discourse conditions. As newer studies demonstrate, high
inclusion can improve the discursive quality of interactions and facilitate
policy outcomes beyond the lowest common denominator (e.g.
Deitelhoff 2003). Hence, polity reforms strengthening the deliberative
democratic legitimacy of network governance can also increase the
effectiveness of outcomes. From the perspective of liberal democracy,
reforms should introduce electoral control mechanisms for all partici-
pating actors opening up network arrangement for ex-post sanctions by
the public. If networks are held responsible for their decisions, the
quality of policy outputs is likely to increase as well.
Thus, polity reform can theoretically increase both: legitimacy and

effectiveness. Whether this will actually work out is ultimately an
empirical question. For assessments based on the liberal polity reform
model, empirical investigations would have to consider on which basis
and by whom representatives are selected and whether this fulfils the
formulated criteria for effectiveness. For examinations according to
the deliberative model, it must be gauged who defines affectedness,
whether all affected actors have really access to decision-making arenas,
and – very importantly – whether inclusion alone does the job or
whether power struggles and differences in bargaining or economic
power of participants within the networks prevent true deliberation. If
so, how can power differences between included actors be neutralized?
Put differently: how is arguing strengthened and bargaining down-
graded? First empirical insights gained in the literature on international
negotiations show that arguing and bargaining coexist (Risse 2000,
2003; Holzinger 2004), indicating that access alone cannot prevent
bargaining. Wide access and inclusion do not automatically translate
into arguing and innovative consensual results. Polity reforms are, thus,
only the necessary and not the sufficient condition for successful
network governance.
Conclusion
Effectiveness and legitimacy are both important standards for good

(network) governance since they are crucial for the maintenance of
political order in the longer run. Hence, successful network governance
on the national as well as on the international level cannot do without
effectiveness and legitimacy. Yet, it is often claimed that there is a trade-
off: one standard can only be improved by comprising the other. The
above discussion showed that legitimate and effective network gover-
nance is not necessarily mutually exclusive but can be complementary
or even reinforcing. Increasing the liberal or the deliberative democracy
in strengthening the accountability or the inclusiveness might very well
contribute to an increase in the effectiveness of outcomes, due to possi-
ble ex-post sanctions or to deliberations. Increasing the legitimacy of
NWGs is all the more important since several empirical studies pinpoint
that network governance on the national and on the international level
is not per se effective. As a consequence, effectiveness cannot compensate
for the lack of democratic legitimacy in network governance arrange-
ments. Likewise, doing away with democratic legitimacy altogether does
not do the trick either. Especially lacking a complete congruency between

the rule-makers and the rule-addressees compliance problems arise in
NWG at least in the longer run. Polity-related reforms could be one way
to escape the dilemma and contribute to the success of network gover-
nance within and beyond the nation state. Whether such reforms
actually work out and which additional variables are of importance, is a
question open for further empirical research.
Notes
1. Next to networks, communities and clans (Ouchi 1980), on the one hand, and
associations (Streeck & Schmitter 1985), on the other, are also subsumed
under the third type of governance. Like networks, they are based on non-
hierarchical coordination.
2. As similar argument is made by (Downs et al. 1996) with respect to interstate
relations.
3. The term ‘within certain margins’ refers to the precondition that all decisions
are taken on the basis of and strictly according to prior defined procedures
(Luhmann 1996).
4. Free-riding is a potential problem in all governance arrangements. It will
become especially likely if self-interest is the only source of compliance.
Hence, NWG could profit from legitimacy as a compliance-fostering mecha-
nism, because this could prevent free-riding activities, especially in the
absence of transparency and monitoring mechanisms (Abbott et al. 2000;
Keohane et al. 2000). Network governance arrangements do, however, possess
two functional equivalents that foster compliance: (1) The number of actors is
relatively small (compared to the governed and the government in states) and
substitutes for institutional mechanisms increasing transparency; and
(2) there are iterative games in network governance arrangements, which
combined with a low number of actors and transparency is favourable for the
development of a norm of diffuse reciprocity.
5. This is not to say that governance beyond the nation state needs a “transna-
tional demos” to be legitimate (cf. Brock 1998; Zürn 2000).
6. Criteria of what constitutes legitimacy develop incrementally (Sternberger
1968, 1986; Habermas 1976). With the beginning of the enlightenment,
religious legitimacy as a shared standard of legitimate governance declined as
a result of which the existence of ultimate reason was challenged or even
denied (Habermas 1976: 43). This process led to the development and
prominence of democratic standards as the only substantive basis for legitimate
governance in modern states.
Part III
Metagovernance
9
Metagovernance
Introduction
An increasingly important issue for governance network theory is how

and to what extent it is possible for governors to regulate self-regulating
networks. It is generally recognized that governance networks must be
regulated if they are to contribute to the efficient governing of society.
But since a constitutive feature of governance networks is self-regulation,
it is not possible to regulate governance networks by means of
traditional sovereign forms of detailed, hierarchical and bureaucratic
regulation. Sovereign forms of regulation would inevitably undermine
the self-regulating capacity of the networks. Governance network
theory describes how efforts to harvest the governing capacity of self-
regulating networks, while still being able to ensure an overall societal
governance, has brought with it a growth in new forms of governance,
and suggest ways in which such forms of regulation can be developed
further (Mayntz & Marin 1991; Kooiman 1993; Scharpf 1997; Pierre &
Peters 2000, 2005; Rhodes 2000b; Milward & Provan 2000b; Richardson
2000; O’Toole & Meier 2000; Van Heffen, et al. 2000; Goss 2001).
These new forms of governance differ from traditional forms of sover-
eign regulation in that they regulate networks and other self-regulating
actors without reducing their space for manoeuvring in any radical way.
Governance network theories term these new ways of regulating self-
regulating networks ‘network management’ (Kickert et al. 1997; Rhodes
1997a) or – the term we have adopted in this chapter – ‘metagover-
nance’ (Kooiman 1993, 2003; Jessop 2002). Interdependency theory,
governability theory, integration theory and governmentality theory all
stress different ways in which to exercise metagovernance. The next sec-
tions give an outline of the perceptions and forms of metagovernance
169
that are discussed in each of the four governance network theories. Then
follow an analysis of the similarities and differences between the four the-
oretical approaches to metagovernance, and a description of the choices
that are to be made when choosing one approach and form of metagov-
ernance from the others. Finally, we pinpoint some of the questions that
need to be answered in order to improve our theoretical understanding
of how metagovernance is, and can be, exercised.
Interdependency theory
Interdependency theorists such as Rod Rhodes (1997), and Walter

Kickert, et al. (1997) share the view that the governance potential of
networks can be increased considerably through metagovernance
because the latter can help to forward the negotiation process through
conflict mediation, process planning, and diplomacy. There is a perma-
nent risk that conflicts between network actors will reduce or even
destroy the self-regulating capacity of a governance network, and thus
the main objective of metagovernance is to avoid a situation in which
conflicts undermine the formulation of shared goals.
However, since conflict reduction is generally difficult to promote at a
distance, the most efficient way of regulating networks is through various
forms of hands-on metagovernance that bring the meta-governing actor
(metagovernor) in direct interaction with the self-governing networks
(Klijn et al. 1995: 442; Kickert & Koppenjan, 1997: 40; Rhodes, 1997: 56;
Klijn & Koppenjan 2004: 203). This direct interaction between the
network and the metagoverning actor can either take the form of process
management or network participation.
Through process management, the metagovernor seeks to facilitate
the network in various ways in order to support and increase its selfreg-
ulating capacity. This form of metagovernance is non-interventionist in
that the objective is not so much to obtain influence on the content of
the self-regulation, as it is to promote negotiated self-regulation as such.
Hands-on metagovernance can also be performed through network
participation, where the metagovernor becomes a member of a self-
regulating governance network. This form of hands-on metagovernance
is more interventionist than process management as it grants the
metagovernor a platform for obtaining influence on the content of the
self-regulation. However, the condition of possibility for obtaining this
influence is that the metagovernor plays by the horizontal rules inher-
ent in network governance, and obtains influence with reference to the
resources that they put into the network and not, as might be tempting
Theoretical Approaches to Metagovernance 171
for public authorities, on the basis of their formal hierarchical authority

vis a vis the network.
Interdependency theory claims that, in principle, metagovernance
can be carried out by anybody who has the necessary resources to do so.
The reason why state actors often play a central role as metagovernors is
that they tend to possess the resources that are central for exercising
metagovernance such as legal, institutional and financial resources
(Kickert, et al. 1997: 168). Interdependency theorists are particularly
interested in the potential role of public administrators as metagover-
nors (Kickert Klijn & Koppenjan 1997: 192; Rhodes 1997a: 52). In fact
many interdependency theorists re-conceptualize public administration
as public management and hence inscribe network processing and
network participation as two new tool kits that can be taken into use by
public administrators in an effort to govern society as efficiently as
possible (Kickert et al. 1997: 168; Rhodes 1997a: 56). What interdepend-
ency theorists suggest is that the traditional ideal typical image of public
administrators as rule-following bureaucrats should be exchanged with
an image of them as creative, pragmatic, and engaged process facilitators
and network participants.
Interdependency theorists argue that public administrators have a
good chance of governing networks efficiently by means of various
forms of hands-on metagovernance, but at the same time they insist
that network governance and the metagovernance of networks will
often fail (Rhodes 2000b: 355; Jessop 2002: 243f). Rhodes (2000b: 355)
argues that governance networks are by nature difficult to regulate
because of the complex and conflictual relationship between network
actors, and because network governance is difficult to combine with
other forms of governance such as hierarchy and market, which both
weaken the horizontal interaction and cooperation within the networks
(Rhodes 1997a: 57). For Jessop (1998: 43; 2002: 236ff), the propensity to
fail is not a specific characteristic of governance networks. All forms of
governance tend to fail. State failure and market failure are well known
facts, and network governance and metagovernance will fail as well
(Jessop 2002: 236–43). Governance networks fail when they are unable
to overcome continuous disagreement among the network actors about
the goal of the governing process, while metagovernance fails when it
does not succeed in enabling a network to overcome such conflicts.
Since all forms of governance fail, there is no one superior form of
governance. Networks are not superior to the state and market. They are
yet another imperfect way of governing society. Accordingly, meta-
governors should neither give up state, market or networks, but make
use of them all in turn or in combination in ways that reduce the

degree of governing failures as much as possible in concrete governing
processes.
Interdependency theory predominantly has a descriptive approach to
metagovernance. A number of empirical studies show how metagover-
nance is carried out and advocated in current public sector reforms and
they evaluate how it contributes to the efficient governance of society.
Governability theory
Governability theorists such as Jan Kooiman (1993), Renate Mayntz

(1993) and Fritz Scharpf (1997) put considerable emphasis on the need
for metagovernance. They actually insist that hierarchical regulation of
self-regulating networks is a precondition for making networks an
efficient form of governance. The need for a hierarchical regulation of
governance networks rests on the assumption that unregulated networks
are highly unstable because the individual network actors tend to choose
individual strategies for optimizing their interests instead of developing
a shared strategy together with the other network actors. Another
destabilizing factor is the negotiated nature of network cooperation
since it constantly places the network actors in the ‘negotiators
dilemma’ that is in a situation where they risk loosing invested resources
if the other network actors choose not to fulfil their part of a negotiated
agreement (Scharpf 1994: 45). Faced with the negotiator’s dilemma,
many network actors choose not to take any risks and fall back on
individual strategies with externalities and lack of coordination as the
aggregated outcome. Due to the propensity of the network actor to
choose individual strategies, networks are continuously on the brink of
collapsing as a collective form of governance. But this inherent instability of
governance networks can be reduced through forms of metagovernance
that increase and illuminate the interdependency between the network
actors: the higher the level of interdependency between the network
actors, the greater are their willingness to search for shared goals and
their eagerness to take risks.
While interdependency theory regards state, market, and network
governance as equally imperfect forms of governance with varying
strengths and weaknesses, governability theory stresses the advantages
of governance networks over state and market. In today’s complex,
dynamic and diversified societies, it is argued (Kooiman 1993: 36;
Mayntz 1993b: 11; Scharpf 1994: 41) governance networks represent the
most efficient means to produce efficient governance, because they are
in themselves complex, dynamic and diversified. However, due to the

inherent instability of governance networks, the potentials of gover-
nance networks can only be realized through intensive metagovernance
that increases their stability and ability to coordinate action (Mayntz
1993a: 15; Scharpf 1994: 41).
While interdependency theory primarily underlines the importance
of hands-on forms of metagovernance for the efficient regulation of
governance networks, governability theory first and foremost stresses
the value of hands-off metagovernance through institutional design.
Institutions structure the conditions under which self-governing actors
interact and, as such, they structure the relations of interdependency
that lead to the establishment of networks among certain actors and
competition among others (Scharpf 1994: 40; Kooiman 2003: 155). In
Kooiman’s words metagovernance through institutional design involves
the construction of ‘the “rules of the game”… the medium through
which actors can act and try to use these rules in accordance with their
own objectives and interests’ (Kooiman 1993: 258). This game structur-
ing is exercised through the deliberate strategic design of institutional
conditions which visualizes and increases the interdependency struc-
tures between specific actors (Kooiman 1993: 251).
Game structuring can be performed in different ways: 1. through the
institutionalization of plus-sum games that enhance network cooperation
by redistributing the resources of the network actors; 2. by making it
clear to the network actors that they network ‘in the shadow of hierarchy’,
which means that the right to self-regulate will be reduced or removed
by the metagovernor if the network is unable or unwilling to solve a
defined policy problem (Scharpf 1994: 40–1, 367); and 3. through vari-
ous forms of incentive steering. These three forms of game structuring
gradually become more and more interventionist the first being the low-
est level of intervention and the third, the highest. While the construc-
tion of plus-sum games merely promotes network cooperation among a
defined group of actors, the existence of a heavy shadow of hierarchy,
and most markedly incentive steering seeks to influence the content of
the self-regulation.
Since governability theorists regard the exercise of metagovernance as
a precondition for efficient network governance, considerations about
the role of governance networks become inseparable from considerations
about the role of the state. Both Mayntz (1993: 18) and Scharpf (1994:
40) emphasise that there is basically a mutually beneficial relationship
between governance networks and the state. Governance networks fulfil
some of the governing needs that would otherwise have to be fulfilled
by the state, while the state establishes the conditions for efficient
coordination of individual action both within and between networks.
Governability theory argues that this interdependency between gover-
nance networks and the state means that the need for hierarchical state
rule is as strong as ever. The growing need for governance networks does
not indicate a goodbye to hierarchical state rule. Rather it calls for new
forms of state rule. Kooiman contends that the central role of governance
networks in contemporary processes of societal governance: ‘has nothing
to do with a withdrawal of the state or with the state disconnecting itself
from society, but with a state that governs in other, more apt, ways. In
many cases “other” will mean more connecting than disconnecting.
The task of governments in contemporary, complex societies is to influ-
ence social interactions in such ways that political governing and social
self-organisation are made complementary’ (Kooiman 1993: 256).
The state is not weaker than before, but in today’s society it must gain
influence through the hands-off metagovernance of social interaction.
Mayntz agrees that the position of the state is by no means threatened
by the increased role of governance networks because of the ability of
the state to define the institutional rules of the game: ‘what we are deal-
ing with is not so much a loss of state control as a change in its form.
Societal self-regulation takes place, after all, within an institutional
framework that is underwritten by the state’ (Mayntz 1993a: 31). Seen
from this perspective, the strength of the state vis-à-vis governance net-
works depends on the ability of the state to metagovern by the strategic
construction of institutional designs that increase the selfregulating
capacity of governance networks, and the coordination capacity of the
state (Kooiman 1993: 255f ).
In order to prepare the state for this new metagoverning role, the state
must reorganize itself. The traditional organizational lines of demarca-
tion between politics and administration and between state and society
make even less sense as the distinction between metagovernance and
self-regulation becomes still more crucial as the core principle for the
distribution of competence between the different actors that participate
in the governing process at different levels and stages be it inside or out-
side the state (Kooiman 1993: 257). Today, the central questions are not
how to separate politics from administration or the state form the mar-
ket and civil society. The central question that should guide considera-
tions concerning the reorganization of the state apparatus must be how
to simultaneously upgrade the capacity of the state to metagovern and
the best possible conditions for self-regulating governance networks. In
Kooiman’s words, the job consists in striking ‘an effective and legitimate
balance between social self-organization and political–administrative

intervention’ (Kooiman, 1993: 255). The ability to strike this balance
calls for organizational patterns within the state that are radically different
from those characterizing the traditional liberal state.
In contrast to interdependency theorists, governability theorists are
generally prescriptive in their approach to metagovernance. They argue
that the hands-off metagovernance of self-regulating networks represents
a favourable alternative to neo-liberal efforts to increase the governing
capacity of the state by means of the market as well as to traditional
bureaucratic forms of state rule. Kooiman (1993: 250) directly criticises
public leaders and others who are engaged in reorganizing the public
sector for paying too little attention to the ongoing changes that are
taking place in society, and the need for new forms of governance that
these changes bring about. In addition, he accuses social scientists for
doing too little to make public leaders aware of the role that meta-
governance and governance networks can play in meeting the new
challenges that these societal changes bring with them (Kooiman 1993:
249). It is not enough to produce convincing research results. These
results must also be disseminated to those who might make use of them
in their efforts to increase the governability of society.
Integration theory
Integration theorists like James G. March and Johan P. Olsen (1989,

1995) and Richard Scott (1995) stress that the political identities and
capacities of the involved actors are crucial for the degree to which
governance networks are able to contribute to the governing of society.
Accordingly, a core element in the exercise of metagovernance – a term
that is not used by the integration theorists themselves – is the strategic
formation and development of the political identities and capacities of
the network actors. The metagovernance of identities is exercised hands-off
through the shaping of institutional rules, norms and logics of appro-
priateness within the self-governing networks; through the production
of specific forms of knowledge; through story telling about ‘best
practices’; through a forceful campaigning of specific frames of meaning;
and through the construction of symbols and rituals. All these identity
producing forms of metagovernance seek to influence the network
actors’ perception of themselves and the context they are part of: Who
are they as social and political actors? Who are the other actors within
and outside the network? What is the overall purpose of the network?
What is appropriate conduct within the network as well as in relation to
external actors? The metagovernance of identities can be seen as an

interventionist form of metagovernance as the identity of the network
participants have considerable impact on the content of what is being
self-regulated.
The metagovernance of the network actors’ capacities is less interventionist
as the main ambition is to supply the self-regulating actors with the
resources that they need in order to be able to regulate themselves and
not to influence the content of the self-regulation. The metagovernance
of the network actors’ capacities is performed hands-off through the
strategic shaping and transformation of political rights, through the dis-
tribution of various social, financial, authoritative and institutional
resources as well as through more hands-on metagovernance that sup-
ports the development of the political know how and general learning
abilities of the network actors (March & Olsen 1995: 223ff).
But what is the precise purpose of the efforts to metagovern networks
through the shaping of the network actors’ identities and capacities?
One of the central objectives that motivates the exercise of identity pro-
ducing metagovernance is the formation of a strong democratic political
community within and between networks March and Olsen (1995:
63ff). Hence, metagovernance should enhance solidarity and a strong
sense of communality among the involved actors within the network
through the creation of shared meaning and common visions that facil-
itate consensus. Metagovernance should also promote the shaping of
democratic political identities among the political actors. This can be
done by means of story telling that refers to the network actors as ‘respon-
sible citizens, interest organizations, firms and voluntary organizations’,
‘attentive politicians’, and ‘responsive administrators’. The purpose of
such references is to spread and consolidate an image of the network as
being a part of a democratic political system that subscribes to a set of
democratic values including values such as mutual respect, reciprocity
and communality.
Metagovernance of the network actors’ political capacities concerns
the mobilization and enhancement of their ability to act individually
and collectively in order to make a difference to the governance of
society. However, from a democratic point of view, it is not enough
merely to enhance the capacities of the involved as much as possible,
and to mobilize them in concrete governance processes. Metagovernance
must deliberately seek to regulate the native of the capacities that are
created, who obtains them, and when and how they are brought into
use (March & Olsen 1995: 130ff). Seen from a democratic perspective, it
is important to ensure that the differences between the political
competences that different network actors possess are not too large.
Metagovernance must ensure a level of equality in the distribution of
competencies that installs a considerable degree of interdependency
between the network actors in order to ensure that all network actors
gain some level of influence. Finally, the metagovernance of political
capacities involves the shaping of democratic ideals that define where,
when, and how specific political capacities can be used. Individual and
collective actors must use their capacities to obtain political influence,
and have a duty to mobilize all their powers to withstand the illegiti-
mate use of public power. At the same time, metagovernance must seek
to prevent network actors from using their competencies to undermine
the democratic process both within and between networks.
It is rather unclear who integration theorists perceive to be perform-
ing metagovernance. Scott (1995: 93f ) emphasizes the central role of
the state in the regulation of organizational fields. The state is not
merely one among a plurality of political actors, due to its monopolized
right to exercise legitimate coercion, it has a unique position for
exercising metagovernance of organizational fields. Hence, it has the
capacity to construct and change the institutional framework within
which governance networks regulate themselves, and to interfere in
various conflicts between self-regulating actors. However, March and
Olsen (1995: 69) argue that the Weberian concept of the state as a uni-
tary political organization with a monopoly on the legal exercise of
coercion is inadequate if we are to grasp the reality in the disintegrated
polities of today. Never the less, March and Olsen, on several occasions,
refer to ‘the political system’ or simply ‘the government’ as the execu-
tor of metagovernance while they, on other occasions, more diffusely
refer to ‘society’ or ‘the modern democracy’ as the executor of metagov-
ernance. All in all, we are left with a relatively confused image of who
the metagovernor is.
In general, integration theory tends to have rather high expectations
as to the possibility of regulating self-regulating governance networks
efficiently by means of identity shaping and capacity creating forms of
metagovernance. This is due to the basic assumption in integration the-
ory that actors and their actions are, by and large, conditioned by exist-
ing structures of meaning, and that it is possible to systematically
influence action through the shaping of these structures of meaning.
Furthermore, integration theory tends to have a prescriptive approach
to metagovernance since it seeks to outline how the metagovernance of
governance networks should be performed in order to increase the
democratic quality of societal governance processes.
Governmentality theory
Even though governmentality theorists such as Michel Dean (1999),

Michel Foucault (1991a), and Nicolas Rose (1996) – like the integration
theorists – do not use the term metagovernance, the regulation of self-
regulation plays a central role in their understanding of how governance
is exercised. The term governmentality refers to the institutionalized
collective mentalities that define what governance is and how it is
performed. Governmentality theory sees the regulation of self-regulation
as central element in the dominating governmentality in advanced lib-
eral societies. We recognize this image of governance as metagoverned
self-regulation in the subjectivations and rationalities articulated in
most contemporary governmentality programs, and we recognize it in
the governance techniques that are currently proposed from bench-
marking to contracting out.
In order to get a more precise idea about how metagovernance is being
performed today, it is necessary to take a closer look at the historically
specific liberal governmentality program of which metagovernance as
the regulation of self-regulating governance networks is a part. Rose
(1996: 43) claims that a core feature of the liberal governmentality
program is the effort to govern at a distance without getting involved in
the detailed governance activities and the conflicts between different
actors. Dean (1999: 165) claims that advanced liberal metagovernance
concerns a simultaneous subjectification and subjection of self-regulating
actors. Advanced liberal metagovernance constructs and mobilizes free
subjects, while at the same time placing them within specific conditions
of possibility and specific power relations. On the one hand, metagover-
nance establishes partnerships, hearings, negotiations, user boards, free
choice and participation incentives (‘subjectification’). On the other
hand, metagovernance constructs a set of norms, standards, bench-
marks, and performance indicators, and various hierarchical instruction
and control mechanisms, which in sum make it possible to measure,
evaluate and sanction the self-regulating actors (‘subjection’). In other
words, metagovernance has two sides. It constructs and mobilizes the
network actors’ energies, resources, capacities and knowledge, and it
ensures that the empowered, selfregulating actors act within the limits
set by specific discursive conditions of possibility, which do not manifest
themselves as external boundaries to self-regulation, but through their
presence in the minds of the self-regulating actors.
Metagovernance mobilizes the energies, resources, capacities and
knowledge of the network actors through the use of technologies of
agency, while it disciplines their minds through various technologies of

performance (Dean 1999: 167ff). Different kinds of contract technologies
make it possible for public institutions to form binding ties with firms,
organizations, groups, and citizens and thereby include them in various
governance networks. These governance networks are then forced to act
within specific social, political and financial conditions of possibilities
through ongoing evaluation practices. The force that is exercised neither
has the character of legal sanctions or financial loss on a free market.
Rather, the force lies in naming and shaming actors as either a success or
a failure. What is at stake is the general reputation of the actors.
Metagovernance through various interrelated forms of subjectifica-
tion and subjection is predominantly exercised hands-off through the
construction of meaning and institutionalized technologies, and it has a
decisive interventionist top-down character. However, it is also possible
to point out forms of metagovernance within the advanced liberal
governmentality program that are less interventionist and more bottom-
up. Hence, one of the central performance indicators that are used to
evaluate self-regulating actors including governance networks is their
responsiveness to affected citizens (Dean 1999: 169). However, since it is
most often centralized government who define the performance indica-
tors even this responsiveness is an outcome of top-down metagovernance.
This conclusion gives a strong hint about who governmentality
theorists view as the primary provider of metagovernance. Even
though Foucault recognizes that metagovernance can be performed by
a wide range of actors including private actors, he emphasizes the cen-
tral role of the state in this respect. The state is not one among many
power holders, but the instance that all exercise of power must relate
to, not in the sense that all exercise of power derives from the state, but
in the sense that over time, more and more instances of power are con-
trolled by the state. With reference to the narrow meaning of the word
‘government’ as state, one can say that the many forms of power
that are exercised in society have become governmentalized (Foucault
1986: 224).
The state plays a central role in setting the scene for the self-regulating
actors and in governing ‘the conduct of conduct’. The starting point for
the metagoverning state is the discursive construction of the objects,
tool kits, identities, and purposes of self-regulation. But this form of
state rule is enforced through the many disciplining technologies and
sovereign forms of rule that the state masters. This understanding of the
role of the state has some parallels to Scharpf’s (1994) claim that
network governance takes place ‘in the shadow of hierarchy’. If the
discursive metagovernance fails it can make use of more direct forms

of governance, even though more direct forms of metagovernance is
likely to meet considerable resistance due to the articulation of citizens
as empowered selfregulating contributors to the production of public
governance.
The approach that governmentality theory takes to metagovernance
is more descriptive than prescriptive. It describes the contingent discursive
conditions of the possibility for exercising governance in advanced
liberal democracies, and indicates that, seen form a historical perspec-
tive, the ability to metagovern self-regulating networks through various
forms of subjectification is considerable.
Similarities and differences between the theories
There are many similarities between the four governance network

theories with regard to their views on metagovernance. They all insist that
self-regulating networks have become an inseparable part of modern
governance processes, and that they contribute to the production of
efficient societal governance. For that reason, societal governance can no
longer be exercised through sovereign bureaucratic forms of governance
but must take the form of metagovernance.
However, there are also many differences between the theories. For
one thing, interdependency theory and governmentality theory are
predominantly descriptive, while governability theory and integration
theory are prescriptive. There are also differences in their approach to
how metagovernance can be carried out. Interdependency theory pri-
marily focuses on how hands-on forms of metagovernance facilitate
network formation, network negotiations and outcome orientation in
networks. Governability theory, integration theory and governmentality
theory on their side take issue with various forms of hands-off metagov-
ernance, which work through different forms of institutional framing of
governance networks. Governability theory suggests that metagover-
nance involves the construction of institutionalized games that promote
coordinated collective action between self-interested actors, while
integration theory and governmentality theory point to how metagov-
ernance is exercised through the cognitive and discursive framing of
identities, capacities and meaning.
Interdependency theory also differs from the other governance
network theories in its view on the degree to which it is possible to
metagovern governance networks. Its starting points in a calculus
actor oriented approach to human action and a conflict view on what
Table 9.1 Differences between the four governance network theories
Approaches to Interdependency Governability Integration Governmentality

metagovernance theory theory theory theory
Perspective Descriptive Prescriptive Prescriptive Descriptive

Form Hands on Hands off Hands off Hands off
Efficiency Low High High High
constitutes governance processes, lowers its expectations to the possibility

of governing society in general and to the possibility of regulating
self-regulating networks in particular. The many conflicts between self-
interested rational actors mean that the metagovernance of networks
fails more often that it succeeds. By contrast, the other theories tend to
regard metagovernance as a highly effective means to govern networks.
Governability theory and integration theory regard the institutionaliza-
tion of incentive structures, normative rules, and cognitive images as a
forceful means to regulate autonomous actors, while governmentality
theory describes metagovernance as a highly efficient advanced form of
subjection and subjectification. The differences between the four
theories are summarized in Table 9.1.
Although the four theories have different approaches to the question
of perspective, form and efficiency, these differences do not entail any
clear cut choices. Descriptive and prescriptive perspectives are often hard
to separate completely; hands on and hands off forms of metagover-
nance tend to function well together; and metagovernance that tend to
be efficient in some contexts, might not be so in others. As such, the dif-
ferent perspectives, forms, and evaluations of metagovernance that the
four theories provide seem to be more supplementary than exclusive.
Where to go from here?
The four theories of governance networks have provided knowledge

about the driving force behind the emergence of metagovernance, how
it is and can be performed, and what the problems and potentials of
metagovernance are. However, some of the central questions relating to
improving our understanding of metagovernance that have been raised
by the second generation of governance network theorists have only
briefly been discussed and analysed by the first generation of gover-
nance network theories. In this book we have selected three questions,
which are in need of further theorizing if we are fully to qualify and
develop our understanding of metagovernance. These questions
concern: 1. how the formation and mobilization of governance networks

is regulated through metagovernance; 2. how network negotiations are
metagoverned; and 3. how metagovernance affects the production of
outputs and outcomes of governance networks. These topics are dealt
with in the next three chapters, which in different ways combine and
transgress the theoretical positions to metagovernance presented in this
chapter.
10
Governing the Formation and
Mobilization of Governance
Networks
Peter Triantafillou
Introduction
This chapter is motivated by a specific concern over the capacity of

contemporary (network) governance studies1 to critically address the
governing of networks. This concern is nurtured by the proximity
between the problem space informing contemporary political action, on
the one hand, and the theoretical imagery informing governance
studies, on the other. When politicians, civil servants, business man-
agers, NGOs and concerned citizens’ attempt to explain the existence
and justify the need for mobilizing governance networks, they often
provide a set of explanations that, I think, are remarkably close to those
provided by contemporary governance studies. Perhaps the reader will
agree that in contemporary society, narratives of the following type are
encountered quite commonly:
● Because society has grown increasingly complex, public problems

cannot be solved by the state alone. Without drawing on the
knowledge and resources of diverse public-and private actors, state
action is doomed to failure.
● While the market may serve as an excellent mechanism for solving
information-demanding problems, it is incapable on its own of
providing satisfactory solutions to a wide range of collective problems,
such as social and economic inequality and environmental problems.
● It is therefore necessary to include and stimulate the participation
and interaction of a wide range of state, market and civil society
183
actors in order to provide legitimate and effective solutions to

contemporary problems.
True, one finds political calls for the state to remain the locus of
policy-making, or others who would prefer the market to constitute the
primary decision-making and problem-solving mechanism. In other
words, we find many conflicting opinions within contemporary politi-
cal discussions regarding the role of the state. Nevertheless, I believe that
the narratives outlined above have become rather widespread among
influential politicians, civil servants, business managers, NGOs and
citizens participating in or attempting to promote various forms of
governance networks.
Moreover, despite the fuzzy and often contradictory character of these
narratives, I think it is fair to assert that they constitute a problem-space
that is remarkably close to the theoretical imagery underpinning many
governance studies. By theoretical imagery, I am referring to types of
research questions, key concepts, objects of enquiry and lines of argu-
mentation. Some academics assert rather bluntly that the state is being
‘hollowed out’ (Rhodes 1994; Saward 1997). Others ask how policy
input and policy delivery are shaped in a society characterized by
increasing complexity, where hierarchical co-ordination is rendered dif-
ficult if not impossible, and the potential for market solutions is limited
because of the problems of market failure (for example Mayntz 1993b;
Kooiman 1993). And yet other, more normative (policy recommending)
governance publications inquire as to when and under what conditions
of interaction diverse public and private actors succeed in mobilizing
resources and capacities in a manner that will improve problem-solving
capacities (Kickert & Koppenjan 1997: 53–8).
These and similar questions are all very relevant and legitimate;
however, I do not view them as adequate for understanding the role
networks play in contemporary forms of government. My concern does
not regard what some view as the insufficient capacity of the aforemen-
tioned studies to explain changes in policy formation and policy
outcome (for example Dowding 1995). Nor does it regard what others
take to be the tendency of the governance literature to overstate the
autonomy of the networks vis-à-vis (central) state authorities (Marinetto
2003).
Instead, my concern regards the proximity between the research
questions informing many governance network studies and the self-
understanding and reflections of contemporary forms of rule, notably
advanced liberal government. This term is unfolded below, but essentially
Governing Formation and Mobilization of Governance Networks 185
addresses the multifarious reflections over – and answers provided to – the

following problem: how can the self-governing capacities of individuals,
groups, organizations and networks be developed and facilitated?
I believe that the proximity between advanced liberal governmental
thinking and governance studies poses a number of important limita-
tions on the ways in which we can analyze governance networks. In
particular, I think we must detach ourselves more fundamentally from
the self-understanding of advanced liberal government in order to be
able to illuminate those modalities of power that work not through
force or coercion, but above all through diverse governing mechanisms
seeking to facilitate the self-governing capacities of diverse actors.
The aim of this chapter is therefore to discuss two interrelated
working hypotheses on the conception of the governing of governance
networks: first, the proximity between the problem space of advanced
liberal government and the theoretical imagery underpinning many
governance studies poses important limitations on the capacity of these
studies to address critically the role played by agency, autonomy and
interdependency in the shaping of governance networks. Second, by
adopting Foucault’s conception of government, we may overcome some
of these limitations and open a space for the critical analysis of the
formation and mobilization of governance networks in our societies. It
should be stressed that rather than attempting to test or verify these
hypotheses, I use them as practical guidelines for my conceptual discus-
sions. Moreover, both hypotheses are based on the (unquestioned)
assumption that the theoretical imagery underpinning many gover-
nance studies actually approximates the problem space of advanced
liberal government.
Governmentality and advanced liberal government
Michel Foucault introduced the notion of governmentality in his 1978

lectures at College de France on ‘Security, Territory and Population’
(Foucault 1978; see also Foucault 1988). While Foucault had explicitly
avoided the question of the state in his studies of disciplinary power, his
problematization of the exercise of power over territories and popula-
tions forced him to deal with the forms of political power evolving
around the modern territorial states in Western Europe. However, the
key category for addressing this issue was neither power nor state power,
but the notion of government. Government referred not to the govern-
ment (the state apparatus), but to the manifold ways in which the
conduct of individuals and groups are directed. Defined as ‘the conduct
of conduct’, government refers to the diverse schemes, programs and

methods seeking to structure and nurture the field of actions of others
(Foucault 1982, p. 221). By implication, the practice of government
is inevitably an exercise of power. However, the notion of conduct
points not only to the directing of the conduct of others, but also to the
ways in which one conducts oneself. If by freedom, we are referring not
to the absence of power but to the diverse and everyday forms of reflec-
tions, choices and actions we make with regard to ourselves, then it
becomes clear that advanced liberal government implies not only the exer-
cise of power, but also the ways in which we (individuals, organizations
and networks) exercise our freedom.
The term governmentality – Foucault’s neologism contracting the terms
government and rationality – was coined to denote a set of historically
specific constellations of problematizations, forms of knowledge and
practices of government (Foucault 1991a: 102; Foucault 1991b: 78–82).
This concept seeks to point out two things: first, that schemes, programs
and methods seeking to shape the conduct of others are always already
informed by particular forms of knowledge regarding the object to be
governed and by particular sets of queries or problematizations. And,
second, that the forms of knowledge turning an issue into an object or a
problem of government is informing and structuring the types of
schemes, programs and methods that are selected and developed for
solving the problem. To take a close example, if contemporary forms of
knowledge depict increased societal complexity and state overload as
the main problem, they obviously influence the type of possible
governmental actions that are deemed proper.
Based on the notion of governmentality, advanced liberal government
may be understood neither as an ideological doctrine, nor as a party-
political programme entailing, for example, privatization. It may
instead be understood as a plurality of often contradictory governmen-
tal schemes, programmes and methods informed by a particular manner
of reflecting on – or more precisely problematizing – the practice of gov-
erning. From this point of view, it is possible to talk about three broad
types of political rationalities that have dominated Western European
societies since the end of the 18th century, namely liberal, social liberal,
and advanced liberal government. These three types ought to be consid-
ered ideal types that aim to capture certain aspects of multifarious and
overlapping reflections on how to govern the wealth, health and welfare
of a population. As such, liberal government, articulated most clearly by
the Scottish moral philosophers Adam Smith and Adam Ferguson,
conceives of civil society as a distinct sphere with a natural/inherent
capacity of self-governance that the state should be wary of disturbing.

Social liberal government, which emerges towards the end of the 19th
century, is above all concerned with the problem of how to ensure
solidarity and devise (diverse insurance) mechanisms promoting social
security (Donzelot 1988).
Advanced liberal government shares the problematization of
excessive state intervention with liberal government. Unlike liberal
government, however, civil society is not regarded as a sphere with a
natural capacity for self-governance. Rather, the latter capacity must be
artificially constructed on the basis of diverse governing mechanisms
that will provide the institutional conditions necessary to stimulate and
facilitate its self-governing capacities (see Burchell 1996: 27–30). Thus,
what characterizes the problem space of advanced liberal government is
not primarily the concern for excessive state intervention, but questions
such as: how can the capacities of individuals, groups, organizations,
and networks be nurtured and facilitated to deal with the problems of
their own and/or the wider community? What sort of schemes, devices
and methods may be brought into play so as to create an institutional
setting that will prove conducive to the self-steering and problem-
solving capacities of diverse actors and networks? Conceived as a prob-
lematization evolving around the question of how best to augment the
self-steering capacities of diverse actors, advanced liberal government
has less to do with de-regulation and more to do with the re-regulation
of the relations between public and private organizations and the
relations within them.
Mobilizing agency
How can we understand and analyse the mobilization of a network and

the capacities of its respective participants to handle policy problems? In
this section, I attempt to argue that we must be very cautious in the
ways in which we address the agency of network participants. I attempt
to explain why we cannot take actors for granted, that is, individuals or
organizations equipped with a capacity to act. I argue that in the case of
governance networks, agency – the capability to act – is something
that always already depends on a series of devices, methods, and tech-
niques seeking to form and mobilize the capacities of individuals and
organizations to act in particular ways.
In the attempt to explain the mobilization of networks, many
network studies tend to attribute significant importance to the interests
of the (potential) network participants.2 Notwithstanding the important
differences in the understanding of interest formation, many network

studies tend to resort to the pursuit of interests when accounting for the
formation of networks. However, as is well argued in the political
science literature, we cannot take the interest of individuals and organi-
zations for granted. Neo-institutionalist theory of a certain construc-
tionist bend, such as March and Olsen (1989), have even argued that the
behaviour of political actors is shaped not by rational self-interest (or
logic of consequentiality), but by various institutions (rules and norms)
forming a logic of appropriateness. In line with March and Olsen (1989:
160–6), I believe that the point is not that individuals and organizations
do not hold interests, but rather, that these interests are always already
informed by – and depend upon – particular norms (such as maximizing
utility), particular forms of knowledge (such as ego psychology or neo-
classical economics), and particular techniques for calculating the
rational as opposed to the irrational path of action (such as cost–benefit
analysis).
While I find March and Olsen’s social constructionist approach
appealing, I do not think their theory is entirely suitable in order to
critically analyse the formation of agency. Not only are March and Olsen
paying quite limited attention to power and conflict, their very ambi-
tion for proposing a more or less universal theory (a truth) about the
formation of agency may prove counter-productive. Thus, if the capacity
to exercise power over a subject depends upon a certain theoretical
knowledge or truth concerning the conditions under which that subject
will act in particular ways, it is obviously problematic to take that truth
as the foundation for studying power. Consequently, if our purpose is to
propose a critical analysis of various workings of advanced liberal
government, such as governance networks, we may be better off with an
epistemologically speaking less ambitious conceptual framework that
explicitly addresses the methods, devices and techniques of power
involved in the formation of agency.
One way of grasping the mobilization of agency in networks is to view
agency as the result rather than the cause of diverse social practices.
During the last two or three decades, our societies have experienced the
emergence of a whole series of practices around norms of activism,
participation and empowerment. This includes opinion polls, surveys,
public hearings, citizen panels, workshops, and consensus conferences
(for example Rowe & Frewer 2005). Such technologies of agency may seek
to bring forth the voice and choice of citizens, organizations and
networks.3 They may also seek to enable and induce these groups to be
active and participate more directly in policy implementation through,

for example, the contracting out of public services, public-private
partnerships and regional development agencies (for example Chalmers &
Davis 2001). A crucial qualification must be made in this regard: my
argument is not that the agency of individuals and organizations are
governmental puppets or by-products of structural forces working
behind our backs. Instead, my argument is that, in real life, our actions
are always already taking place in relation to particular understandings
of the issue calling for action. Moreover, such actions always take place
in relation to the attempts of other social groups seeking to make us act
in particular ways.
By the same token, I think we ought to be cautious in applying the
term meta-governance, at least if that term suggests that attempts at
mobilizing networks can be reduced to the orchestration of a unitary
governor standing above or behind network governance. While the
state may act as an important – if not the most important – relay for net-
work governance, it would be highly problematic to reduce network
governance to the intentions or will of the state. Thus, if power works
through the mobilization of diverse networks, I suggest we see this not
as the orchestrations of a unitary meta-governor, but as the polyvalent
and multifarious technologies employed by a wide range of public and
private actors for often quite different reasons.
Subsequent to these clarifications, we may return to the ways in
which technologies of agency may be instrumental to the formation
and mobilization of governance networks. While the formation, shape
and functioning of these networks vary immensely, which is an impor-
tant feature of their governmental abilities, they nonetheless often seek
to formulate policies based on a certain level of consensus and legiti-
macy by organizing clusters of stakeholders in loose, issue-based net-
works that may cut across national boundaries. These networks need
neither be formalized in the sense of being governed by more or less
universal rules or procedures, nor do they need to be initiated or organ-
ized by public authorities, though the latter may play an important
role by urging certain individuals and organizations to participate and
by providing expertise, facilities and political opinions on proposals
for possible ways of handling the issue in question. In sum, if networks
are regarded as characterized by close interactions between various indi-
viduals and groups with regard to the handling of a specific (policy)
issue, we ought to study the means by which such interactions are
stimulated.
Governing through the formation of autonomy and

interdependencies
A wide range of network scholars tend to emphasize the self-organizing

and/or autonomous character of governance networks. For example,
March and Olsen explain that: ‘contemporary democratic polities …
involve relatively stable, self-organizing networks of interdependent but
partly autonomous actors with resource bases and rule structures of their
own’ (March & Olsen 1995: 70). Bob Jessop takes self-organization to:
include self-organizing interpersonal networks, negotiated inter-

organizational co-ordination, and decentred, context-mediated
inter-systemic steering. The latter two cases involve self-organized
steering of multiple agencies, institutions, and systems which are
operationally autonomous from another yet structurally coupled due
to their mutual interdependence (1998: 29).4
Characteristic for both of these definitions of governance networks is

that they equate autonomy with insulation from state power. A
governance network must have ‘its own’ power distinct from that of the
state to qualify as a governance network. Because of this understanding,
we repeatedly end up spending time resolving the question: who is
governing the network’s actions? The network itself or – through more
or less subtle avenues – the state?
That which is excluded from this line of who-questioning is the how
of governing: how, by what devices, methods and techniques, are the
autonomy and interdependencies of networks formed? This line of
questioning implies that we understand autonomy not as the more
or less perfect insulation from state power, but as a particular form of
governing. More precisely, it entails conceiving autonomy as a specific
constellation between the exercise of power and the exercise of freedom.
My basic premise is that autonomy is never the freedom to do nothing
nor to do anything, but to do something. If we accept the premise that
this ‘something’ is not something merely invented by the network in
isolation from the surrounding world, then it becomes pertinent to ask:
what activities is the ‘autonomous’ network undertaking and what rela-
tions can be identified between these activities and activities carried out
by other groups? In particular, what sort of rapport exists between the
activities of the network and the devices, methods and techniques
involved in the attempt to form and mobilize the self-steering capacities
of that network?
By understanding autonomy not as the absence of power but as a

particular form of power-freedom relation, a space is cleared for analyzing
the diverse instruments of mobilizing, forming and thereby governing
the capacities of networks to undertake particular tasks, functions, and
services ‘on their own’. In other words, the question is not how
autonomous networks really are, but instead: for what purposes and by
what means are the self-organizing and self-steering capacities of this or
that network being formed and mobilized?
This brings us to the question of interdependence, which is a key tenet
in many definitions of governance networks (see above; see also Kickert
et al. 1997: 6). What is defining governance networks, we are told, is not
only that they are (partly) autonomous, but also that they are interde-
pendent in terms of resources, such as manpower, expertise and finance.
In fact, some tend to regard these interdependencies among network
actors as the functional requirement of the complex nature of society
characterized by functional differentiation (Mayntz 1991) or by the
complex nature of the problem to be solved, which leaves public policy-
making in a state of permanent uncertainty (Klijn & Koppenjan 2004:
1–16). I am not disputing the empirical accuracy of these diagnoses;
nevertheless, I believe that by basing our research agenda solely on the
complex or uncertain character of our world in favour of other aspects,
we risk overlooking the ways in which networks are created and shaped
by the manifold attempts to make individuals and organizations partic-
ipate in problem-solving activities by teaming together in a network.
Network formation through legal and financial methods

A few examples may illustrate how dependencies among network
participants can be created through framework legislation or through
the allocation of financial resources. In the Nordic countries, framework
laws have been widely utilized for several decades to support the public
service activities of local authorities and various private actors. In
Denmark, for example, a municipal reform in 1970 entailed the enact-
ment of laws stipulating the powers of local government to collect taxes
and decide on a wide range of key welfare policy areas (Esping 1994;
Villadsen 2000). In some cases, such laws have more or less deliberately
supported the formation of local governance networks. In Sweden, the
central government has applied a combination of framework laws, eco-
nomic incentives and evaluations to structure and steer networks of
regional and municipal authorities involved in the implementation of
care for the elderly (Johansson & Borell 1999). While the network did
not attempt to include new (private) actors, the re-structuring of
existing relations between the public authorities resulted in quite

dramatic changes (not all positive) in the provision of care for the elderly
(ibid. 590). In Denmark, Local Coordination Committees on Preventive
Labour Market Measures were created in 1998 by law with the aim of
increasing the possibilities of employment (Damgaard 2002). The
Committees are supposed to provide advice to the municipal councils
and others (for example, enterprises and unions) and to improve local
cooperation between different authorities and actors. The committees’
activities are financed by an annual grant. A case study of the establish-
ment of a Local Coordination Committee in a Danish municipality
showed that not only were new actors brought into play, including the
organization of disabled people and the medical practitioners, the engage-
ment of the traditional local labour market partners (the municipality,
representatives of employers and employees, and local enterprises) was
effectively modified so as to address a wider social policy agenda
(ibid. 21).
Second, the interdependencies solidifying a network may also be
formed through the allocation of financial resources. In recent years,
governmental ambitions appear to rely increasingly on projects requir-
ing several groups to team up in the identification and/or implementa-
tion of policy issues. The requirement that two or more organizations
team up is found in more than just EU projects and national research
funding. Within urban and housing policies, we also find several exam-
ples of (central) state authorities using the allocation of financial
resources to promote the formation of so-called partnerships between
local groups, such as residents and shop owners (for example Harding
1998; Bache 2000). Needless to say, the network relations formed
between actors applying for public funding for a particular social issue
may be very fragile and last no longer than to the termination of the
collaborative project. Moreover, the actors’ motivation for forming a
(tentative) network may have more to do with pecuniary concerns than
with the embracing of a particular normative agenda (Sherlock et al.
2004). Nevertheless, the point remains that a network of hitherto discrete
actors and the interdependencies between them may be inaugurated and
formed – however unevenly – through the allocation of financial
resources.
Network formation through norms

Objection might correctly be raised that my example is not particularly
representative of the formation of governance networks. By focusing on
a case in which state action, laws or the allocation of public funds
played a key role in mobilizing the network, I have neglected the many
instances where governance networks are formed under the presence of
little or no state intervention. Nonetheless, the formation and mobiliza-
tion of networks may be shaped not only through state action, but also
through norms. I conceive here of norms as codes of conduct that are
always already inscribed in the social practices they inform (see
Triantafillou 2004: 494–7). While certain norms may be propagated by
the state apparatus, it is usually more than problematic to reduce their
existence to the intentions of the state. Even if the state favours certain
norms, it will often be unable to control the diverse practices that they
may give rise to. In particular, norms of participation, transparency and
social responsibility permeate our societies. These norms have informed
the formation of various governance networks, the mobilization of
which cannot be reduced to the actions of the state.
For example, the norms or regulative ideals informing public
organizations and their relations to citizens appear to have changed
significantly over recent decades. The dominating norm informing the
tasks of public organizations no longer seems that of the neutral imple-
mentation of more or less universal welfare programmes formulated and
adopted by politicians, but rather that of specific service provision
and an institutional setting for the nurturing of the self-reliance and
problem-solving capacities of individuals, communities and networks.
Correspondingly, citizens are no longer primarily regarded by public
authorities (and by citizens themselves?) as legal subjects entitled to the
more or less passive reception of welfare services, but rather as a stake-
holder with a legitimate interest in ensuring that public money is spent
efficiently and/or the participation in the formulation and/or imple-
mentation of a certain policy. Citizens increasingly make demands for
transparency in decision-making processes. Thus, one finds that
individuals and groups often establish temporary, issue-based networks,
such as concerned taxpayers, environmental protection groups, disability
groups and labour organizations. On the one hand, these groups may
seek to promote their interests by taking advantage of the new image of
public organizations as being attentive to the concerns of society by
including societal groups in the decision-making processes. On the
other hand, public authorities may take the initiative to mobilize such
networks in order to gain (input) legitimacy to make policies easier to
implement by ensuring that the most powerful, relevant societal groups
are more or less accepting the policies.
The norms informing private (for profit) companies also appear to
have recently undergone a significant transformation. Under headings
such as Corporate Social Responsibility and Management Ethics, it has

become increasingly difficult for private companies to publicly depict
their mission solely as that of seeking to maximize profit (Whitehouse
2006). Instead, ‘serious’ private companies, including many of the
largest transnational corporations in the world, are publicizing glossy
mission statements on their own responsibility for labour standards, the
environment, the local community, etc. Correspondingly, shareholders
are no longer regarded as the sole legitimate purveyors of company
policies, but the larger and much more diffuse group of ‘stakeholders’.
The latter category is open for contestation, and various labour unions,
environmental groups, local community groups and many others seek to
convince the companies via an array of different strategies that they
ought to be considered legitimate stakeholders. Again, the point is that
the changing norms informing the activities of the private companies
appear to favour the formation of diverse networks of citizens and groups
that make claims to be heard and included on specific political issues.
In sum, the conduct of both public and private organizations has
changed significantly during the last few decades and in some cases this
has given rise to more or less institutionalized governance networks.
While the formation of some of these networks – and thereby the inter-
dependencies between its participants – can be rather directly traced to
the actions of the state, the norms of participation, transparency and
social responsibility are clearly informing the actions of numerous other
(private) organizations who may effectively initiate – or at least contribute
importantly to – the formation of governance networks.
Governing the performance of networks
As already noted, various technologies of agency seeking to mobilize

governance networks are always, to a certain extent, shaping the latter
in a certain direction. At the same time, however, at least in principle
they provide a wide room of manoeuvre for the network. One answer to
the problem of ensuring that networks actually undertake the tasks they
are supposed to in a qualified and effective manner is the wide array of
devices and methods that may be dubbed technologies of performance.
This term includes the devices and methods that seek ‘to penetrate the
enclosures of expertise fostered under the welfare state and to subsume
the substantive domains of expertise to new formal calculative regimes’
(Dean 1999: 169).
These technologies were developed, inter alia, in the context of the
contracting out and outsourcing of formerly public activities. In a wide
range of institutional settings, we are witnessing the dispersion of tech-

niques for the devolution of budgets, benchmarking, (self-) evaluation
and assessment techniques that all seek to make organizations and
networks themselves strive to perform better in relation to more or less
stable norms of performance. By urging – and at times even forcing –
networks to constantly evaluate their own performance according to
more or less officially recognized standards and norms, these technolo-
gies may be crucial for the functioning of issue-based networks. Not
only may they allow for a certain room for external steering by central
public authorities, they may also establish a more or less stable common
conception among participants in the network of how things ought to
be conducted and thereby ensure the internal cohesion of the network.
The notion of technologies of performance cannot be neatly separated
from the notion of technologies of agency, as presented above. It is not
that technologies of agency simply nurture diverse capacities to act,
whereas technologies of performance seek to shape and restrict these
actions in particular ways. On the one hand, I take technologies of
agency as always already nurturing a particular form of agency in a par-
ticular direction. On the other hand, I see technologies of performance
as always assuming and at times even seeking to facilitate the capacities
of networks to handle certain issues and problems by themselves.
As an illustration of the interrelationship between network governance
and technologies of performance, I turn to the case of the European
Employment Strategy (EES). In the absence of common legally binding
employment policy, the EES depends on annual National Action Plans
on employment (NAP). These NAPs, which are structured by common
guidelines, are subjected to annual benchmarking, peer-reviews and
form the basis for recommendations issued by the Commission to the
member states on how to improve their employment policy (Hodson &
Maher 2001). While the NAP clearly is not a legally binding policy-
making instrument, the guidelines nonetheless urge member state
governments – in dialogue with local public authorities, representatives
of employers and employees, and possibly other non-state actors – not
only to take actions, but also to reflect on and address their employment
policies in terms of the guidelines. This clearly does not entail a harmo-
nization or standardization of employment policies. Yet when combined
with the annual benchmarking, peer review and recommendations from
the Commission, the process around the NAP urges member states to
conceive their employment problems and the solutions to them in a
particular manner (Haahr 2004). Moreover, the guidelines directly urge
member state governments to develop and stimulate the mobilization of
employment policy networks (Esmark & Triantafillou, forthcoming).

Thus, the guidelines, benchmarking, peer reviews and the Commission’s
recommendations seem to some extent to influence not only the
processes determining who should participate in the network (and who
should not), but also what themes and what broad goals ought to be
addressed by the network.
A word of caution may be appropriate here. While there can be little
doubt about the governing ambitions of the EU and other meta-
governors, one ought to be careful to avoid exaggerating the ability of
meta-governance to control the actual outcome of political processes of
networks. Guidelines, benchmarking, peer reviews and other devices
can doubtlessly generate important political effects, but these can
hardly be controlled by any single actor. For example, the data produced
through the Commission’s benchmarking analyses of the member
states’ employment policies may be used and abused by all parties in the
network to favour their particular interests.
Conclusion
This paper was motivated by a concern regarding the capacity of

governance studies to critically address the governing of networks. This
concern was based on what I perceive to be the proximity between the
problem space of advanced liberal government and the theoretical
imagery underpinning many governance studies. Both tend to base
their understanding of the need for state/public sector reform on a
(structural) functionalist narrative of increasing societal complexity and
the hollowing out of state capacities due to pressure from above
(‘globalization’) and from below (calls for influence by increasingly
vocal public and private groups). In particular, this problem space fits
well with the many political reforms in our societies seeking to mobilize
the self-steering capacities of citizens, organizations and communities
through governance networks.
My first hypothesis stated that the proximity between the theoretical
imagery of many governance studies and the problem space of advanced
liberal government poses certain limits to the capacity of the former to
illuminate the forms of power involved in the latter. First of all, there is
a danger that governance studies, at least in the crudest functionalist
versions, may reduce the calls from our politicians to establish
public–private partnerships or governance networks to little more than
functional reflexes. While this clearly is not representative of all
governance studies, it is worth noting that many of these have little con-
cern for power, except in a purely instrumental fashion, that is, as a
device for improving governance.
Moreover, even if several governance studies do discuss issues of
power, they often end up discussing gains and losses in the actors’ power
resources: how does the introduction of network governance affect the
distribution of power, in particular between central state authorities and
the various networks? One of the problems with this conception of
power as a resource, as a thing that can be held by a particular actor, is
that it entails viewing power as a zero-sum game: if a local network gains
power, it must be at the expense of the power held by a central authority.
However, by framing power in this manner, it tends to overlook the ways
in which advanced liberal government multiplies and spreads through
more subtle (non-legally binding) mechanisms. While legally binding
mechanisms may be less important than in the past – though I would
doubt even that – the capacity to pursue and implement social and polit-
ical goals may actually be enhanced through a wide array of more or less
new governing technologies that stimulate the self-steering capacities of
networks. In short, the governing of governance networks may consti-
tute a plus-sum game if attention is directed to the devices, methods and
techniques of governing rather than the actors. By focusing rather nar-
rowly on actors and their interactions within a network, as many gover-
nance studies do, it becomes very difficult to bring to the fore those
workings and effects of network governance that depend on a wide set of
both legally binding and non-legally binding governing instruments.
My second hypothesis was that Foucault’s conception of government
may redress some of the blind spots of governance studies and critically
address the problem space of advanced liberal rule. Rather than taking
for granted the diagnosis that our societies are characterized by state
overload and increasing complexity requiring new forms of governance
(be that through markets or networks), we may apply Foucault’s
conception of government to address the actual devices by which the
governing of governance networks hinges on the formation and mobi-
lization agency, autonomy and interdependence in very instrumental
ways. My discussion was supported by examples of some of the ways in
which networks may be formed and mobilized by a rather diverse set of
legal, economic and (political) devices and instruments. These devices
not only aimed to mobilize and structure the agency and autonomy of
particular individuals and groups, but also to form particular interde-
pendencies between them so as to form a network with the capacities to
handle particular policy issues. In this manner, the Foucauldian concep-

tion of government may draw attention to some of the instrumental
and productive aspects of network governance that often go unnoticed
in many network studies. The point of such a conception is, of course,
not to condemn network governance or other forms of advanced liberal
rule, but to enhance the space allowing us to address critically what cur-
rently appears to be one of the fastest growing forms of governing in
contemporary societies.
Notes
1. I admittedly use the rather vague term ‘(network) governance studies’, or just
‘governance studies’ to denote the rapidly growing and variegated (political
science / public administration) studies dealing with recent forms of
governing practices in general and the role of networks in these practices in
particular.
2. Some even conceive networks primarily as an instance of interest intermedia-
tion (e.g. March & Rhodes 1992).
3. I use the term technologies of agency in a manner similar to Barbara
Cruikshank’s notion of technologies of citizenship (Cruikshank 1999). However,
I have chosen the term technology of agency to underscore the norms of
activism and participation embedded in these practices rather than the norms
for deciding what is regarded the proper conduct of a citizen. For the same
reasons, I choose to distinguish the term technologies of agency from tech-
nologies of contract and performance (for an alternative categorization, see
Dean 1999: 167–8).
4. For a similar emphasis on the role of autonomy in network governance, see
Rhodes (1997a: 15).
11
Meta-governance as Network
Management
Erik-Hans Klijn and Jurian Edelenbos
Introduction: a network management

perspective on meta-governance
There have been countless publications that have attempted to

conceptualise the trend towards governing through and in networks. In
the analysis of public policy and management, this literature is distinc-
tive because of its focus upon an entire network, rather than the actions
of individual actors (Hanf & Sharpf 1978; Scharpf 1978; Milward &
Wamsley 1985; Kaufman et al. 1986; Kooiman 1993; Kickert et al.
1997; Rhodes 1997a; Agranoff & McGuire 2001; Mandell 2001; Klijn &
Koppenjan 2004a).
Network management and governance

When these publications discuss guidance activities, they make frequent
use of terms like (meta-) governance, management and network man-
agement. Network management has become the most popular of these
referents within the literature, but before elaborating further it is
necessary to clarify the phenomena to which these terms apply.
Meta-governance, often roughly described as ‘the regulation of social
processes’ (see introductory chapter), involves many control and steering
activities. As networks are made up of many actors with different resources
and are characterized by complex interaction processes, they cannot be
easily controlled; this is why we prefer the word ‘guidance’ or ‘facilitation’.
The word (meta-) governance covers a wide array of guidance mechanisms
including self-steering (Rhodes 1997a). This corresponds with the observa-
tion that many outcomes in networks are the consequence of strategic
interaction between actors who have negotiated policy content or desired
measures (Rhodes 1997a). If explicit policy is needed to solve societal
199
problems or to organize service and policy implementation, then most of

the time, self-steering will not be enough. Since we are dealing with
autonomous actors, each with their own perception of the problem and
their own chosen strategy, we need to address the question of collective
action. Cooperation and the coordination of actors efforts requires more
active and deliberate managerial strategies. There remains, however, some
scope for self steering within interaction processes.
This chapter focuses on these strategic attempts to manage (not
control!) ongoing interaction processes between the actors in networks.
It is concerned with attempts to guide not only the pursuit of goals of
individual actors, but rather, influence the interaction between actors.
We call these activities ‘network management’. While network manage-
ment can be identified as strategies or governance, it is significant to
note that not all strategies or governance approaches are equivalent to
network management. Meta-governance, seen as network management,
attempts to deliberatively guide governance processes in networks.
This requires the manager to work out strategies to deal with the
different perceptions, preferences and strategies of actors. Network
management aims to initiate, guide and facilitate interaction processes
between actors (Friend et al. 1974), to create and change network
arrangements with the aim to achieve better coordination (Scharpf
1978; Rogers & Whetten 1982; Mandell 1990; Kickert et al. 1997). It
assumes that a satisfactory outcome is often impossible without active
network management.
Structure of this chapter

In this chapter we address the following main question: ‘What is
network management and what strategies can be employed to realize
it?’. The literature describes various network management strategies. We
shall draw a distinction between process design and management on the
one hand and institutional design on the other (Gage & Mandell 1990;
Klijn & Koppenjan 2004a). Process design and management is con-
cerned with facilitating interaction between actors in policy process. It
requires steering strategies aimed at bringing different perceptions of the
actors together, coordinating interactions, and other activities. The
question to be addressed is what strategies seem to be most suitable and
acceptable in networks situations? The manager accepts the characteris-
tics of the network (position, rules, historical traditions etc.) as given,
and tries to stimulate cooperation between the actors in order to reach
concrete and acceptable outcomes. This type of network management is
discussed in detail in the following section.
Meta-governance as Network Management 201
Network strategies focusing upon institutional design are employed

when the goal is to change the institutional characteristics of the net-
work, e.g. the actors’ positions, the entry rules or other more drastic
forms of intervention geared to adjusting the structure of the network.
The focus is upon identifying suitable institutional design, as well as the
available strategies for effecting it. We deal with these strategies in the
penultimate section. We conclude with a discussion of the skills and com-
petencies that a network manager needs to implement both strategies.
Process design and management: setting up

and facilitating network interactions
Given the problem of collective action, network interactions lack a ‘self

executing’ character (Olson 1965). Cooperation often needs a push from
a process manager, usually an independent person (or group), who can
facilitate the interaction process as impartially as possible. At the same
time, the facilitator should be a person who receives benefits from the
possible results of the interactions or from involved actors.
Process design and process management

According to earlier research, complex processes in networks can only
deliver good and satisfying outcomes if they are intensively supported by
independent process management (Friend et al. 1974; Hanf & Scharpf
1978; Gage & Mandell 1990; Klijn & Koppenjan 2000; Mandell 2001),
and are based upon well-designed organizational guidelines (a process
design) for interactions (de Bruijn et al. 1998; Edelenbos 2000; Edelenbos
& Klijn 2006a). In practice, inter-organizational processes often evolve
together with agreements regarding the nature of the interaction process,
as well as, the participation and ground rules. In network theory these
are known as the process design. Since the process design supports
the interaction of the actors, it is essential that they accept it. The process
design is made interactive by the stakeholders, which also enables it to
become authoritative. There is no standard design or blueprint. The
actual design of the process depends on the situation-specific features in
which the interactive process unfolds. The literature identifies several key
principles of process design (see de Bruijn et al. 1998):
● Openness: all relevant actors are involved in the process at an early

stage. This reduces the risk of a blockage, delivers more information
and thus enhances the quality of the solutions.
● Safety: actors are often wary about cooperating because they are afraid
of undermining their interests. They are only prepared to participate
if their core values and interests are protected. This aspect also has to
be reflected in the process design by, for example, an obligation to
inform the actors of new developments, exit rules for the actors, veto
powers, or (hard or loose) linkages between decisions in the process.
● Progress: provisions need to be made for an ongoing process (e.g.
agreement on timetables, important decisions, the activities of vari-
ous actors, conflict resolution etc.). These criteria need to contain
incentives designed to keep the actors in the process.
● Content: although network processes do not start with a fixed con-
tent (like a solution), content does plays an important role in the
process design. A creative and appealing idea can mobilize actors into
deploying their resources in the network processes. To achieve this, it
is necessary to create variety in ideas. This will enrich the initial ini-
tiative that subsequently evolves with the process.
Every process design is a balancing act between these different

principles, which must also be attuned to the specific characteristics of
the interaction process. The tension between these four principles also
require considerations of trade offs. These can take the form of a
dilemma. The principles should be combined and elaborated in such a
way that they meet the specific needs and problems of the process.
Process design is not ‘self-executive’. It must be developed during the
interaction process, applied, and if necessary, adapted. Together with
other activities, this forms part of process management (Gage & Mandell
1990; Kickert, Klijn & Koppenjan 1997; de Bruijn et al. 1998; Edelenbos
2000). In other words, there is a constant interplay between process
design and process management, which is further complicated by
the fact that the environment in which the process unfolds is in a con-
tinuous state of flux. Hence, the design is not static, but evolves with the
process (Koppenjan 2001). Process management fulfils a crucial role
here. On the basis of theoretical insights, we may expect interactive
processes to yield the best results when the design is constructed well
(ground rules for timetabling, conflict management, responsibility,
roles, etc.) and active process management is employed, which uses the
process design flexibly and focuses on the specific interaction situation.
Empirical research (Edelenbos & Klijn 2006a) shows that an adaptive
process-management style is required to realize satisfactory outcomes
for inter-organizational cooperation. Adaptive process management
implies a reasonably detailed process design that evolves flexibly with
the developments in the inter-organizational process.
Process-management strategies
The literature features an enormous number of process management
strategies for guiding stakeholder interaction (Hanf & Scharpf 1978;
O’Toole 1988; Gage & Mandell 1990; Kickert et al. 1997; Agranoff &
McGuire 2001, 2003; Mandell 2001; Klijn & Koppenjan 2004). These
may be classified as follows:
1. activation of actors and resources: selective activation, coalition

building, building common ground, resource mobilization;
2. creation of organizational arrangements: realizing ad hoc coopera-
tion structures, like sounding boards, project groups, consultation
groups, etc.;
3. guidance of interaction: mediation, brokerage, conflict resolution, etc.;
4. goal-achieving strategies: creating divergence and convergence in the
perceptions of problems and solutions;
5. joint knowledge production: the strategy of searching collaboratively
for information questions and ways to answer them; and
6. trust creation: the strategy of developing a relationship of trust with
other interested actors in the network.
These process management activities do not stand alone, they are

practised in combination. The process manager has to simultaneously
implement a range of different strategies (see Kickert et al. 1997; Agranoff &
McGuire 2001). In other words, he has to activate actors and resources
(Scharpf 1978; Mandell 1990), he has to coordinate goal-achieving mech-
anisms, which includes influencing the perceptions and goals of other
actors, he has to foster or create organizational arrangements to facili-
tate interactions between actors (Agranov & McGuire 2003) and, last but
not least, he has to coordinate the flow of actions and interactions
between different actors (Kickert et al. 1997).
The strategies that are ultimately selected often depend on the course
followed by the interaction process. If the process stagnates because of
conflicting perceptions, energy needs to be invested in developing a
mutually acceptable compromise (goal-achieving strategies). This can be
attained by, for instance, analysing differences, searching for common
ground, generating new solutions or changing and influencing existing
perceptions. When the process stagnates, organizational arrangements
or intensive process management are necessary to address the lack of
linkages between the actors’ interactions. Hence, to improve interactions
one must invest in temporary organizational resources, such as project
groups, consultation platforms or mediators who can re-vitalize the inter-
action.
Activation and arranging: getting the process started

Strategies to activate actors or resources are necessary to start the
process. The literature stresses that the network manager has to identify
the actors needed to set up an initiative and actually create a situation in
which they become interested in investing their resources (see also Lynn
1981). Scharpf (1978) calls this ‘selective activation’. He notes that the
correct identification of participants, as well as the absence of opposition
from other actors who possess the resources to block initiatives, is
absolutely crucial in inter-organizational policy-making. Sometimes the
manager has to try to deactivate actors because their involvement is not
productive. Once the process gets started it is necessary to clarify the
actors’ goals and perceptions (Forrester 1989) and to invest time and
money in developing solutions that create opportunities for participa-
tion Klijn & Koppenjan 2004a). Sometimes, however, the process lacks
creative solutions capable of satisfying the various actors. In such cases,
more variation needs to be introduced by, for instance, inviting different
teams of experts to compete against each other in formulating solutions
(Teisman 1997).
The creation of temporary organizational arrangements to facilitate
interaction is also important (Rogers & Whetten 1982). The transaction
costs of these arrangements must, of course, be kept to a minimum
(Williamson 1996), but at the same time the actual arrangements have
to be acceptable to the actors.
Goal-achieving and guiding strategies: managing
content and interactions
Goal-achieving strategies are primarily focused upon generating
solutions and influencing the actors’ perceptions of the nature of the
problem and the possible solutions. If different actors turn out to have
different perceptions and favour different solutions, then a process of
exchanging and converging perceptions is often necessary to achieve
joint action (Klijn & Koppenjan 2004a). This is achieved by strategies
aimed at, for example, clarifying perceptions, seeking goal congruence
or combining goals in different packages. It can also be achieved by
introducing new actors or new ideas to get more variation in the
perceptions.
Whereas the management strategies for goal achieving have a strongly
cognitive character, the strategies for managing the interactions have a
strongly social character. They include, among other strategies, mediating
between actors, maintaining an attractive agenda, maintaining commu-
nication with the outside world, and mediating and mitigating conflicts
(or trying to use them for ongoing interactions). Basically, all these
strategies are aimed at facilitating and speeding up the interaction
process.
Joint knowledge production: supporting

the search for a shared content
More and more articles are appearing on joint knowledge production as a
network management strategy (Ehrmann & Stinson 1999; van Bueren et al.
2003; van Bueren & Edelenbos 2004). Joint knowledge production is a
type of process strategy that helps to develop the content of interaction
processes in networks. It is based on the assumption that non-recognition
of unreliable knowledge can generate certain problems when knowledge
needs to be produced for inter-organizational cooperation. Experience
has shown that it is very difficult to produce knowledge that is accept-
able to most parties (Lindblom & Cohen 1979; van Bueren et al. 2003).
In complex processes actors spend most of their time challenging one
another’s research, trying to prove the contestability of premises or the
inadequacy of a database and so on. This results in ‘knowledge disputes’
and piles of reports presenting contradictory conclusions, both of which
seriously hamper the quest for well-negotiated and shared knowledge in
networks. Stakeholders with differing viewpoints and interests must
work together in joint fact-finding exercises to develop data and infor-
mation, analyse facts and forecasts, develop common assumptions and
informed opinions. Finally, they must also use the information they have
collected to reach decisions together (Ehrmann & Stinson 1999: 376).
To fill gaps in knowledge and to eliminate knowledge disputes,
independent experts and interested actors (such as private parties,
authorities and interest groups) draw up a research design in mutual
consultation. The research design is, in effect, the outcome of a process
of discussion and negotiation between stakeholders and external experts,
rather than something that has been laid down in advance. Together
they search for workable methods in the quest for knowledge and the
guiding principles, assumptions and suppositions on which these
methods are based. They also ratify by mutual agreement the (funda-
mental, temporal and geographical) system boundaries and the scope of
the study (e.g. When will the study begin and end? Which effects will be
included in the study? Which criteria will be used to evaluate these
effects? Which subjects will be part of the study, and which will not?). If
the stakeholders are unable to decide which methods should be used,
they may opt for several (competing) methods and/or sensitivity
analyses to determine the extent to which the outcomes will vary for the
different assumptions on which the different methods are based. They

may also decide to integrate various research models. Finally, they may
decide to set up a ‘Committee of Wise Men’ comprising independent
experts from different disciplines and charged with the task of settling
persistent knowledge disputes.
Generating trust: facilitating and

consolidating cooperation
Trust is often cited as a contributory factor in smoothly running net-
work processes. It is seen as the answer to complexity in networks (Lane &
Bachman 1998; Rousseau et al. 1998; Nooteboom 2002). Basically, trust
implies a more or less stable perception by the actors of one another’s
intentions; hence, they refrain from opportunistic behaviour. Trust has
several potential benefits. First, it can facilitate and cut the costs of coop-
eration (fewer transaction costs). Second, it consolidates cooperation
and makes the actors more willing to invest in it, thereby improving its
robustness. Third, it enhances the results of cooperation. More problem-
solving and innovation capacity is generated because the resources of
the actors are combined (Edelenbos & Klijn 2006b).
If trust is so important, then we need to know how it can be realized so
that we can use it to advantage. A number of process-management strate-
gies have been distinguished in this domain (Edelenbos & Klijn 2006b):
● Intensify interactions: the expectation of beneficial transactions in the

future stimulates the emergence of trust (Axelrod 1984). Actors must
also feel that there is reasonable reciprocity in the efforts they make and
the risks they run (Chiles & McMackin 1996). It is also important to dis-
play the efforts and performance of individual actors and to formulate
common criteria for judging them. This way, the perception of perform-
ance, reputation and mutual trust can grow (Deakin & Wilkinson 1998).
● Stabilize and manage interactions: vehement disruptions and frequent
contract renegotiations can negatively influence performance and trust
(see e.g. Milward & Provan 2000a). It is important to stabilize interac-
tions and expectations. Research has revealed a relationship of mutual
reinforcement between trust and collective problem-solving (see e.g.
Sako 1998). Active process management is needed to achieve this.
Institutional design: changing the network
Institutional design is the second network-management strategy

addressed in this chapter. It is an indirect strategy (unlike process
management which is more hands-on and can therefore be seen as a
direct strategy), which focuses on realizing institutional changes in the

network. Institutional design is based on the assumption that the insti-
tutional characteristics of the network influence the strategies and
cooperation opportunities of the actors. It attempts to change one or
more of these characteristics (see Klijn & Koppenjan 2004a). Very little
has been written about institutional design in networks.
What is institutional design?

Institutional design strategies are usually aimed at changing formal or
informal rules in networks (Ostrom 1986; Dimaggio & Powell 1983;
Klijn 2001). Rules lie at the heart of institutions. Many authors even see
them as the defining characteristic. For instance, Scharpf (1997: 38)
describes institutions as ‘systems of rules that structure the course of
actions that a set of actors may choose’. Institutions are thus, in short,
sets of rules, which influence, guide and limit the behaviour of actors. In
this sense, networks may also be regarded as institutions: they are not
only patterns of social relationships between mutually dependent actors
but also systems of rules. Therefore rules can be considered a part of net-
works that also give them institutional meaning. Networks are charac-
terized, after all, by specific and unique sets of formal and informal
rules. Each network has its own history, in the course of which, rules
have been formed and subsequently undergone development (Klijn,
1996, 2001; Burns & Flam 1997; Scharpf 1997).
The rules of the network are ‘activated’ by the actors, who apply them
in their interactions. This does not mean that these rules are always clear
or even fully known to the actors. On the contrary, rules are often
ambiguous and require translation in the interactions (March & Olsen
1989). Just as a judge tries to interpret a specific case (offence) in the light
of an existing rule, an actor in the network tries to interpret the meaning
of events in the light of the network rules that he is familiar with. It is in
this way that network rules influence the interactions in the network.
Although rules may be the product of deliberate design behaviour,
they are still only valid if the other actors in the network recognize them
as such. This immediately reflects the essential difference between social
and physical rules. Social rules, and thus also rules in networks, can only
continue to exist if they are adhered to by the actors and are truly
respected in concrete game behaviours. Rules that are broken by the
actors, either consciously or unconsciously, or are not (or no longer)
respected, lose their validity (Burns & Flam 1997).
Institutional design aims specifically at bringing about changes in the
institutional characteristics of networks. It therefore needs to be seen in
a separate light from the gradual changes in institutional characteristics
that occur through adaptations over time or because institutional arrange-

ments lose their regulative power. These emergent, almost unconscious,
processes of (de-) institutionalization are more or less by-products of
the strategic games that actors play (Giddens 1984; Ostrom 1986).
Institutional design refers both to the activity of trying to change the
institutional features of policy networks, and to the content of the insti-
tutional change that is targeted. Given that we defined institutions in
terms of rules, institutional design is the deliberate attempt to change
the rules (either formal or informal) of networks.
Since most rules develop gradually over time or, in the case of formal
rules, are created in complex institutional arenas, institutional design
strategies are time-consuming and difficult to implement. As a result,
they are unsuitable for achieving changes in policy games that are
already underway.
Types of strategy for institutional design

There are roughly three types of strategy for institutional design (see
Klijn & Koppenjan 2004a). First, strategies aimed at the network composi-
tion: these strategies attempt to change or influence the composition of
the network. They are based on the premise that (changes to) the com-
position of the network has (have) an influence on the interactions (and
thus the outcomes) occurring within it. There are various ways in which
the composition of a network may be changed, for example, by consol-
idating or changing the actors’ positions or by adding new actors.
However, strategies may also be aimed at changing the access rules for
actors or at influencing the network as a whole. This can be pursued by
promoting network formation and self-regulation, or through modifica-
tions to the system. The strategies range from relatively light interven-
tions, such as laying down the actors’ positions (which merely confirms
achieved and existing rules) to more encompassing interventions, like
system modification. System modification not only involves a wider
range of rules, it also has a deeper impact on the position and identities
of the actors. This generally inspires greater resistance.
Second, there are institutional strategies aimed at the network
outcomes. These strategies try to influence the standards or the logic of
costs and benefits in a sustainable way, which has the consequence that
the games within networks then evolve differently because other strate-
gic choices are made. The target of intervention here is therefore not the
actors as in the previous example, but their choices. A sustainable
influence needs to be exerted on the actors’ strategic choices and the sub-
sequent outcomes. The most important institutional-design strategies in
this category are strategies to change the pay-off structure (financial or

other rewards that are connected to strategies and decisions), strategies
to change professional codes (standards by which the actors see their
professional activities and identities) and strategies to change evaluation
criteria (standards by which the actors judge the outcomes). The two last
types are closely connected.
The third category consists of strategies aimed at network interactions.
These strategies try to influence the rules that regulate the process in
networks. The aim is to facilitate interaction between the actors, create
a framework and establish linkages. Interventions in this category
include conflict-settlement mechanisms (to regulate conflicts between
actors) and the introduction of certain procedures (thus fixing a specific
interaction or decision sequence). Strategies such as certification
(standards of quality attached to the characteristics of an actor or his
relation to other actors) and influencing supervisory relationships also
fall into this category.
Institutional tensions: the example of interactive

policy development
As said before, strategies to change the institutional context of the
network are not easy to implement and therefore need to be seen as
highly delicate operations. Although many definitions and descriptions
of institutions stress perpetuity and stability, we expressly wish to counter
this with the possibility of short-term variants a.k.a. ‘proto-institutions’
(see Lawrence et al. 2002). This is because our network perspective
recognises the dynamic character of institutions. Deinstitutionalization –
rather than institutionalization – comes into play. In our view, a temporary
process like interactive policy development (but other processes as well)
often imposes a temporary institutional structure (methods, phasing,
rules and roles for stakeholders) on top of, or alongside, existing institu-
tions. This then has implications for the current situation (Edelenbos
2005).
Experience (in the Netherlands) of the introduction of interactive
policy development (citizen involvement) illustrates this observation.
The introduction of interactive policy development may be accompa-
nied by the re-evaluation and repositioning of political and administra-
tive institutions. Interactive policy development admits a form of direct,
deliberative democracy (King 2003; Verweij & Josling 2003) in a repre-
sentative democratic system. This may cause a certain amount of insti-
tutional tension within the existing administrative system (Edelenbos
2005). New ground rules and actors’ roles accompanied by interactive
policy-making are often rejected by the existing administrative system.

This is because it is primarily based on norms of representative democ-
racy, such as hierarchical accountability and the ‘primacy of politics’.
Empirical analysis (Edelenbos 2005) has revealed that institutional
dissociation occurs between the interactive process and the existing
administrative structures and procedures. We were struck by the way the
temporary procedures of the interactive process differed greatly from
the existing procedures, and moreover often existed in isolation from
one another. In addition, the roles of the various actors appeared to be
extremely rigid. The way in which public servants, politicians and
administrators, but also citizens and interest groups, perceived their
roles, did not seem to be all that open to change. Traditional role pat-
terns dominated the fulfilment of new roles in interactive policy devel-
opment. Politicians adopted a very reluctant attitude and remained
safely entrenched behind their familiar institutions. Public servants saw
the interactive process primarily as a political whim, which has to be
tolerated and skewed wherever possible in the direction of standard
procedure in the course of the process.
If a form of interactive policy development is permitted within an
existing institutional framework, then roughly speaking two responses
may occur during and after the interactive process, institutional intran-
sigence or institutional change. In the former case, institutional intran-
sigence, the existing institution has a decisive impact on the new
institutions of the interactive process. The existing institutions sweep
away the new ones or ignore them. The representative democratic sys-
tem continues to dominate. In the latter case, institutional change, the
new institutions of the interactive process have a decisive impact on the
existing institutions. The new institutions are absorbed by the existing
ones, merging into a hybrid, where the two exist side by side. Another
possible scenario is that the new institutions completely overrule the
existing ones. A third scenario could be that the existing and new insti-
tutions merge into a symbiosis, something new that neither can be
deduced from the existing nor the new institutions.
Hence, when we look at the question of institutional change we
assume that there is a reciprocal relationship between existing and new
institutions. New institutions of interactive policy development have an
impact on existing institutions, but existing institutions, likewise, have
an impact on the new institutions of interactive policy development.
Conclusion: institutional design as process

The strategy of institutional design requires special attention for the
institutional interrelationships between existing and new rules. New
institutions need to be somehow adapted and incorporated into existing

ones if they are to have any chance of being ‘picked up’ (Weimer 1995;
de Jong 1999). But how can such an adjustment take place? There is a
dilemma between, on the one hand, the risk of rejection of new institu-
tions because they differ too much from the status quo. And on the
other, the risk that the change will be so small (when new institutions
are too similar to the old ones) that experiments with interactive policy
development, which are often announced with a great flourish, will
have very few institutional consequences (Edelenbos 2005). New insti-
tutions must, to a certain extent, be in line with existing ones, but at the
same time they have to be sufficiently different to be capable of realiz-
ing institutional change. Healey (1997: 268) argues along the same lines:
… new ideas and organizing routines need to grow from the specific
concerns of stakeholders. (…) They must develop with the grain of
local contingencies. Yet, to carry transformative power, they must
have the capacity to challenge existing conceptions and re-frame
ways of thinking, ways of valuing and ways of doing things.
By definition, institutional design is a search in each individual

situation for the norm of the minimum allowed institutional deviation.
A workable balance has to be struck between the rejection and absorp-
tion of new institutions in creating networks (Edelenbos 2005).
Good network management: skills and competencies
Publications on the network approach invariably stress that network

management is far from easy. It requires knowledge of the network and a
range of competencies, not least negotiation skills. This is because net-
work-management strategies are conducted in a context of mutual
dependency. Thus, as said earlier, a network manager is not a central
actor or a director, but rather a mediator and a facilitator (Forester 1989;
Gage & Mandell 1990). A network manager cannot (and should not try
to) control all the interactions in the complex network. The interactions
in networks are far too numerous and diverse for that. Complex
processes often have their own dynamics, and are to some degree self-
organizing. However, the self organizing aspect of complex processes
does not mean that there is no need for deliberate network management.
Complex processes are often lengthy due to deadlocks, disputes and so
on. There is often a desperate need for actors who are committed to the
process to be guided to generate fresh and promising solutions for press-
ing societal problems. This guidance includes connecting the various
actors and the decisions within the network. To this extent, network
manager must be selective and pick the right moment to influence inter-
actions and proposed solutions. Network management is often the art of
bringing together different streams, such as perspectives of actors, their
problem definitions and solutions, at a time when this can make a dif-
ference (see also Kingdon 1984). Network management becomes dirigisme,
when the network manager’s presence is too dominant in the interaction
process and blocks emergent positive movements and couplings.
Required managerial skills

Interaction management in networks is a difficult job that requires a lot
of effort, interaction and commitment. It also requires the manager to
know his network. First of all, a network manager needs analytical skills
so that he can comprehend and fathom a complex network of actors
with different sorts of relations, perceptions of problems, solutions, etc.
Second, he must have some substantive knowledge of the issues dealt with in
the network. This involves being able to ask the right questions without
personally having to come up with the right answer (de Bruijn et al.
1998). Third, a network manager must have the ability to make a workable
and authoritative design for network interactions. He must be able to use
the network analysis to construct a design (with process agreements) for
the network interactions. Fourth, he must have the skills to estimate,
choose and implement the right network strategies. Design and management
are two sides of the same coin and lie at the very heart of network man-
agement. This calls for an adaptive management style; the process man-
ager must first design the network interactions carefully by paying close
attention to the time phases of the process, the determination of the
budget, the role allocation, the method of conflict resolution, accounta-
bility, substantive frameworks, auxiliary conditions, etc. All of this
implies that the network interactions are regulated and choreographed
somehow prior to the actual implementation of the process. The process
manager must also apply an adaptive style in the implementation of the
process design, which must be carried out with an eye to the specific
conditions and developments in the network situation. He must change
the ground rules and the roles to ensure that the network processes
unfold smoothly (Edelenbos & Klijn 2006a). A network manager
(or more than one network manager) needs to play an active role to
achieve interesting results (Gage & Mandell 1990; Alter & Hage 1993;
Agranoff & McGuire 2003). We are not suggesting very heavy and
intense network management; the (self-regulated) dynamics of the
network processes determine for a great part the evolving interactions.

However, a carefully thought-out process design can prevent non-
productive dynamics. Basically, the design must be continuously adapted
to the complex dynamics of the network processes. Design and manage-
ment must be seen (and treated) as a continuous interplay.
We mentioned in the previous section that institutional obstacles can
also hinder successful interaction (Klijn & Koppenjan 2004a; Edelenbos
2005). Many studies indicate that a lack of active strategies on the part
of the network manager contribute to failures in interaction (Hanf &
Scharpf 1978; Mandell 2001; Bueren, Klijn & Koppenjan 2003). A good
network manager has the ability to very quickly assess the network condi-
tions and to identify and assess the explicit and – more importantly – the
implicit rules. This is the fifth skill required of the network manager. On
the basis of this assessment the network manager must consider the fea-
sibility of the newly formulated rules and roles for the network game. It
all boils down to a balancing act between trying something new,
together with ‘business as usual’.
Research challenges
We conclude this chapter by pointing out that there are still many
unanswered questions about the skills and competencies required of a
network manager. Meier and O’Toole (2001) conducted research on
success and failure in different educational networks and found that
managerial networking was positively correlated with primary goals, but
also with other indicators of organizational performance. They did not,
however, look at specific managerial strategies discussed in this chapter.
Nor did they ask questions about the skills and competencies needed for
successful network management.
Available case study material demonstrates the importance of network
management strategies, but we still know very little about which strate-
gies seem to be most effective in which situations and – more impor-
tantly – how they can be successfully implemented. We therefore need
further research which can provide systematic information on the actual
behaviour and the tool kits of network managers. We also need to gain
more insight into the conditions under which managers make a success
of steering complex decision-making processes in networks, as well as
the actual effects of network-management strategies. The multifarious
characteristics of networks and the wide range of strategies within the
network games make this a difficult research assignment. It will

probably necessitate additional research methods besides surveys and
case studies. We suggest the deployment of action research methods,
which will allow the behaviour, skills and competencies of network
managers to be closely observed and monitored.
12
Governing Outputs and
Outcomes of Governance
Networks1
Laurence J. O’Toole, Jr
How is it possible – indeed is it possible – for democratically elected

governments to shape the outputs and outcomes2 of network gover-
nance? What role can be sketched for public authorities who may have
reason to attend to the direct or ultimate products of governance
networks? How are networks likely to respond? Can public authorities
operate at a meta-level to shape network results in the interests of the
broader public? What dilemmas and complications can be foreseen?
This chapter analyses the interaction between governance networks
and public authorities, particularly with regard to the latter’s efforts to
regulate or otherwise influence policy outputs and outcomes in network
settings. It does so by situating the subject in the context of a model of
governance that suggests the likely influence but not control available
to public authorities over such results. Once this framework is sketched,
the chapter examines the notion of meta-governance and distin-
guishes two such concepts. Working primarily from one of these, the
chapter then offers a game-theoretic heuristic to analyze the ways that
governments can shape the outputs and outcomes of networks.
Framing the subject
Governance networks can offer advantages in delivering outputs and

outcomes – enhancing political legitimacy among interested stakeholders,
increasing technical capacity to produce valuable results, and leveraging
administrative capacity to generate reliably successful implementation
(O’Toole 1997). In addition, it is clear, reliance on governance networks
is often inevitable and is quite routinely an explicit choice of legislative
bodies (for systematic evidence from one country, see Hall & O’Toole
2000, 2004).
215
Despite their utility and relative ubiquity, nonetheless, governance

networks offer certain significant drawbacks. In a sense, these can be seen
as the other side of the coin from the advantages of bureaucratic systems.
The latter are often unresponsive to direct influence by stakeholders,
whereas governance networks have their outputs and outcomes shaped
by the networked stakeholders who coproduce results. Whereas that form
of action may seem more accessible and responsive to those in the
network, it can also be less sensitive to those not represented directly
there, including the relatively diffuse or unorganized interests in the rele-
vant community (see Lowi 1979), and less mindful of any overall notion
of the public interest. Second, whereas hierarchical institutions can some-
times move slowly and pay considerable attention to procedural formality
and regularity, governance networks are less regularized and transparent
arrays. They can be fluid and responsive to stimuli, but at the expense of
certain kinds of fairness and usually at the expense of coherent and visi-
ble forums for access by those who are not extant parts of the club. Third,
bureaucracies tend to operate as autoregressive systems, that is, they oper-
ate today (and produce outputs today) in fashions that greatly resemble
the patterns of yesterday. Bureaucracies are typically considerably more
stable systems than are networks (O’Toole & Meier 1999).
Meier and I have developed inductively, from the empirical literature,
a general model of public management and performance that expresses
this reality. Explicating the model briefly can help to focus on some
relevant features of governance and networks:
Ot 1 (SM1) Ot1 2 (Xt/S) (M3/M4) t
where
O is some measure of output or outcome,
S is a measure of stability,
M denotes management, which can be divided into three parts
M1 management’s contribution to organizational stability through
additions to hierarchy/structure as well as regular operations,
M3 management’s efforts to exploit the networked environment,
M4 management’s effort to buffer shocks emanating from the
network,
X is a vector of environmental forces,
t is an error term,
the other subscripts denote time periods, and
1 and 2 are estimable parameters.
Governing Outputs & Outcomes of Governance Networks 217
The model is inertial, nonlinear, and contingent. It represents formally

the production of outputs or outcomes and indicates, for instance, that
system stability (ranging from rigid hierarchical patterns to fluid,
networked arrays) is a critical determinant of performance. It also indi-
cates some points of leverage in shaping such results. Several managerial
functions play roles, when operating in and on such systems from one
or another level, and also that forces external to (for instance) a network
itself – the ‘X’ vector – can be crucial in shaping what happens. One
component of this vector can consists of the effort and direction of
influence applied by public authorities onto the functioning of the
system of interest.3
Stability has its obvious advantages, as analysts since Weber (1946)
have noted, as well as some recognizable drawbacks. Less stable network
patterns can adapt more flexibly but are also more likely to be subject to
pressures for cooptation from organized interests in their interdepend-
ent environment (see O’Toole & Meier 2004). In general, then, their net-
work outputs are more easily shaped, or attenuated, by strategic action
of parties seeking to manage from within the network (or pressing on it
from outside).
Contrariwise, this logic suggests that governmental ‘control’ over
network outputs and outcomes is likely to be a chimera. Multiple actors,
including public authorities, may have the capacity to influence the
results of network action, but none of them can successfully exert
unilateral direction.
Furthermore, the logic also suggests that whatever influence a
government is actually able to exert is subject to attenuation over time.
In short, there are no autopilot possibilities. Rather, one can expect a
continuing tendency, unless checked, toward weakened governmental
direction over the network production.
There are several reasons. First, governments sometimes intentionally
seek and prefer attenuated influence. Public authorities have reasons to
distance themselves from certain controversial or difficult issues (for an
analysis, see O’Toole & Meier 2004). They may even prefer to outsource
production to a network rather than handling these responsibilities in
house. Second, sometimes governments coproduce outputs and out-
comes with networked others; if so, the use of lengthy principal-agent
chains almost certainly results in reduced influence over actors closer to
the final stages of production. Third, cooptative tendencies in networks
can work to reinforce inequities present in the broader social setting,
and governments are likely to have difficulties reducing such dynamics
(O’Toole & Meier 2004). Overall, then, stable and active governmental
regulatory paths can tend toward atrophy during processes of imple-
mentation via governance networks.
This conclusion does not mean that government is powerless when
dealing with networks. Later in this chapter, a number of points of
governmental leverage in and over networks are identified. It does mean,
however, that realism should counsel public authorities to forsake strate-
gies of control for more nuanced and sometimes indirect approaches
based on an understanding of how networks actually operate.
In the remainder of this paper, the notion of meta-governance itself is
first explicated. A heuristic is then offered to help analyse the potential
for governments to shape the outputs and outcomes of networks.
Finally, some conclusions can be offered.
Two notions of meta-governance
If one is interested in meta-governance of networks, it can be important

to distinguish issues connected to two different albeit related types of
‘levels’. The first alludes to public authority as a kind of superordinate
decision maker. The idea of the democratic state as holding, at some
point, a monopoly of coercion to be exercised in the public interest
seems to require that governing authorities do more than preside in the
midst of or participate in joint decision making but rather stand at some
remove, at least on critical issues.
Various ways of conceiving of this normative, and perhaps empirical,
‘differentness’ represented by public authorities can be sketched. One,
linked to the traditional notions of a liberal society, would be to con-
sider public authorities as providing room for discretion on the part of
other social actors – whether networked or otherwise – in the ‘private’
realm. The public authorities, from this perspective, set the parameters
within which other actors may exercise freedom to maneuver and
decide, and the authorities operate to check any excesses on the crucial
issues. Another would be close to the game-theoretic concept of last
mover, coupled with a default mechanism. That is, social actors can
work jointly to decide and act, but public authorities are free to make
the last or definitive move in the multi-actor game; and should social
actors not be able to jointly produce acceptable outputs and outcomes,
government decides by default. A third related way of characterizing
this sort of meta-role for public authorities is to think of government as
sitting at the apex of a societal hierarchy, with ultimate responsibility to
stop or undo inappropriate actions by others. Various concrete special
powers accorded to public authorities – ratifications, vetoes, amend-

ment powers, and so forth – are then presumably justified by this kind
of supervisory role. In any of these notions, the issue then becomes: how
can governments operate to shape or check the decisions and actions of
governance networks when such an effort is justified?
In considering this question, however, it can be helpful to make use of
another meaning of meta-governance. In their well-known analysis,
Kiser and Ostrom (1982) sketch three ‘levels of action’. The operational
level encompasses concrete moves to take action in the world. The leg-
islative or policy level – which can entail moves by actors far beyond
actual formal legislatures – sets rules (that is to say, policies) about how,
where, and when various operational moves can happen. Finally, the
constitutional level of action sets the understandings about how policies
are made or changed. Although later work by Ostrom emphasizes that
social systems may operate on more levels than even these three (1999),
those enumerated here are sufficient to make the key point here regard-
ing meta-governance.4
In particular, this formulation emphasizes distinctions among levels
of action rather than among types of actors. Any social actor might
spend time and effort working in two or even all three levels; indeed,
some specific moves or strategies might even function simultaneously on
multiple levels of action. This idea carries a number of ramifications, only
one of which needs to be emphasized here. Whereas meta-governance in
the earlier-summarized conceptualization conjures the notion of public
authorities as sitting above or apart from the actions of networks, the
Kiser–Ostrom formulation suggests that the ‘apart-ness’ to be empha-
sized is best considered a kind (level) of effort developed by whichever
social actors become involved in setting or changing the basic rules of
the game. That is, thinking of meta-governance as constitutional deci-
sion making, and perhaps certain kinds of policy decision making
(including certain moves by managers), brings to the fore a more com-
plex reality of meta-governance that needs to be treated seriously: pub-
lic authorities may not be the only ones working at the meta-level. In
fact, it becomes an empirical question as to whether public authorities
are at all involved at the meta-level; an alternative might be that other
social actors sometimes set the basic rules of the game that ultimately (at
another level of action) produces outputs and outcomes.
So what? There are several ramifications. Networks themselves may be
capable of anticipating (or responding) and enacting rules to shape
outputs and outcomes. For instance, when operating at the legislative
level (in Kiser’s and Ostrom’s sense), networks of actors can collectively
agree to impose restrictions on certain operational actions allowable to

all the actors in the network to reduce second-order collective action
problems – free riding, overuse of common pool resources, and so forth
(Ostrom 1990). In this type of situation, self-organizing efforts of net-
works constitute, in effect, meta-self-governance. Indeed, it should be
clear that under appropriate circumstances, meta-governance need not
be conducted solely or primarily from outside; a network, or parts of a
network, can meta-govern the network. As Ostrom herself has docu-
mented in great detail (1990), cases spanning hundreds of years on sev-
eral continents demonstrate that governments need not necessarily
even be involved in some successful kinds of meta-governance.
In addition, various network actors may themselves seek to take
action at one of the meta-levels, even if not collectively. Different actors
may have agendas that would be furthered by, for instance, an alteration
in a legislative-level decision. Pursued separately, meta-moves may be
attempted by several network actors even simultaneously. One could
think of this possibility as involving different actors pulling, or enticing,
their network collaborators in different directions, more or less as force
vectors operating at varying strengths and in various directions. The
ultimate meta-outcome in such an action space would be the sum of the
vector forces, which would be a resultant rather than a consciously
selected end state that had been chosen by any actor. Note that this vari-
ant of meta-governance could also develop without public authorities
being directly involved in any of the meta-moves.
Finally, there is the role of government itself in actions at the meta-
levels. Certainly governments can become active in shaping legislative
and even constitutional choices in governance networks. If and as they
do so, meta-governance in these two distinct senses becomes closer to
the same notion. Even here, nonetheless, matters are unlikely to be sim-
ple. Multiple different public authorities of the ‘same’ government may
seek a role, and often formal superiors refrain from intervening to
impose a coherent governmental position. Also, a given government
often sits in a nested array of multilevel governmental systems –
whether as in a federal arrangement like that structuring the 85,000
governments within the USA or in the evolving system of the European
Union.
In addition, when public authorities do become involved in efforts at
meta-governance of the Kiser–Ostrom variety, they are unlikely to be
alone. Other actors are also interested in shaping actions at the
meta-levels, even if a public authority is primus inter pares. Actions
aimed at meta-governance thus typically involve actions by public
authorities to shape the context and operations of the setting in which

they themselves are interdependent. This statement holds, whether one
is talking about corporatist settings in which some notion of partnership
is explicit, or in pluralist systems, where conventional wisdom has it
that government should not be considered ‘just another’ social partner. In
a wide variety of democratic settings, the possibilities for meta-governance
via government are real but hardly unilateral.
I approach the issue of meta-governance primarily by adopting the
Kiser–Ostrom conceptualization, albeit with the clear recognition that
some empirical circumstances allow for public authorities’ adopting a
role that involves a kind of standing-apart, and some sensible normative
claims justify such authorities being able to do so on key issues. The two
notions are not explicitly at odds, but the Kiser and Ostrom explication
is more likely to capture the actual, consequential, rules-grounded
moves that in turn shape the context of networked action.
Meta-governance in action
The foregoing coverage strongly suggests that a consideration of the

prospects for meta-governance by public authorities should not be con-
ceived as a top-down affair (for instance, Mazmanian & Sabatier 1989),
nor as a radically bottom-up process (Hjern & Hull 1982a). In the field of
policy implementation, most researchers have sought to work from a
theoretical perspective combining insights of both approaches while
also avoiding some of the weaknesses of each. Similarly here, such a
strategy seems apropos.
Unfortunately, it is easier to prescribe an appropriate melding of
top-down and bottom-up ideas than to develop such an approach in
detail. Several candidate approaches, however, have been put forward. A
promising route has been sketched by Bressers and colleagues via a
‘contextual interaction theory’. Here, the basic insights of bottom-up
analysts about the importance of interaction processes has been linked
in detail to a well-developed set of ideas about policy instruments
(Bressers 1983, 2004; Bressers & Klok 1988; Bressers et al. 2000).
In the current chapter, a framework linked to this last-mentioned one
is adopted as a heuristic. Game theory, I argue, can be employed to help
understand the options available to governments if they are interested
in shaping performance in and through networks. In this sense, there-
fore, the analysis in the chapter is built upon an implicit foundation in
governability theory, as explicated in the introductory chapter. In
particular, the heuristic used here clearly does emphasize calculation on

the part of network actors, although such an assumption does not preclude
the notion of cultural ties. Similarly, the perspective assumes at least
occasional conflict (differences in interest) among the actors, although
‘coordination games’, to use the game-theoretic notion, are also possible
under some circumstances.
The emphasis here, it is important to note, is on game theory used
heuristically rather than, in a strict sense, as a theory aimed at prediction
(see O’Toole 2004). Game theory has been explored for similar purposes
in other kinds of settings (see Lynn 1993; O’Toole 1996; Scharpf 1997).
While game-theoretic approaches implicitly assume equality of power
between or among actors or actor constellations, and thus represent a
set of circumstances somewhat narrower than could be expected in the
real world, where governmental actors may have at least formal power
advantages on some decisions (Bressers et al. 2000), this constraint does
not pose a serious limitation to the use of the approach heuristically.
Where power is shared among a set of interdependent actors, a relatively
common circumstance in networks, the approach does offer some
important advantages. First, it builds a melding of top-down and
bottom-up insights into the logic of the perspective itself. All actors are
treated as strategic decision makers and therefore must have their pref-
erences taken into account. Second, the outputs and outcomes of game-
theoretic interactions are driven by the full set of actors and their
preferences – the result may, indeed, be something that no one had pre-
ferred. These fundamental notions fit the circumstances of many real-
world efforts to produce action via networks. Third, the foregoing
coverage suggests that government as actor is best represented as
enmeshed among other social actors rather than as standing apart from
them; either way, however, game theory offers appropriate analytic
representations. Government-as-standing-apart can be represented as a
two-party game (government and network) or, more complicatedly, as
a two-level set of nested games (a within-network game connected to
a linked game between network and government). Government-as-
network-member can be modeled as an n-party game within the
network.
Finally, while the heuristic of game theory is rather abstract, it is
employed here in preference to a catalogue of policy instruments avail-
able to government for shaping network results. Many different typolo-
gies of such instruments can be found in the literature, but virtually all
are ad hoc and oversimplified depictions of what are actually a large set
of much more nuanced and varied tools. These, in turn, can be
understood and compared by representing them on several different

dimensions – provision or removal of resources to target groups, pro-
portionality of governmental response to target-group behavior,
amount of information needed to apply the instrument, and several
more (for a full exposition, see Bressers & O’Toole 2005). The game-the-
oretic treatment is, in effect, an analysis at a higher level of generality,
and consideration of particular instruments can be derived from the
more general coverage.
The emphasis here, then, is on game theory as a tool to help identify
alternatives that may be useful for meta-governance, rather than as a
theory employed to predict actions, outputs, or outcomes. The reason is
straightforward. As a branch of applied mathematics, game theory treats
as ‘solutions’ the determination of (for instance) Nash equilibria, even if
such ‘solutions’ constitute failures of cooperation. The game-theoretic
solution to the well-known Prisoners’ Dilemma, to note but one exam-
ple, is mutual defection – a result that represents not only a failure of
interdependent actors to cooperate with each other, but also a self-
interest-driven outcome that does not even represent a Pareto-optimal
situation. In analyzing the practical world of governance, one is neces-
sarily more interested in governance solutions than mathematical ones.
Treating the theoretical apparatus as a heuristic provides considerable
leverage in understanding practical problems of interdependent action.
The reason is that the theoretical structure suggests potential difficulties
as well as leverage points for improving the odds of appropriately coop-
erative action. Changing or at least influencing outcomes is precisely
what public authorities have in mind when they engage in efforts at
meta-governance.
Possibilities for public authorities to shape

network outputs and outcomes
What practical possibilities are available to address the challenge of

meta-governments faced by public authorities? The options are dis-
cussed in terms, and from the perspective, of relevant and legitimate
public authorities.
Meta-governance via policy formulation
If governance involves more than governments, it is nonetheless true that

governments have a major point of leverage that can shape what happens
via networks: public policy. Leaving aside for now the questions of how
policy gets made and the extent to which governments may either be
open to network influence during formal policy adoption, or delegate
policy choices to governance networks, governments typically possess
formal authority to operate at the meta-level across a broad range of
governance challenges. While governance networks must be treated very
seriously for a host of reasons – including political, technical, resource,
and/or managerial advantages that networks bring to addressing the
practical problems of governance – the opportunity to set the policy game
in motion gives public authorities a considerable channel of influence. In
game-theoretic terms, this sort of move sets parameters for one or more
ensuing games yet to be played. The point should not be taken too far, for
network characteristics themselves can shape the policies that govern-
ments adopt (Bressers & O’Toole 1999); but that limitation by no means
signals that governments operate as mere puppets of networks.
Policy formulation is important, but it only begins the game (or adds
a game to an extant mix, or modifies an already-established game).
Additional kinds of moves available to public authorities for meta-
governance can be organized into three categories. Coverage begins
with the least overt type and then moves to others of increasing
complexity and intrusiveness.
Assisting in the play of the game
Public authorities can often initiate moves that do not shift the game-
theoretic structure of the situation. Such moves may seem at first to be
relatively trivial but can be important when the game being played in
the network is not far out of synch with the preferences of democratic
governments. If the main thing needed is to help the interdependent
actors coordinate their efforts – that is, if network actors’ own prefer-
ences are close to the government’s objectives but the actors face diffi-
culties in concerting themselves for the right kind of action – several
forms of governmental assistance can increase the probability of success.
Public authorities can use informational instruments to signal moves
by some network actors for the coordinative benefit of others, in such a
way that a relatively benign scheme of networked operations can be sta-
bilized around the status quo. Assisting in the development and main-
tenance of transparent information systems, similarly, can be a step
taken by public authorities to help facilitate the sensible production of
results. A simple coordination game could nevertheless shift toward
suboptimal production if one or more players perceive the game to be
something different – as would be the case, for instance, if actors have
insufficient information about others’ motives and commitment.

Encouraging iteration in well-functioning or benign network settings
can also be an important contribution by a public authority. None of
these moves seems remarkable, but assisting the play of a sensible game
rather than changing the pattern of interdependence can sometimes be
crucial.
Linking and segmenting games
Activities involved in policy formulation, along with those designed to

assist in the play of the game, can be rather successful varieties of meta-
governance under circumstances in which networked actors do not have
widely divergent preferences. Under different scope conditions, for
instance, when preferences are not so easily meshed, other approaches
can be necessary. When differences in points of view are larger, meta-
governance may require that simpler moves be supplemented by efforts
to link and/or segment the multiple games among the actors.
Although analyses of governance networks are sometimes framed as if
a given network has influence over one, or one type of, output or out-
come, the real world presents a more complex picture – one that can be
amenable to meta-governance by public authorities. Again, the reasons
can be understood by using game-theoretic logic. It has been shown
mathematically that in patterns of multiactor interdependence, con-
necting two or more issues (games) under consideration by essentially
the same set of actors (with the same sets of preferences) can have a
major impact on the choices made. Identifying and explicitly connect-
ing the linked games in a given network, or building connections where
they did not exist originally, can nudge the choices of governance net-
works. Connecting games, therefore, so that network actors understand
that the links are being explicitly considered, can shape collective
action. Similarly, two or more games may be causally connected but
collectively unrecognized. Here a public authority can change network
operations by identifying the connection and highlighting its impor-
tance. Such a move can shift actors’ preferences and/or cognitive
structures.
Governance networks may generate outputs and outcomes in successive
games over time. Public authorities can assist in the achievement of
public objectives by brokering agreements across these games. A
government could offer network actors flexibility on some issues now,
in exchange for concessions on another to be considered later. (Consider
the possibilities presented by covenants with industrial groups on
environmental issues, as developed in the Netherlands.) Public authorities

have particular advantages in serving in such a broker role, particularly
in network settings lacking substantial institutionalization.
Similar points can be made regarding the option of ‘disconnecting’
games among network actors. Sometimes network actors are reluctant to
commit themselves and their resources to the fulfillment of a rather
complex program, given their sense of uncertainty in the venture. If the
reasons have to do with the perceived riskiness of investing a great deal
in collective action, it may be possible for a public authority to disag-
gregate the task into a series of smaller ones, each involving less risk.
Doing so can help in several ways. First, not too much is gambled at any
one time. Second, if participants see that others are likewise investing in
the early parts of the operation, they may perceive less risk. And third,
this dynamic can encourage a sense of trust among the full set of actors.
(Note here the thread connecting the point to governability theory and
not merely interdependence theory. I believe that each of these perspec-
tives has merit.) In short, by operating at the meta-level in disaggregat-
ing the elements of large and potentially costly network operations,
public authorities can increase the odds of success.
Finally, governmental authorities can also link games across different
levels, in the Kiser-Ostrom sense. In such situations, a public authority
can assist in congealing a solution to the second-order collective action
problem well known to game theorists (enforcing and/or monitoring
adherence to commitments that are collectively in the interests of all
participants, even though each can be tempted to shirk while others
contribute) – including by extracting contributions from network mem-
bers to create a monitoring and enforcement mechanism.
The meta-governance moves discussed thus far can be important in
inducing the production of network outputs and outcomes desired by a
government. Not infrequently, however, the game within a network, or
between a governance network and a public authority, is not structured
so benignly. In such situations, rather than merely assisting the play of
the game, or helping to link or decouple network games from each
other, public authorities may need to consider changing the game that
is being played or contemplated.
Changing the game: active meta-governance

by public authorities
When the matrix of preferences in a governance network is closely

aligned with the agenda of the government, the foregoing steps may be
sufficient to help shape network results appropriately. When the only

feasible game-theoretic ‘solutions’ essentially involve practical failures,
from the standpoint of public authorities, more vigorous options can be
considered.5 And while multiple actors may be able to contribute to the
shaping or re-shaping of the game, public authorities are clearly among
these potentially influential meta-governors.
One obvious way of changing the game is by influencing preferences.
Public authorities can use persuasion (perhaps supplemented by incen-
tives) to seek a shift in the ordering of preferred options among network
actors – in the direction of an acceptable or superior overall result. Such
authorities can also seek to operate at a more general level on network
activities, by persuading participants of the value to them of long-term,
stable cooperation rather than a short-term, narrowly self-interested cal-
culus. Such persuasive efforts can focus on constitutional-level contex-
tual features like the understood norms of conduct within network
operations – transparency, promise keeping, good faith, reciprocity,
civility, and trust. Encouraging the development of such norms can shift
the choices of participants and gradually develop greater prospects for
stable, appropriate network solutions. Trust, for instance, converts a
Prisoners’ Dilemma game into a solvable cooperative effort. It also
reduces transaction and monitoring costs, and lowers the rate by which
future payoffs in repeat-game situations are discounted to present value.
The importance of public authorities’ role in promoting the develop-
ment and mutual strengthening of such norms, including by operating
in accord with them from the meta-level as well as using them in assess-
ing and contributing to policy and operations in the network, should
not be underestimated.
Employing persuasion to increase the odds of network agreement and
cooperation amounts to changing the game, but it does so by building
on the assumption that agreement and concertation among the partners
generates acceptable outputs and outcomes, rather than an inappropri-
ate conspiracy. (It is no accident that the English word ‘collaboration’ is
double edged, signaling both cooperative action and also a sinister, sub-
rosa plot.) What if this relatively benign condition does not obtain? In
such situations, public authorities may need to resort to other methods
for changing the game.
The most obvious involves using formal authority to render certain
outputs illegitimate (for instance, illegal). Another straightforward pos-
sibility invokes formal authority to change the value of certain choices
available to networks – either positively (for instance, via subsidy
schemes, or by linking certain outputs to additional attractive payoffs
that government can create) or negatively (through the use of penalties

such as fines, reporting requirements, regulatory oversight, or enhanced
connections to additional policy requirements).
Not to be overlooked is the opportunity often afforded public author-
ities to modify the network’s structure. Depending on the circumstance,
such authorities may be able to mandate links for certain purposes, or
buffer nodes against the need to coordinate with certain others. If the
networks of interest are purely self-organizing, and must remain so, pub-
lic authorities may find their role in altering network structure relatively
limited. But if the networks are subject to at least some mandated
participation or some governmental review of the acceptability of
participation, the influence of public authorities can be substantial.
Designation of a lead actor, or insistence that certain actors should or
should not be regular participants, can shape decisively the results of
network operations. Similarly, devices as simple as accreditation and
licensing can be used by public authorities to craft the composition of
likely networks. As Schattschneider pointed out long ago (1960), deter-
mining who sits at the table to contribute to a joint decision heavily tilts
the decision in one direction or another.
Conclusion
Amid the institutional intertwining characteristic of a networked world,

public authorities are neither in control nor impotent. Nor do they oper-
ate as merely another participant in interdependent action. By virtue of
special status, expectation, and formal authority, governmental agencies
can shape not just the processes but also the results of network produc-
tion. There is no guarantee that such authorities will enhance the public
value of such outputs and outcomes; indeed, they can sometimes
exacerbate inequalities and abet unmonitored and self-interested behavior.
Nevertheless, governments can play key roles in governance, including
in and on governance networks.
Public authorities have the potential to operate at the meta-level in
both senses of that term: by making moves that shape action at policy
and constitutional levels, not only as a direct participant in operations,
and in influencing at one remove the games of interdependent choice
characteristic of networked decision making. Such efforts can be under-
stood as moves to set games in motion via governance networks, assist
in their execution, connect or disconnect strategic interactions across
the range of games in play, and change a game toward a more playable –
or acceptable – collective result. A variety of instruments, and also
instrumental moves, can assist. Although a simple typology does injustice

to the range and subtlety of instruments available, it is clear that infor-
mational tools and suasion, regulatory options, incentive-based possi-
bilities, and combinations of all these can be effective. They can operate
with outputs as targets, or on the range of options available for legitimate
consideration, or on the networked actors themselves – individually or
collectively – with a focus on operations, policy, constitutional-level
aspects, or all three.
While public authorities must adjust to a world of networked governance,
and while such a world inevitably means a diffusion of influence in exchange
for increased capacity, it does not necessarily signal a disappearance of
public authority and governmental influence. Meta-governance in net-
works may be a complex and thus far inadequately understood craft, but
its potential should be clear – and clearly worth developing.
Notes
1. An earlier version of this chapter was presented at the workshop on ‘Democratic
Network Governance: Theoretical Puzzles’, in Roskilde, Denmark, 28–29
April 2005. Thanks for helpful comments are due to the workshop partici-
pants, especially Diana Panke, Allan Dreyer Hansen, Eva Sørensen and Jacob
Torfing.
2. These terms are used here in the fashion typical in the research literature.
Outputs are the direct products of a governance system (decisions, regula-
tions, subsidies, and tangible results like water or air cleaned, roadways paved,
or criminal suspects arrested). Outcomes refer to the ultimate result of such
interventions: increased overall employment, cleaner ambient air, reduced
crime rate. The distinction is important, but for reasons of space the issue is
treated generally in this chapter, so that both types are included simultane-
ously in the analysis. The difference to be kept in mind for present purposes is
this: The generation of outcomes involves a more complex causal chain, and
usually more coproducing actors, than does the production of outputs. In
general, the networks involved in influencing outcomes are larger, the ability
of any actor or set of actors to ‘govern’ the network more constrained, and
thus the metagovernance challenges more significant.
3. Modeling the performance of networks of organizations is considerably more
complex than modeling the performance of an organization operating within
an interdependent setting. We are at work on aspects of the first-mentioned
subject as well, although this work goes beyond the reach of the present paper
(for initial modeling efforts, see Meier and O’Toole 2004).
4. The distinctions developed by Kiser and Ostrom are similar in some respects
to the three ‘orders’ of governing explicated by Kooiman (for instance 2000:
154–61). The operational level corresponds roughly to Kooiman’s first-order
governing, while the legislative or policy level is somewhat similar to his
notion of second-order (institution-building) governing. The meta-governing
referenced in Kooiman’s third-order concept of a normative ‘mortar’ seems a
somewhat looser idea than the rule-based constitutional level sketched by

Kiser and Ostrom. The two approaches are similarly multilevel, but I employ
that by Kiser and Ostrom because its explicitly rule-based institutionalism
is clear and offers rather direct links to the game-theoretic perspective devel-
oped here.
5. Here and elsewhere, some parallels to the analysis provided by Klijn and
Edelenbos (2006) should be evident. In general, their treatment of process
management corresponds roughly to the less intrusive forms of meta-
governance covered earlier in this chapter. Their explication of institutional
design largely fits with this section on changing the game. The current
treatment hews more closely to game-theoretic concepts, and thus to the
frequently multi-actor aspect of meta-governance. This chapter provides
heuristic guidance in identifying the variables and points of leverage, and
the chapter by Klijn and Edelenbos probes more deeply the specifics of the
managerial moves that may be apropos.
Part IV
Democratic Network Governance
13
Democratic Network
Governance
Introduction
Governance network theorists have mainly been interested in the role of

governance networks in enhancing governance efficiency while less
attention has been directed towards the impact of governance networks
on democracy. Their reflections about the democratic problems and
potentials of governance networks have been relatively brief and sketchy.
However, the content of these reflections are highly intriguing as they
suggest that while governance networks certainly represent a threat to
representative democracy they do not necessarily threat democracy as
such. They could become a central building block in the development of
new forms of democracy (Goss 2001; Jessop 2000b; Kickert et al. 1997:
174; Pierre & Peters 2000: 67; Rhodes 1997a: 21; Skelcher 2004; Wolf
2002). Bob Jessop (2000b: 17) claims that governance networks might
prove valuable as a means of linking different units of democratic gover-
nance in the global, polycentric political systems of our time. Rod
Rhodes (1997a: 9; 1997b: xiv) suggests that governance networks repre-
sent an important functionally organised supplement to the territorially
organised institutions of representative democracy. Jan Kooiman (1993:
36; 2000: 143) regards governance networks as a promising way of
enhancing the flexibility of democratic institutions in order to increase
the adaptability of democracy to the specific characteristics of concrete
governance processes. Fritz Scharpf (1999: 18ff) points to governance
networks as an important means to increase democratic outcome legiti-
macy because they increase knowledge exchange and shared under-
standing between policy makers and stakeholders.
As such governance network theorists have opened the door to a com-
plex and radical discussion about the future of democracy and the role
233
that governance networks might play in it. However, as several gover-

nance network theorists contend, governance research still needs to take
the full step across the doorstep into the realm of democratic thought
and address the question of the relationship between governance net-
works and democracy head on (Rhodes 1997b; Scharpf 1999; Pierre
2000; Peters & Pierre 2000; Mayntz 2003; Skelcher 2004).
This part of the book focuses on the democratic problems and poten-
tials of governance networks. In order to reflect on this issue we need to
consider more closely what is meant by democracy. Currently, we wit-
ness a marked shift in the perception of what democracy is and how it
should be institutionalized. As we shall see, this shift can be described as
a move from a liberal to a postliberal perception of democracy.
Evidently, the assessment of the democratic implications of gover-
nance networks depends very much on how democracy is defined.
Accordingly, it is necessary to take a closer look at the competing notions
of democracy. First, we give a brief outline of the traditional theories of
liberal democracy and their interpretations of governance networks.
Then follows a description of how four emerging postliberal theories of
democracy relate to governance networks. Finally, we analyse the differ-
ences and similarities between the old and new theories of democracy
and pose three central questions regarding the democratic problems and
potentials of governance networks that need to be addressed.
Governance networks and liberal democracy
The large and heterogeneous body of literature that theorists on democ-

racy denote liberal theories of democracy (Macpherson 1977; Holden
1993) share the view that democracy can be defined as a political system
in which the members of a territorially defined political community –
i.e. a nation state – govern themselves either through direct citizen
participation or through the election of representatives. One way of
obtaining some degree of structure in this complex body of liberal dem-
ocratic thought is to group the theories according to different criteria.
Many such efforts have been made in order to produce clarity, but, as
always, with some degree of oversimplification as its price. Knowing
that the price of clarity is a loss of detail and complexity, we adopt James
G. March and Johan P. Olsen’s (1989: chapter 7) division of liberal theo-
ries of democracy into two groups: the predominantly aggregative and
the predominantly integrative theories.
The common feature of the aggregative theories is that they regard
democracy as a way of dividing political power equally among the
Theoretical Approaches to Democratic Network Governance 235
citizens by means of general elections, and a way of ensuring the maxi-

mum level of individual liberty by means of various forms of minority
protection and a large private sphere. The traditional institutions
of representative democracy are expected to produce this political equal-
ity and individual liberty. By contrast the integrative theories regard
democracy as a means to supplant political battles for power with
reasoned deliberation directed towards identification of the common
good. Accordingly, the main purpose of democratic institutions is to
develop a sense of communality and obligation to act in the interest
of the common good among the citizens. The means to do so is
intensive citizen participation in the many voluntary organizations,
institutions and groups that constitute civil society. Hence, a proper
functioning of the traditional institutions of representative democracy
calls for a strong civil society that is capable of developing the citizens
into democrats that make decisions in the interest of the common
good.
Although the aggregative and the integrative theories of liberal
democracy differ in several respects they share the view that democracy
calls for a sharp borderline between state and society: aggregative theo-
ries of democracy regard it as a precondition for ensuring an equal dis-
tribution of political power in the state as well as for ensuring individual
liberty in the private sphere; integrative theories regard it as a means to
ensure an autonomous realm of reasoned deliberation that can foster
the development of democratic citizens.
Seen from the perspective of the theories of liberal democracy the
question about the democratic problems and potentials of governance
networks is quickly answered: Governance networks represent a threat
to democracy because they undermine the borderline between state and
society. Aggregative theories of liberal democracy see governance net-
works as a threat to political equality because they undermine the sov-
ereign position of the elected government and a threat to individual
liberty because collective decision-making is spreading into the private
sphere. Integrative theories contend that governance networks con-
tribute to expanding the scope of public participation in processes of
political decision-making, but stress that networks are more oriented
towards the promotion of particularistic interests than on the promo-
tion of the common good of society as such. In so doing, they threaten
the development of democratic citizens instead of promoting it. In sum,
both aggregative and integrative theories of liberal democracy see gov-
ernance networks as a danger to democracy. Governance networks
might be effective but they are certainly not democratic.
Governance networks and postliberal democracy
While a traditional liberal approach to democracy leads to an overly

negative evaluation of the democratic problems and potentials of gover-
nance networks, a more complex and optimistic view is reached when
taking a postliberal approach to democracy (Sørensen & Torfing 2005c:
218). In the last two decades, new theories of democracy have surfaced
that in different ways surpass the traditional liberal theories of democ-
racy. They (1) seek to reformulate the basic ontological positions of the
aggregative and integrative theories of liberal democracy; (2) renounce
the traditional institutions of representative democracy, including its
call for a sharp demarcation between state and society; and (3) challenge
the idea that democracy is confined to decision making within a given
polity. Hence, post-liberal theories of democracy search for new percep-
tions of democracy, and new institutions of democracy that do not take
the existence of a well-defined homogenous nation-state and a sharp
demarcation between state and society as their starting point. Post-
liberal theories of democracy share the ambition of developing such
new perceptions and institutions of democracy, but they do so from very
different theoretical starting points. As was the case with governance
network theory, it is possible to divide the emerging postliberal theories
of democracy into four groups according to whether they take departure
from a conflict or a coordination approach to governance, and whether
they hold a calculation or a culture view of human action.
Each of these theoretical approaches to democracy indicates ways in
which governance networks could contribute positively to the develop-
ment of new form of democracy. Below we shall see how.
Competitive democracy
The competitive theories of democracy share the view that democracy is
basically a means of regulating battles for power among self-interested
Table 13.1 Four theoretical approaches to postliberal democracy
Calculation Culture
Conflict Competitive democracy Agonistic democracy

[Etzioni-Halevy 1993; Hirst [Connolly 1991, 1995; Mouffe
2000] 1993; 2000]
Coordination Outcome democracy Community democracy
[Fung & Wright 2003; [March & Olsen 1995; Sandel
Young 2000] 1996]
actors with the aggregative theories of liberal democracy, but they reject
the idea that it is possible to contain political battles within the realm
of the state and the institutions of representative democracy. Among
these theorists we find Eva Etzioni-Halevy (1993) who seeks to reformu-
late traditional democratic elite theory and its call for competition as a
means of balancing elite power in a society where semi-public sub-elites
play a significant role. Etzioni-Halevy (1993: 53) insists that the core
feature of representative democracy is not primarily that it allows the
people to appoint and control political elites through elections. All
the way back to Charles Montesquieu, representative democracy has
been celebrated for its ability to institutionalize competition among
autonomous elites, and thereby establish a situation in which various
elites within the political system keep each other at bay. Etzioni-Halevy
(1993: 53–4) takes a further step in claiming that semi-public sub-elites,
such as social movements and other actors placed in the grey zone
between the public and the private sphere, can in fact enhance democ-
racy. Hence, they can supplement the horizontal balancing of powers
between different political elites and parts of the state apparatus with a
vertical balancing of powers between the established elites and sub-
elites. While ordinary people do not have the necessary knowledge and
capacity to keep track of the decisions and actions made by the ruling
elites, sub-elites have the capacity to ensure public control over the rul-
ing elites between elections. According to Etzioni-Halevy, sub-elites,
who have been more or less ignored by traditional elite theory, are not
only central to democracy because they increase the pressure on the
ruling elites. Sub-elites are also important because they establish an
intermediary level between the people and the ruling elites that facili-
tates mobility between the people and the political elites (Etzioni-Halevy
1993: 194). This level gives citizens an opportunity to play an active role
in policy processes without themselves becoming political elites, and it
serves as a training ground for the citizens in which they learn to play by
the rules that regulate processes of political competition.
Paul Hirst (2000) is another important contributor to a postliberal
competitive theory of democracy. His associative model of democracy
suggests that representative democracy at the national level ought to be
supplemented with publicly founded, but relatively self-regulating
voluntary associations at the local level that provide a host of public
services to the population. The establishment of such self-regulating
voluntary associations will increase the vertical balance of power
between democracy from above (representative democracy) and democ-
racy from below (self-regulating voluntary associations). The role of the
state will be to define the overall political goals and financial frames
within which the voluntary associations operate, while the task of the
associations will be to produce public services in competition with one
another.
Hirst claims that in the world of today, the principle of affectedness
must play a vital role in considerations concerning how to ensure equal
access to political channels of democratic influence. The equal right to
vote in national elections will remain important, but the territorially
defined representative democracy must be supplemented with a func-
tionally defined democracy for the affected stakeholders. In such a
model of ‘associative democracy’ (Hirst, 1994), the access to channels of
influence should be distributed equally, not among all citizens, but
among those who are affected by the decisions taken by the local
associations. (Hirst 2000: 29).
Hirst operates with a number of mediating consociational institutions
linking the self-governing associations with each other and with the
state. These consociations play a central role in balancing and connecting
the local and national levels in the associative model of democracy
through a process of negotiated governance (Hirst 2000: 30). The conso-
ciation has many of the same characteristics as governance networks, as
they link interdependent, but relatively autonomous, actors in an effort
to reach solutions to shared problems through horizontal negotiation
and mutual coordination.
As such, the two contributors to a postliberal theory of competitive
democracy mentioned above agree on the need for a vertical balancing
of powers between political elites and strong local actors, and an ongoing
competition, contestation, and coordination between them. Furthermore,
they point to the importance of providing all affected citizens access to
channels of influence that supplement those made available by the
institutions of representative democracy.
Seen from the perspective of the competitive theory of democracy,
governance networks might under certain conditions contribute to the
strengthening of democracy. First, they could be seen as a way of recruit-
ing, nurturing and organizing political sub-elites. Second, governance
networks could function as a stepping-stone that enhances the mobility
between political elites and ordinary citizens. Third, they could prove to
be a fruitful way of organizing coordination processes among a plurality
of autonomous actors and between top-down state rule and bottom-up
self-regulation.
However, the competitive perspective on democracy also points to a
number of potential dangers for democracy when involving networks in
governance processes. Etzioni-Halevy (1993) fears that governance net-

works between elites and subelites might undermine elite autonomy
and elite competition. In addition, she stresses that the informal char-
acter of many governance networks is likely to result in a low level of
accountability and publicity in the decision-making process. Hirst
claims that there is a risk that governance networks might encroach on
tasks that are to be dealt with exclusively by the state. These tasks
include: (1) the distribution of powers and responsibilities between the
national, regional and local governments and civil society; (2) the pro-
vision of the main source of internal democratic legitimacy for the citi-
zens; and (3) the establishment of a primary and democratically
legitimate actor in the dealings with other nation states and political
entities (Hirst 2000: 31). Hence, competitive democracy does not see
governance networks as the saviour of democracy: they pose a range of
democratic pitfalls, but contrary to liberal theories of democracy, they
are also seen as having certain democratic potentials.
Outcome democracy
A number of postliberal theorists on democracy share the calculus
approach to human action with the aggregative theories of democracy,
but tend to measure the quality of democratic governance more on its
ability to produce desired outcomes through various forms of coordina-
tion than on how decision makers are democratically legitimized.
Archon Fung and Erik Olin Wright (2003) have developed a model of
democracy they refer to as ‘Empowered Participatory Governance’ (EPG),
which in some respects is inspired by Jürgen Habermas’ deliberative
model of democracy. However, while the EPG-model regards deliberation
as a core ingredient in a well-functioning democracy, Fung and Wright
take a far more pragmatic and praxis-oriented approach to democratic
governance than that found in Habermas’ ideal-typical model of deliber-
ative democracy. The guiding principle in the EPG-model is that demo-
cratic institutions ought to be judged by their ability to solve policy
problems experienced by the people ‘more effectively than alternative
institutional arrangements’ (Fung & Wright 2003: 25). Fung and Wright
argue that three guidelines are to be followed when designing democratic
institutions capable of effectively solving defined problems (Fung &
Wright 2003: 16ff). First, democratic governance institutions must be
geared to dealing with practical concerns and concrete situations. The
suggestion here is that we ought to bid farewell with specific characteris-
tics to the generic image of democratic institutions as permanent enti-
ties. Institutions should be designed for concrete situations. Second, the
capacity for effective problem solving is increased if the political process

is organized in a manner ensuring the bottom-up participation of the rel-
evant stakeholders. The stakeholders possess valuable insights and a con-
crete engagement that enhances efficient problem solving and the
responsible and engaged implementation of decisions. Third, democratic
governance must take place through deliberative problem solving where
‘the participants find reasons that they can accept in collective action’
(Fung & Wright 2003: 17). Deliberation is enhanced in situations in
which power is shared among the participants since power sharing con-
tributes to the mutual respect between the participants (Fung & Wright
2003: 23). Conflicts that hamper effective problem solving can be
avoided through this form of balanced, or negotiated, deliberation only.
Iris Marion Young (2000) is also intensely concerned with the quality
of the outcome of democratic governance processes. She is heavily
inspired by John Rawls’ claim that democracy ought to be measured in
terms of its ability to provide just solutions (Young 2000: 27). Like
Rawls, she argues that since there is no common good according to
which it is possible to measure whether an outcome is just or not, justice
must be ensured procedurally. The outcome of a process of democratic
decision-making can be deemed just if:
all significantly affected by the problems and their solutions are

included in the discussion and decision-making on the basis of equality
and non-domination, and if they interact reasonably and constitute
a public where people are accountable to one another (Young 2000:
29–30).
Hence, the central goal in Young’s efforts to develop democratic institu-

tions is to establish procedures capable of producing just outcomes that
is outcomes, which are to an equal extend in the interest, not of all citi-
zens but of the affected by the decision being made.
The outcome democracy defined by Fung and Wright and Young
point to several ways in which governance networks could enhance
democracy. Governance networks tend to follow all of the guidelines
advanced by Fung and Wright: they are usually constructed around the
solution of specific problems or tasks; they tend to facilitate bottom-up
participation; and they provide an arena for negotiated deliberation
between autonomous and mutually dependent actors. As such, gover-
nance networks contribute to effective policy making and policy out-
comes. In so far as governance networks respect the procedural demands
advocated by Marion Young, they will also help to enhance just

outcomes.
However, outcome democracy also points to a number of potential
dangers for democracy connected to the use of governance networks.
Young (2000: 52) is worried that situations occur where specific groups of
stakeholders are systematically excluded from the governance networks,
and that the level of publicity will become too low, while Fung and
Wright (2003: 33) point to the danger that a few strong actors will dom-
inate the networks. In addition, Fung and Wright (2003: 21) claim that
the establishment of a plurality of self-governing networks might lead to
an atomization and fragmentation of the political processes that will
make it difficult to ensure a democratic governance of the overall polit-
ical development of society. However, according to Fung and Wright,
the potential dangers for democracy can be avoided through a system of
centralized supervision and coordination:
Unlike New Left political models in which concerns for liberation

lead to demands for autonomous decentralization, empowered par-
ticipatory governance suggests new forms of coordinated decentral-
ization. Driven by the pragmatic imperative to find solutions that
work, these new models reject both democratic centralism and strict
decentralization as unworkable (Fung & Wright 2003: 21).
This call for centralized supervision is not unlike the demand for
meta-governance expressed by many governance network theorists.
Community democracy
In line with the integrative theories of liberal democracy, the postliberal
theories of community democracy maintain that a democratic polity
cannot be reduced to a legally defined unity. In a democratic society cit-
izens are linked together by open dialogue and public debate taking
departure from a shared sense of connectedness and collective identifi-
cation. However, the postliberal theories of community democracy
reject the idea of a unified democratic community so outspoken in the
integrative theories of liberal democracy. The ongoing displacement of
political power upwards to inter- and supranational institutions and
downwards to local political institutions, voluntary organizations, and
private firms has lead to a situation in which the nation state can no
longer play the role as the unifying point of identification that defines
the group of individuals that belong to a given polis. Moreover, no other
overarching point of collective identification is ready to assume its

place. As Michael Sandel states:
Since the days of Aristotle’s polis, the republican tradition has viewed
self-government as an activity rooted in a particular place, carried out
by citizens loyal to that place and the way of life it embodies. Self-
government today, however, requires a politics that plays itself out in
a multiplicity of settings, from neighbourhoods to nations to the
world as a whole. Such a politics requires citizens who can think and
act as multi-situated selves. The civic virtue of our time is the capacity
to negotiate our way among the sometimes overlapping, sometimes
conflicting obligations that claim us, and to live with the tension to
which multiple loyalties give rise (Sandel 1996: 350).
In a multi-sited governance process of multi-situated selves, we require

democratic institutions and norms capable of helping people to navi-
gate within a patchwork of collective identifications and orientations.
Postliberal theories of community democracy also relinquish the
notion that reasoned debate leads to the identification of an all-embracing
common good. Instead, they argue that reasoned debate leads to the
construction of shared stories of past, present and future that render
meaningful collective behaviour possible (March & Olsen 1995: 63ff;
Sandel 1996: 350). Democracy is promoted through the telling of stories
that construct democratic rules, norms and logics of appropriateness in
and between political communities. Community democracy theorists
have also shown an interest in identity formation as a means of enhanc-
ing the level of democratic political empowerment among the citizens.
Democracy involves the shaping of political capacities and political
identities that support a democratic ethos of reciprocity and a high level
of political engagement (March & Olsen 1995: 91ff; Sandel 1996: 333).
In sum, community democracy renounces the notion that a demo-
cratic political process presupposes a unified and homogenous political
community; that political engagement is directed towards the promo-
tion of a pre-political common good; and that the development of
empowered citizens with a strong sense of communality must necessarily
proceed within an autonomous civil society. They perceive the nation
state as one out of many competing and overlapping points of political
identification, and they regard the establishment of linkages and bridges
between political identities, narratives and communities as an important
democratic task that postliberal theories of democracy must take into
account.
Governance networks might prove to be an important means of estab-

lishing such linkages and bridges, as they bring together stakeholders with
different points of identification in an autonomous environment, which
enhances communication, coordination, negotiation and cooperation
between them. Furthermore, governance networks increase public partici-
pation and engagement and thereby contribute to the empowerment of
citizens.
However, governance networks also pose a threat to democracy. First,
overlaps and ties between political communities might turn out to be
systematically structured in ways that tend to marginalize certain com-
munities and identities. Second, governance networks might empower
and engage the few while disempowering and disengaging the many.
Third, there is a danger that the loosely coupled nature of governance
networks will prevent the development of a point of identification and
orientation in the governance networks that makes them capable of
working towards a shared objective.
Agonistic democracy
Finally, one can identify the emergence of a group of postliberal theories
of agonistic democracy. These theories combine a conflict perspective of
democracy with a cultural understanding of the nature of human
action. Like the aggregative theories of liberal democracy they regard
democracy as a way of regulating political conflicts while rejecting the
idea that actors are basically driven by rational calculations of costs and
benefits of alternative forms of action. Instead, they combine the con-
flict approach to democracy with an integrative understanding of sub-
jectivity as an outcome of social processes of identity formation that are
shaped by contingent hegemonic articulations.
Agonistic theories of democracy claim that traditional theories of
liberal democracy have overly focussed on the regulation of the three
faces of power: direct power through decision-making, indirect power
through non-decision-making, and ideological power through the
shaping of other actors’ perceptions. But they have ignored the ques-
tion of how discursive power that constructs the actors’ identity, world
view and scope for legitimate action can be democratically regulated.
The key questions raised by these theorists concern how individuals
discursively construct themselves and others as democratic actors; how
the best conditions for discursive contestation are created, and how
discursive images of the polity and the political issues at stake produce
patterns of inclusion and exclusion (Mouffe 1993, 2000b; Connolly
1995).
William Connolly (1991) and Chantal Mouffe (1993) argue that the
core objective of democracy is to facilitate an ongoing discursive con-
testation. Politics consists of battles between competing discursive
images of society, its borders, and the identity of those inhabiting it.
One of the key objectives for theories of democracy is to consider how
these battles are to be democratically regulated. Connolly (1991) and
later Mouffe (1993) argue that the principal democratic task consists in
transforming antagonistic friend-enemy relations into agonistic rela-
tions, whereby people disagree on substantial and procedural issues but
respect one another’s right to voice dissimilar opinions. In Mouffe’s
words: ‘the aim of democratic politics is to construct the “them” in such
a way that it is no longer perceived as an enemy to be destroyed, but as
an “adversary”, that is, somebody whose ideas we combat but whose
right to defend those ideas we do not put into question’ (Mouffe 2000b:
101–2). Therefore, democracy’s worst enemy is essentialist beliefs that
stipulate the existence of a pre-political common good. Such beliefs
might serve to legitimize efforts to remove democracy and establish a
totalitarian regime in situations where the outcome of a democratically
regulated political process does not realize that which a minority defines
as the common good. The best protection against totalitarianism is
the recognition of the political, and therefore contingent, character of
the common good.
In line with this argument for the primacy of politics, agonistic theo-
ries of democracy emphasize the political character and effects of the
constitution and demarcation of the polity. The tendency in traditional
theories of liberal democracy to regard the polity as a pre-political entity
means that the political patterns of inclusion and exclusion it estab-
lishes tend to become invisible, and thus escape political contestation
and democratic regulation. This is problematic because the scope of the
polity has severe political implications not least through the patterns of
inclusion and exclusion it installs (Connolly 1995: 135ff).
The theories of agonistic democracy open up for different ways in which
governance networks can contribute positively to democracy. First, gover-
nance networks might widen the scope of discursive contestation because
they involve a plurality of public and private actors in the governance
process. New actors obtain a platform through which they can raise their
voice and claim to be heard. Second, network governance promotes a poli-
tization of the construction of the polity because the construction of gov-
ernance networks takes place as an integrated part of the policy process.
Hence, patterns of inclusion and exclusion and the constitutional rules
that regulate networks are settled not in advance but as an integrated part
of the policy processes within the networks. Finally, the fact that gover-
nance networks are held together by relations of interdependency enforces
a negotiation process that may help to transform antagonistic relations
into agonistic ones. Negotiation processes force hostile actors to commu-
nicate and thus to get a better and more nuanced image of each other.
However, agonistic democracy also points to ways in which gover-
nance networks might undermine democracy. The overall worry
concerns the danger that governance networks integrate the network
actors to an extent that does not only lead to a transformation of antag-
onism into agonism, but transforms agonistic disputes into rational
technocratic problem-solving discourses that evaporate any chances of
democratic contestation from competing discourses.
Similarities and differences between the theories
Each of the postliberal theories of democracy points to ways in which

governance networks might contribute to enhancing democracy in a
society that becomes increasingly multi-levelled and multi-centred.
Governance networks provide a means to increase the flexibility of our
democratic institutions and to increase the number of citizens who
participate actively in processes of political decision-making.
Postliberal theories furthermore agree that governance networks can-
not stand alone. They should rather be seen as a supplement to the tra-
ditional institutions of representative democracy and thus as making
way for the development of a political system that combines territorially
and functionally organised democratic institutions. One of the specific
tasks of the institutions of representative democracy is, as argued by
Hirst and by Fung and Wright, to regulate governance networks and the
relationship between them in order to enhance their democratic qualities
and avoid the democratic dangers that they contain. Hence, the demo-
cratic potentials of governance networks are not harvested automatically,
but must be nurtured through a deliberate effort to regulate them through
various forms of metagovernance.
While the postliberal theories of democracy agree that governance
networks bring with them a number of potential problems for democ-
racy, they tend to focus on different problems. Young and Sandel argue
that governance networks might result in a systematic exclusion of
specific groups of stakeholders and particular political identities while
including others. This might result in a situation where a number of cit-
izens are excluded from participating in networks and thus from
exercising influence on decisions that affect them.
Other postliberal theorists are more worried about the democratic

quality of the decision-making processes within networks once they are
formed. Hence, Fung and Wright argue that there is a risk that strong
network actors dominate the network processes. March and Olsen on
their side question the extent to which network actors reach a level of
integration that makes them able to develop a common point of identi-
fication and orientation. Etzioni-Halevy, Mouffe and Connolly have the
opposite concern, namely that the network actors establish so strong a
hegemony that little space is left for discursive contestation and openness.
Finally, Etzioni-Halevy and Young are worried that the informality of
governance networks might make it difficult to ensure the level of
publicity regarding the governance networks, which is necessary in
order to ensure their democratic accountability.
Where to go from here?
It should now be clear that taking departure form a liberal approach to

democracy leads to the conclusion that democracy is threatened by gov-
ernance networks because they undermine representative democracy.
However, seen from a postliberal approach to democracy one reaches a
much more complex conclusion. Governance networks contribute in var-
ious ways to enhancing democracy but they also raise serious problems
that must be dealt with.
With reference to the postliberal theories of democracy, at least three
questions must be answered if governance networks are to become a
positive contribution to democracy: 1. how can it be ensured that all
affected are allowed the access to participate in governance networks?;
2. how can it be ensured that network actors participate in the networks
on equal terms, and develop a collective orientation, and that the net-
works allow for discursive contestation and openness between network
actors and networks; and 3. how can publicity be promoted in and
around governance networks in order to ensure democratic accountabil-
ity. These problems will be addressed in the next chapters.
14
Governance Networks and
Participation
Allan Dreyer Hansen
In this chapter we shall take a closer look at governance networks from

the perspective of participatory theories of democracy. The aim is
twofold: First, to contribute to the overall debate regarding the demo-
cratic legitimacy of governance networks. What are their potentials and
shortcomings when viewed in light of participation? Second, the aim is to
specify and reformulate the participatory tradition in ways that help spec-
ify concrete analyses of the democratic quality of governance networks.
The second aim occupies most of the space in the article on the
grounds that it is very difficult to assess the general democratic standing
of governance networks. It is not possible to deduce the democratic
quality of governance networks from the concept itself, neither with
negative or positive conclusions. Even though certain (probably) wide-
spread features of governance networks – such as the limited access and
absence of publicity around them – are definitely not democratic, such
features are not inherent to the notion of network, as such (which some
positive experiences also suggest).1 Moreover, even if they are not com-
pletely outweighed, then they are at least partially countered by other
aspects contributing to the advance of democracy. Any serious
encounter must be more specific, posing questions of a more detailed
character than simply whether network governance is democratic or
not. Such questions should focus on the extent to which governance net-
works live up to a different, more precisely specified participatory norm.
The participatory tradition includes internationally established
authors such as Carole Pateman (1970), C.B. Macpherson (1977),
Benjamin Barber (1984), as well as the somewhat less renowned Danish
theorist Hal Koch (1981 (1946)). The term ‘participatory’ was first
coined by Arnold Kaufmann in 1970 (Cunningham 2002: 123), but as
the inclusion of Koch indicates, this way of thinking has roots from well
247
before the 1970s and stretching back to the civic republican tradition
headed by Jean Jacques Rousseau (Cunningham 2002: 123f).
The participatory tradition can be understood to consist of three dif-
ferent but interrelated claims, namely that democracy is something to
be learned; that it involves equality among citizens; and finally, that the
results of democratic processes promoting widespread civil participation
lead to the realization of some kind of a common good. As Pateman
states, ‘In the participatory theory “participation” refers to (equal)
participation in the making of decisions, and “political equality” refers
to equality of power in determining the outcome of decisions … [T]he
justification for a democratic system … rests primarily on the human
results that accrue from the participatory process’ (Pateman 1970: 43).
Participatory theory claims that participation can reduce problems
such as apathy among the citizenry and their general orientation
towards particular interests as often claimed by so-called realists
(Schumpeter 1976 (1943)). As Barber writes, ‘people are apathetic,
because they are powerless, not powerless because they are apathetic’
(Barber 1984: 272). This leads to a claim that democratic participation
ought to be broadened to as many social spheres as possible; it should
not be restricted to the state or the parliament.
This positive valorisation of participation speaks in favour of gover-
nance networks, because they blur the sharp distinction between state
and civil society by bringing together representatives from both public
and ‘private’ organizations in efforts to govern society. However, the
participatory tradition has linked the norms of equality and democratic
learning to the notion of the promotion of a ‘common good’ (in the sin-
gular) defined through consensus as the overall purpose of democracy
(I substantiate this claim in the next section). This demand could easily
make most governance networks problematic due to their narrow scopes
and aims.
The aim of this chapter is to argue that the notion of a ‘substantial’
common good inherent to democracy can not be maintained.2 This does
not invalidate all claims made by the participatory tradition, as such;
rather, it necessitates their reformulation. Moreover, a problematization
of an ideal of a singular common good allows for consideration of the
democratic potentials and problems of governance networks, opening a
potential space for positive effects in spite of their narrow perspective.
Since the meaning of the common good has direct impact on the
articulation of the other norms, the article starts here. The next section
presents the norm or principle of learning, which I argue must be parted
into two subfields, namely learning the skills of democracy and acquiring
Governance Networks and Participation 249
a democratic ethos of ‘listening to the other’. I then discuss the possibil-

ity of reformulating the principle of equality. This discussion includes
two considerations: that of equality in terms of access (by the affected)
to networks and that of equality among the participants ‘within’ a net-
work. Each section concludes with a discussion of how these norms may
be utilized in concrete analysis, as well as of how governance networks
can generally be assumed to meet the different democratic norms. The
article concludes with a number of general considerations regarding
governance networks when viewed from the perspective of participation.
The common good
The idea of a common good plays a central role in the tradition of

participatory democracy. As Pateman states, through ‘the participatory
situation … each independent but interdependent individual is
“forced” to appreciate that there is only one right answer, to apply the
word “each” to himself’ (Pateman 1970: 24). Hal Koch (Koch 1981
(1946)),3 makes the same type of claims when arguing that ‘the essence
of democracy is not the vote, but the dialogue, the negotiation, the
mutual respect and the sense of the general interest, that grows from it’
(23) and further, that the democratic task is to ‘find the right solution to
the conflicts’ (27), where both the general interest and the right solution
is in the singular. Even Benjamin Barber, who explicitly distances him-
self from a ‘unitary democracy’ (being kept together by pre-political ties)
and its ‘substantive consensus’, maintains that ‘a strong democracy’
must be able to transform conflicts into ‘creative consensus’ (Barber
1984: 9).4
The question is how literally such statements are to be taken. Most
participatory theorists would likely introduce a set of qualifications stat-
ing that the common good cannot be reached, as such, but that partici-
pation would help citizens to move in that direction without ever
actually reaching it.5 However, my contention is that consensus has rep-
resented an ideal for the participatory tradition and that the ideal of
consensus can be perceived as a democratic ideal of identifying one
common good present underneath actual democratic processes and func-
tioning as a (counterfactual but still present) aim that the participatory
processes should strive for and be evaluated against. The problem of the
idea of a common good in the singular is that it gives the impression
that the extent to which a process can be viewed as democratically
healthy depends upon its ability to produce one (and only one) result.
When Pateman argues with Rousseau that there is only one right answer
(resulting from the ability of seeing one self as ‘each’), this positions the
researcher as one who can judge the pertinence of the process by exam-
ining its outcome. One would have to decide on what the right answer
in a particular situation is, and then judge the distance from the actual
results.
My claim is that a reformulated participatory theory of democracy
should instead be grounded in the notion that democracy involves
power, politics and conflict, and that the formulation of a universal
common good in the traditional meaning of the phrase is ultimately
unsustainable (see Mouffe 1992). Unanimity and consensus is not only
unlikely, but actually democratically undesirable (Laclau 1996a; Laclau
2001). Stated succinctly, the reason for this is a radical pluralism charac-
terizing the modern world, that is, a pluralism that does not render the
disappearance of differences in identities and interests a likely scenario.
A non-exclusionary consensus of the common good is simply not per-
ceivable. Rather than hiding exclusions behind a claim of consensus of
the common good, democracy consists of keeping the exclusionary
moments open, making them possible arenas of renewed democratic
struggle (Mouffe 2000b).
But does this mean that the very notion of a common good must be
abandoned? I do not think so, nonetheless, the common good must
be reformulated as regards the procedures and ethos supporting demo-
cratic politics. The common good is not a matter of the existence of one
just result, but rather democracy itself: allowing all parties and interests
to have a say in the democratic struggles and accepting all democratic
positions as legitimate opponents or ‘adversaries’ in ongoing ‘agonistic’
struggles and exchanges (Connolly 1995; Mouffe 2000a, b; Mouffe 2000).
There can be no democracy without some notion of a common good;
however, the point is that the common good remains empty and con-
tested (Mouffe 2000b). Participants might well agree that they are aiming
for just results (as Iris, e.g. Marion Young suggests [Young 2000: 50]). The
problem is that they are most likely to disagree on what that means
should be in a given situation. Decisions and (temporary) exclusions
(of points of view) are therefore here to stay – also within processes that
meet democratic standards.
As regards governance networks, this reformulation of the notion of a
common good raises two questions: first, one must consider the extent to
which a governance network functions according to a democratic ethos,
that is, accepting democratic rules of the game, including a willingness to
listen to the other, equality amongst members, and so forth. I return to
this question later in the discussion of learning. Second, one must
consider the fact that even though the notion of a substantial common
good (that is, actual unanimous agreement on decisions) is abandoned as
a desirable outcome of any political process, one ought to continue to
engage in more substantial questions regarding the way the participants
argue their case. Hence, a democratic ethos demands that the network
participants are willing to argue their case with reference to a broader
perspective than that of their own particular interests. When evaluating
the democratic quality of concrete governance networks, one must then
look for forms of argumentation that establish links between particular
points of view and the wider interests of the society in which the net-
work is a part. Such interests could include that of justice and defending
the weak, or at least the concrete aims of the network as something that
goes beyond the spontaneous interests of the different members. These
higher principles (as well as the aims of the network) are ultimately
‘empty signifiers’ and the links established with them by the network
members are of a hegemonic kind (Laclau 1996b). That which makes
such battles democratic is exactly the acceptance by the participants that
any such link is temporary in nature and can legitimately be contested.
If we accept such a reformulation of the common good, how do
governance networks then fare? In the first place, such a move appears
to be a prerequisite of the very possibility of judging governance net-
works democratic at all. Since most governance networks are formed in
order to pursue narrow objectives, they will rarely serve a common good
in any strong sense. It is only to the extent that we accept that democ-
racy cannot be about the realization of any substantial common good
that governance networks can be thought to make a positive contribu-
tion to democracy. This does not imply that any governance network is
functioning in ways that are democratically legitimate. Rather, it means
that what is required are concrete analyses of how individual networks
function in practise in order to measure their democratic quality. What
must be analyzed is the extent to which the network takes broader socie-
tal aims into consideration or reflects on its own (possible) ‘externalities’
and the like (which many networks probably do not). If such forms of
argumentation and consideration can be traced, it indicates the working
of a norm of a (de-universalised) common good and ought to be read as
something positive in terms of democracy.
Learning democracy
As mentioned above, the participatory tradition regards learning as a

central contribution of democratic participation. In Pateman’s words,
‘the major function of participation … is therefore an educative one,

educative in the very widest sense, including both the psychological
aspect and the gaining of practise in democratic skills and procedures’
(Pateman 1970: 42).6 The argument is based on the persistent (and con-
vincing) claim that democracy is not something naturally given, but
represents a set of practises and identities that must be acquired, learned
and re-learned. I take this argument to point to a central aspect of
democracy, but in need of considerable reformulation in the light of the
above-presented claim for a de-universalization of the notion of a
common good.
Acquiring democratic skills through participation is a classic idea in
participatory democracy (Macpherson 1977). In order to function in a
decision-making context, one requires an array of skills: the ability to
form an opinion; capacity to determine and defend one’s interests; abil-
ity to state an opinion; experience with participation in meetings; the
ability to make public statements and take issues to the vote; and so
forth. All of these skills remain pertinent, and governance networks
should also be assessed according to the degree to which they promote
such learning and ‘democratic empowerment’. Since governance net-
works often provide opportunity for participation to actors who are not
trained politicians, this is the point where they have one of their
strongest democratic potentials.
Changing identities and listening to the other

However, acquiring the practical skills is only one part of the story. A
learning process also affects the participants’ ‘psychological aspects’, in
Pateman’s words, or, as I choose to call it, their ‘democratic ethos’.
Development of a democratic ethos will reveal itself in the form of
changes taking place in the participants, regarding their interests, their
identities and their views.
The problem with the defined goal of such changes as defined in the
participatory tradition is that it was directed towards the identification
of a predefined substantial common good. This would lead the
researcher to raise inquiry as regards the extent to which psychological
changes moved in the desired direction or not. If not, or if not to a suf-
ficient degree, it could be concluded that the process has not realised
(all) it’s democratic potentials, and could be so criticised.
In the absence of such a ‘the only one right answer’-standard for
measuring the democratic impact of democratic learning in governance
networks, things become slightly more complicated. Changes of points
of views and identities are still appreciated to be seen as central to
democracy. As Connolly puts it, ‘[d]emocratic politics, then, is probably

the site of a tension or productive ambiguity between governance and
disturbance of naturalized identities. It thrives only while this tension is
kept alive’ (Connolly 1993: 208). Based on the thesis that all identity is
ultimately contingent, what we should appreciate about democracy is
not merely its capacity to produce ‘governance’, that is, the actual deci-
sions and results. The very disturbance of ‘naturalized identities’ result-
ing from the interaction with others having different interests and
perspectives (might) lead to (greater) awareness of one’s own contin-
gency and therefore a greater acceptance of others. Rather than looking
for changes in the involved participants’ identities that are in accor-
dance with a pre-established ‘correct answer’, one ought to be sensitive
to changes in democratic identities that move towards an increased
acceptance of the contingency of one’s own perception of the common
good, and hence an increased acceptance of different views. Even though
such democratic standards are difficult to measure, they ought to be
taken as signs of participatory democracy working well.
Of course not all changes bear witness to ‘a learning democracy’, just
as the absence of such change does not necessarily point to a lack of
democracy. In judging which does and which does not, the analysis
must refrain from stating a substantial common good to judge the regis-
tered changes against. The failure of a given process to lead to what the
researcher (or single parties within the process) perceive to be the com-
mon good is not a proof of it’s undemocratic form. Decisions involve
excluding certain interests and points of views, and presumably the
parties holding the excluded views do not experience the decision as
rational or in accordance with the common good. Rather than directing
the analysis towards substantial outcomes of the processes, one should
focus on some of the aspects of the process leading up to the decision
making into account, the democratic ethos.
Participants of course have a legitimate right to stick to their points of
view and well established identities. What should be demanded is, first,
that the participants accept the possibility of loosing, but of course with
the right to keep fighting for their beliefs; and second, that the partici-
pants are willing to listen to each other. Even though change is merely a
likely – as opposed to a necessary – outcome of a democratic process, the
elements making such change possible are something that must be part
of any democratic process. For a process to be democratic, all points of
view should be made clear, and the participants should be willing to lis-
ten to each other, rather than forcing one’s will trough. This is what
I refer to as ‘a democratic ethos’.
Participants actually persuading one another can hardly be a demo-

cratic demand, and many issues cannot be settled in agreement;
however, to be democratic, a (deliberative) process of stating points of
view and listening to the others should precede decision-making. When
analyzing concrete governance networks, attention should then be paid
to the degree to which all of the affected parties are heard and are able to
influence the setting of the agenda and the degree to which others are
willing to listen and take different views into consideration. This focus in
the participatory tradition on equal access and the debate of all points of
view is of great importance for democracy, but it should be stressed that
consensus cannot be elevated to a sign of a well-functioning democracy.
In concrete empirical analysis, it is difficult to measure the willingness
to listen to one another in a given network. The proof of one’s willing-
ness to listen cannot be measured by the degree to which one changes
one’s mind, since that is not the necessary outcome of a democratic
process. Subsequent to thorough consideration, one might choose to
maintain one’s original view. So the proof of the willingness to listen
cannot be applied in terms of signs of change in the involved partici-
pants. Accordingly, this standard remains a somewhat ‘fluffy’ demand,
probably only open to highly interpretative forms of judgement and
analysis: is it the overall impression that there is a culture of debate and
mutual acceptance? One might seek insights of this nature, either by
observations of networks ‘in action’ or through interviews. Group inter-
views will likely be revealing, but individual qualitative interviews can
also help highlight this aspect (Hansen et al. 2006; Sørensen and
Damgård 2006).
Learning democracy and governance networks

Learning democracy and listening to the other are probably the demo-
cratic norms established by the participatory tradition according to
which governance networks are most likely to score well. Obviously
there are no guarantees, but since networks often include actors from
outside the established set of political actors, such networks are likely to
increase the democratic learning of those involved.7 However, such pos-
itive ‘internal’ processes might be countered by other aspects of gover-
nance networks, such as the patterns of exclusion they bring with them.
This is the topic for the next section.
Equality
Equality is one of the basic norms in the participatory tradition. For

example, Pateman terms ‘full participation’ the situation in which ‘each
individual member has equal power to determine the outcome of

decisions’ (Pateman 1970: 71). Efforts to ensure equality raise two prob-
lems, each significant for governance networks. The first is the question
of how to ensure equal access to participation in decision-making bod-
ies, which is the general problem of inclusion. The second problem con-
cerns ensuring equality between the members within decision-making
bodies, in this case governance networks.
Access to/inclusion in networks

According to the participatory tradition, in order to ensure equal access,
participation must be widespread, something close to ‘full participation’
or participation by all. Not necessarily participation everywhere, all the
time, but in order to lead to the desired goals of learning and delibera-
tion, a significant number of the citizens must participate. However, the
ideal of ‘full participation’ has met substantial criticism8 and is indeed
difficult to imagine in a contemporary differentiated mass society. How
could such a society function if based on active participation by all
citizens? Equal participation must be reformulated, and Iris Marion
Young’s work represents an effort to redefine the participatory claim of
the full access of all in a manner that might apply to the question of
how to evaluate patterns of inclusion and exclusion in governance
networks (2000).
Young formulates the problem of equality in terms of the inclusion of
the affected: ‘A democratic decision is normatively legitimate only if all
those affected by it are included in the process of discussion and deci-
sion making’ (Young 2000: 23). Governance networks often have a nar-
row functional scope that would quickly be at odds with a claim of ‘full
participation by all’. Young makes it possible to rearticulate the logic of
equality in a manner rendering it more sensible to the way governance
networks function: access ought to be granted to all of the affected, not
to all citizens. It does not follow, however, that governance networks
actually do meet Young’s norm of inclusion of the affected, but this way
of posing the question makes engaging in concrete analysis much more
productive.9
When dealing with the notion of ‘inclusion of the affected’, however,
two problems must be taken into account. First, we must consider how
it is determined who the affected are. This is not an easy task: in one way
or another, we are all more or less affected by most of the decisions made
somewhere in the world. If we were to follow the consequences of this
point, Young’s reformulation would not really help us formulate criteria
for equal network participation. Her answer to this problem is to state
that one should be ‘significantly’ affected, that is, ‘that decisions and
policies significantly condition a person’s options for actions’ (ibid.).

While this qualification does help somewhat, it merely displaces the same
problem. Whose conditions are significantly affected remains an open
question, and with it, who should therefore be included in the decision-
making process.10
The notion of ensuring the inclusion of the significantly affected
generally points to the problem of representation in, and publicity
around, governance networks. These are serious democratic questions;
however, they lead to some extent outside of the participatory tradition.
I shall therefore return to them towards the end of this article. What we
must consider at this point is how to approach the idea of the inclusion
of the significantly affected in concrete analyses of governance net-
works. First, one ought to attempt to establish whether different actors
themselves believe that they are affected – but not included – in the net-
work. If this is the case, if there are explicit criticisms of exclusion in the
field under study, the conclusion is rather straightforward: in such cases,
the network does not meet the participatory standard of inclusion.
Often, however, one might be facing a situation without explicit criti-
cism of exclusion. In the case in which people are well aware of the net-
work and its effects for them – but simply choose not to participate –
this absence of participation cannot be regarded as democratically
illegitimate exclusion. What matters here is whether the choice to
participate or not is an ‘empowered one’, that is, based on an actual
opportunity to be included.
There are two basic forms of being disempowered vis-à-vis governance
networks. First one might be unaware of the existence of the network. If
such unawareness is caused by lack of public information about it, it
might effectively mean that all those affected by it, are disempowered.
Even in situations where not all affected are ‘disempowered by unaware-
ness’, lack of publicity around the networks is definitely an undemocra-
tic trait. Secondly, affected actors might be aware of the network, but are
unable to influence it. This might be due to a more general lack of
resources, including of course the case where one due to a relatively
weak position one is unaware of the existence of the network. In both
cases the networks fail to live up to the norm of equality.
When doing concrete analysis, it will often be difficult to judge
whether lack of influence are caused by undemocratic disempowerment.
Evaluation must rely on concrete judgements from case to case. However,
it is important to keep in mind that absence of participation is not in
itself necessarily a sign of democratic illegitimacy. As long as the choice
not to participate is an empowered one, there is nothing undemocratic
about it. It is simply a result of the high number of potential channels of

influence facing us all in modern societies.
Equality within networks

The second part of the norm of equality regards the processes transpir-
ing within the governance network, that is, the relations between its
members. Equality among the participants within decision-making
bodies are often pointed to as one of the (potentially) democratic
aspects of governance networks, since members of networks are said to
be in a less hierarchical relationship than is (often) the case in institu-
tions of representative democracy. Networks are said to be linked
together by the interdependence of the members rather than by formal
rules. Even though the relationships between the participants might be
less hierarchical in a traditional sense, it might be assumed that asym-
metries in resources among the participants might produce asymmetric
patterns of influence within a network. There will undoubtedly always
be asymmetry in the distribution of resources. There will always be
imbalances, and the interesting problem is to what extent such imbal-
ances overlap, creating more less systematic inequality in resources and
influence. Rather than demanding strict equality (which is in all cir-
cumstances impossible), critical assessment should raise inquiry as
to the degree to which asymmetries in the patterns of interdependen-
cies within a governance network lead to significant differences in
influence.11
Equality and governance networks

Governance networks hold a mixed relation to the norm of equality. To
a certain extent, easy access or broad inclusion could be said to run
contrary to the ‘purpose’ or reasons for the formation of most networks,
that is, most problem solving calls for limited participation. I take it that
most governance networks are not ‘spontaneously participatory’, but
rather rest on (often unclear) forms of representation (see Esmark’s con-
tribution to this volume). There might well be networks that are rela-
tively open to broad participation (as in the city renewal programme
mentioned earlier), but such networks tend to be a minority. Viewed
from the perspective of the participatory tradition, this is obviously a
shortcoming. If limited participation is the result of empowered
decisions by the affected, this is not a democratic problem; however, if
it is the result of the exclusionary manner of functioning in the network
and/or a lack of publicity surrounding the networks, this is quite
problematic from any democratic point of view, and highly so in the

participatory perspective.
If we turn to the processes inside the networks, we might expect them
to fare better. Networks are founded on interdependence between the
members, but even though the interdependence probably rules out
cases of complete dominance by some members over others, equality
between them, on the other hand, cannot be taken for granted. In
relation to differences in resources, the answer must also be concrete
analysis asking questions in terms of degrees: to what extent does this
network live up to the standards of inclusion? To what extent do the
differences in resources lead to undemocratic differences in influences?
Conclusion
I have specified three norms of participation in this article that can be

utilized in the concrete analysis of networks within a de-universalised,
re-articulated version of the participatory tradition: aiming for some-
thing common; learning democracy in terms of rules and practises and
of a democratic ethos; and equality as the inclusion of the affected in
governance networks and as relatively equal patterns of influence
between the members of the networks. Even though they have been
disarticulated from their original framework of consensus around a
substantial common good, these norms remain essential to democracy.
Moving beyond the traditional participatory perspective, such a frame-
work has two major consequences for the study of governance networks.
In the first place, it generally makes it meaningful to investigate the
democratic quality of concrete networks. Raising the traditional univer-
salistic claims to networks would presumably render them illegitimate
from the outset. Secondly, it changes the way in which one questions
the democratic legitimacy of governance networks. Instead of asking
whether networks are legitimate or not, one poses questions about the
extent to which they function in accordance with the different norms
from the participatory tradition. Obviously such questions can only be
answered in concrete analyses of governance networks and will proba-
bly turn out to be quite a complicated matter involving a set of pro
et contras.
One set of problems that will conceivably turn up in such analyses is
the extent to which the weakness of one or more of the principles might
be cured by the strength of one or more of the others. Even though it is
not necessarily so, it appears likely that networks scoring high on the
principles of participation do so in terms of ‘learning’ democracy, that
is, on empowering the members and providing forums that promote

willingness to listen and mutual respect due to the existence of
(although of the asymmetrical) interdependencies between the network
members. In many cases, however, these positive sides are likely to be
countered by unequal patterns of access (strongly limited inclusion in)
the network and/or highly limited publicity around the decision-
making processes (Pedersen et al. 1994). This is not an unlikely scenario
and raises the question as to whether the norm of learning democracy
might weigh more than the norm of equal inclusion.
Even though democratic norms can be achieved to a greater or lesser
degree, I find it difficult to see how even the most empowering network
relations can compensate for a lack of equal access of the affected to
participate in governance networks. However, broader means of provid-
ing access or inclusion than those traditionally pointed out by the
participatory tradition must be considered. As I have briefly outlined in
the above, democratic inclusion does not necessarily involve active
participation, but rather an informed or empowered choice.
Obviously, a central precondition for such an empowered choice is
the presence of publicity in and around governance networks. It is only
possible to form an empowered choice to the extent that the affected
know what is going on in and around the networks. It is not difficult to
understand why the participatory tradition has not really placed
emphasis on publicity. Publicity appears as a prerequisite only in the
case of representation, making a distinction between representing and
the represented. This distinction is ultimately what the participatory tra-
dition attempts to do away with, and therefore publicity is not really
required. However, modern democracy in general – and governance net-
works in particular – is, as pointed out by Esmark, of a representative
nature. It is therefore pertinent to ask whether governance networks
require representative forms of appointment, such as voting and the
like. Not providing a conclusive answer to this question, it is worth
emphasizing that publicity is a necessary prerequisite for representative
democracy as well as for participatory democracy; without that, not
even the best ‘learning processes’ within the governance networks
qualify them as democratic.
Notes
1. One example is the Danish ‘city renewal programme’, explicitly trying to
establish network politics in order for the renewal (better) to meet the inter-
ests and expectations of the inhabitants. The democratic merits of these
experiments were to a large extent based on the absence of formal rules of

access, forms of decision making etc. (Bang and Hansen 2005).
2. A critique of a substantial common good opens the question whether the
position adopted in this chapter is a procedural one? The answer to that
question is affirmative if by procedural we only mean that democracy (even in
principle) is unable to lead to only one correct ‘substantial’ answer. The
answer is in the negative, if procedural is taken in any strong sense, simply
because to agree on a set of procedures requires a whole set of substantial
agreements as well. Agreements on procedures cannot be established strictly
procedurally. My position is not strictly anti-consensual, but rather accepts
Chantal Mouffe’s argument that democracy needs consensus on the ‘rules of
the games’. But firstly, such a consensus cannot be ‘universal’, but will always
be established through exclusions; secondly, it will always give rise to a set of
plural or particular interpretations, leading to quite diverse ‘substantial’
outcomes (Mouffe 1993).
3. All translations of Hal Koch are my own, ADH.
4. Cf. Macpherson who talks about the demising room for compromise as a pre-
requisite of participatory democracy (Macpherson 1977: 114). See David Held
for a general critique of the idea of ‘the end of politics (Held 1987: 105f).
5. Though not all. E.g. Iris Marion Young states that ‘knowing that they are
answerable to others, and that they are mutually committed to reach agree-
ment, means that each understands that his or hers best interests will be
served by aiming for a just result’ (Young 2000: 50 italics added). She also crit-
icises the aggregative model for being ‘sceptical about the possibility of
normative and evaluative objectivity’ (ibid. 21). Cf. Pateman’s analysis
of the notion of Rousseau’s notion ‘forced to be free’, as pointed out by
Cunningham. Pateman interprets this as ‘the strength to look to the com-
mon good’ (rather than something one is ‘forced to’) gradually brought about
through the participatory process’ (Cunningham 2002: 132).
6. Jon Elster has criticised a position that claims that the primary aim of partici-
pation is the educative one, i.e. the process itself (Elster 1983). His point is that
learning democracy is necessarily a bi-product (Elster 1985), something that
can only be achieved to the extent that it is not seen as the primary purpose
of the action: we do not learn democracy if we participate with the intention
of learning – we only learn it if we participate with the intention of making
politics, decisions etc. Even though some proponents of participation defi-
nitely are struck by his critique I think that Pateman (mostly) avoids it. C.f.
her criticism of J. S. Mill (1991 [1861]), that exactly connects influence and
power to learning (Pateman 1970: 33).
7. In a Danish context this was also some of the conclusions from a large project
on ‘democracy from below’ (in which the concept of democratic empower-
ment is also developed). C.f. (Bang, Hansen et al. 2000; and Sørensen and
Torfing 2000).
8. C.f. from an otherwise sympathetic position, Walzer (1989). It should be
noted that both Pateman and Macpherson are aware of the impossibility of
‘full participation’ in modern complex societies (Pateman 1970; Macpherson
1977: cp. 4).
9. Young’s focus on the affected are also productive beyond the focus on
governance networks, since it opens the possibility of moving beyond the
boundaries of nation states that (so far) constitute citizenship, raising issues
of democracy across formal divides of ‘demoi’ as Bohman puts it (Bohman
2005).
10. In the final analysis such determinations are undecidable, and as such deci-
sions involving exclusions. This makes them strictly political, something
that Young does not pay sufficient attention to. However, this does not
change the potential productivity of the concept in concrete analysis. It only
makes the political aspects of such analysis themselves more obvious.
11. One should bear in mind that the participatory tradition never really it self
has come to terms with the presumably high demands placed on participants
in terms of what Young calls ‘articulatedness’ which in case of its generalisation,
itself would lead to exclusion (Young 2000: 37f).
15
Networks and Democratic Ideals:
Equality, Freedom, and
Communication1
John S. Dryzek
Democratic theory has historically proceeded under the assumption

that the proper – and perhaps exclusive – locus of political authority is
the sovereign state claiming exclusive political authority over a defined
territory and population. A well-defined demos can therefore accom-
pany the sovereign state, with a claim to popular control over policy
decisions that is fairly straightforward – at least in theory, if rarely in
practice. The democratic ideal of political equality can then be defined
in terms of the equal capacity of all citizens in the demos to exercise
control over policy decisions. Additionally, state democracy in practice
is almost always liberal democracy. And liberal democratic theorists can
specify a number of rights – freedom of thought, expression, association,
and assembly, more controversially rights to private property and sub-
sistence – necessary to make such a system work.2 Public authority so
constructed constitutes a relatively neat package. Those wedded to such
a picture greet any departure with horror. So for example Lowi (1999)
condemns the cooperative environmental governance applauded by
Sabel et al. (1999) as an abdication of public authority that allows stake-
holders to generate outcomes that suit themselves – but at the expense
of a public interest properly defined at the highest levels of state
government.
Governance networks (as defined in Chapter 1) complicate this neat
picture of public authority quite substantially. Networks can be general-
purpose, but often they are issue-specific, often informal, occasionally
transnational. But if they are consequential, then like all sources of
authority in today’s world their legitimacy ought to rest on their demo-
cratic qualities. So the question asked in this chapter is: can the key
(liberal) democratic ideals of popular control, political equality, and
rights to basic political liberties be redeemed in the context of governance
262
Networks and Democratic Ideals 263
networks? The simple answer is ‘no’. I will outline the reasons for this
simple answer, then proceed to a cautious ‘maybe’. But what the rise of
governance networks ultimately demands is a conversation with demo-
cratic theory about what democracy can and should mean (though not,
I hasten to add, any new model of democracy; there are already too
many models of democracy).
Applying the standard democratic principles to networks
‘Political equality’ is on the face of it hard to operationalize in a net-

work. Equality as an ideal is contested within democratic theory, but
most conceptions of political equality define it in terms of citizens’
proper positions in relation to the construction of sovereign authority:
citizens stand formally equal before this authority, even if they are
unequal in material endowments. Equality is a crucial aspect of democ-
racy largely because individuals cannot easily exit the state; given that
the state looms large and inescapable in the lives of members of the
demos, the democratic ideal demands that they have equal capacity to
influence what it does to them. (Of course, the state does different
things to different people, so in practice the pursuit of equality will
often have to be issue-specific.) Networks generally do not loom so large,
and it may be possible to escape the consequences of their decisions.
Individuals can choose to exit networks in relatively costless fashion.
They can less easily choose to exit the state, except by emigration, and
even if they turn their back on the state, it is likely to retain a very strong
interest in them. Political equality as an ideal is less compelling in
voluntary associations in which individuals are mostly only part-time
participants.
‘Popular control’ would normally be operationalized as the degree to
which policy decisions are responsive to the informed wishes of citizens
assembled in the demos. But in a network there is no single demos.
Networks may produce outcomes that apply to a demos as convention-
ally defined (coterminous with the state). Or their outcomes may apply
to a subset of the members of the demos; or apply to multiple demoi
across state boundaries. Applying conventional notions of popular control
over networks would imply their complete subordination to the state,
such that popular control could be exercised at one remove. Democracy
at one remove is problematic to begin with; and any such subordination
would reduce governance networks to tools of the state. This may be
how some networks are set up, though governance networks can extend
well beyond such instrumental use.
‘Freedom’ hardly makes sense as an attribute of a network itself. Or at

least the liberal concept of freedom as negative liberty, guaranteed by
rights against the arbitrary power of the state and against other members
of the polity, hardly makes sense. A network may be parasitic upon free-
doms secured by a liberal state or states; but the idea that freedom can
be somehow secured within a network itself is problematic, given the
absence of sovereign authority. A republican conception of freedom as
non-domination (Pettit 1997) perhaps fares somewhat better.
Republicans have typically been as wedded to the sovereign state as have
liberals, but their conception of freedom travels better to networks. We
could evaluate networks in terms of the degree to which non-domination
is institutionalized in their practices. Given any relatively informal
character of networks, we should not expect non-domination to be
institutionalized in laws and rights; rather it is a matter of observing the
degree to which non-domination is respected in the actual operation of
the network, and the degree to which individuals feel absolutely confi-
dent that nobody has the capacity to exercise arbitrary authority over
them. Republican freedom rests on the idea that no actor has the capac-
ity to dominate another – so merely observing the absence of overt dom-
ination is not enough to be at all sure that it is present, for some actors
may still fear others and adjust their behavior accordingly. Analysis
must cut deeper, to ascertain that no actor is indeed capable of domi-
nating another, and that no actor feels this threat. Obviously this is an
ideal, from which real-world networks will depart to greater or lesser
degree.
Beyond lingering statism in democratic theory
On the face of it, then, it seems hard to transport into networks some of
the familiar constituent criteria of democracy from their traditional
home in the demos tied to the sovereign state. But democratic theory is
a lively and multifaceted enterprise. So is there any way to move beyond
the application of some very traditional criteria, and so bring demo-
cratic theory into more fruitful conversation with networks? The answer
is a very qualified yes, but only once we move beyond the lingering sta-
tist disposition in democratic theory. Consider three cases in point
where this disposition lingers: deliberative democracy, cosmopolitan
democracy, and critical theory.
Deliberative democracy has been the dominant theme in democratic
theory for the past decade and a half. Deliberative democracy began as a
challenge to established aggregative models of democracy. Indeed, it was

quite common to contrast deliberative and aggregative democracy, with
aggregation being the dominant theme in those variants of liberal dem-
ocratic theory (notably those influenced by market economics) that take
preferences as given, defined prior to political interaction. In its empha-
sis on informed and critical citizen deliberation, deliberative democracy
was a successor to models of participatory popular in the 1960s and
1970s, perhaps requiring innovative institutional homes. However, by
the late 1990s, many deliberative theorists were emphasizing the insti-
tutions of the liberal state – notably legislatures and courts – as key
homes for deliberation. This change was celebrated by Bohman (1998)
as a central aspect of ‘the coming of age of deliberative democracy’. John
Rawls was identified as one of the sources of deliberative democracy,
even though his kind of deliberation requires only cogitation on the
part of a single individual. In the 1990s Rawls (1997: 771f; 1999: 139)
did eventually declare himself a deliberative democrat – but saw the
United States Supreme Court as an exemplary deliberative institution. If
deliberative democracy is moving to a stress on courts and parliaments,
it is correspondingly less capable of engaging governance networks.
Cosmopolitan democracy as developed by David Held (1995) and his
associates at first looks as though it escapes the statist confines of estab-
lished democratic theory. The basic justification for cosmopolitan
democracy is that power has escaped sovereign states, and now lies in
the international system, where democracy must follow. However, the
democratically constituted transnational authorities that cosmopolitans
seek look very familiar: legislatures, courts, referenda, and executives,
leading eventually to a global parliament that holds all global bodies
accountable, accompanied by a global legal system. Cosmopolitans still
seek a single overarching demos, even though it is no longer tied to the
state. The international system as currently constituted contains many
governance mechanisms, such as networks, regimes, and discourses,
that help constitute what James Rosenau (1992), Oran Young (1994) and
other international relations scholars call ‘governance without govern-
ment’. Cosmopolitans want to wipe these mechanisms out or subordi-
nate them in the name of very traditional conceptions of political
authority. (I note in passing that international relations scholars who
praise governance without government typically have no interest in
democracy; like all international relations scholars, they are preoccupied
with order, not democracy.)
Critical theory has in the past decade (with Habermas 1996 central)
turned its focus to democracy, with a crucial test of its claims being how
it can cope with a globalizing world. In discussing this condition,

Scheuerman (2006) faults ‘overly defensive’ versions of democracy that
stress communication in the public sphere as a defense against corporate
globalization. Scheuerman poses the issue as either popular control over
institutions capable of exercising binding authority, or influence with-
out power, which is all the public sphere yields. He dismisses in passing
‘An insufficiently critical homage to (non-state) “governance” [that]
should not lead us to obscure the indispensable functions existing state
and new state-like institutions will need to perform in achieving novel
forms of self-legislation, and the rule of law.’
In short, deliberative democrats, cosmopolitan democrats, and critical
theorists often seem unwilling or unable to let go of centralized state or
state-like institutions; and to that extent, their theories have limited
applicability to governance networks.
Networks and the communicative aspect of

democratic theory
However, if we take seriously the fact that today, at least to a substantial

degree, ‘Talk-centric democratic theory replaces voting-centric demo-
cratic theory’ (Chambers 2003: 308), matters begin to look more prom-
ising. Given their polycentric character, networks rest on relatively
egalitarian communication across different actors (that is, they are
much more symmetrical than hierarchies, where communication up
and down levels takes on quite different form). So it ought to be possi-
ble to apply critical standards to both the procedure and content of
interaction within networks. When it comes to procedure, it is possible
to apply standards of communicative rationality from Habermas. That
is, interactions should be as free as possible from domination (the exer-
cise of power), strategizing, deception and self-deception. The ideal
would then be a network composed of competent and reflective actors
engaged in a search for mutual understanding to coordinate their
actions. Of course, networks in practice will all fall short to greater or
lesser degree.
The Habermasian standards are controversial, and would be rejected
as excessively rationalistic by agonistic democrats. For agonists, the key
to democratic communication is respectful engagement across differ-
ence. Agonism is intended in Mouffe’s words to convert enemies into
adversaries, fighting into critical engagement, and so antagonism into
agonism. The required attitude comes about in moments of conversion
rather than persuasion, as the other is seen in a new light. Agonists are
preoccupied with engagement across deep difference, though in principle

many kinds of differences can be affirmed and processed in agonistic
fashion. In many cases deep difference will not be an issue in a network;
in other cases it will (for example, in developing policy on what to do
with minority non-liberal cultures in a liberal society).
One case where deep difference does arise in a network context is
analysed by Schlosberg (1999). The environmental justice movement in
the United States developed from a series of local actions against corpo-
rations and governments generating hazards such as incinerators, waste
dumps, and toxic pollutants, especially as inflicted on low-income and
ethnic minority communities. In its early days in the 1980s the move-
ment confronted mostly unresponsive governments, as well as uninter-
ested mainstream environmental organizations. Thus it organized as a
kind of counter-network, very different to the hierarchical organization
of the mainstream environmental groups. Bodies such as the Citizens’
Clearinghouse on Hazardous wastes and Southwest Network for Economic
and Environmental Justice were crucial in mobilizing the network. Deep
difference existed between the different communities facing common
hazards: for example, Hasidic Jews, African-Americans, and Latinos in
Brooklyn. To operate effectively, the movement had to develop informal
norms of respect across difference, a kind of critical pluralism. This kind
of movement network might seem a bit different from governance net-
works, which if they are based on communities of interest or culture are
unlikely to face the same kinds of challenges negotiated (with some
success) by the environmental justice movement. However, with time
this movement itself developed connections with both government
(for example, through the US Environmental Protection Agency’s Office
of Environmental Justice) and the mainstream environmentalist
organizations. Sometimes these connections were problematic (as when
mainstream groups stepped in late in particular campaigns and negoti-
ated deals with polluters, leaving out local grassroots groups). But
the kind of democratic ethos developed in the more clearly opposi-
tional days of the movement did persist into this engagement with
government.
This kind of engagement with government would not necessarily
please agonist theorists. The agonistic ideal of democracy concerns how
individuals and groups relate to each other. It is first and foremost the
respectful content of interaction that matters. Agonists have little inter-
est in collective decision of the sort that states and networks can pro-
duce, and which is seen as the essential task in more conventional
models of democracy as decision-making procedure. However, agonists
might have less of a problem in connecting to governance than they

would to any linkage to government of the sovereign statist variety.
The standards of agonists and Habermasians are not the only ones
that could be applied to communication within networks. A relatively
relaxed approach would go beyond the Habermasian privileging of
argument to permit all forms of communication. Iris Young (2000:
52–80) refers to ‘inclusive political communication’ that would also
cover greeting, rhetoric, and narrative. Performances, jokes, and gos-
sip might also be welcomed. However, all such communication
should be (a) capable of inducing reflection, (b) non-coercive,
(c) capable of connecting particular experiences to some general point
(Dryzek 2000: 68). If no reflection is induced, the communicative task
is incomplete. If communications are coercive, basic democratic (and
liberal) principles are violated. If communications cannot connect the
particular to the general, they may reinforce and harden the position
of an enclave or subculture, but not reach those with different
viewpoints.
An emphasis on communicative sources of democratic standards is
not however completely adequate, because one might get exemplary
communication within an unrepresentative elite (just as in Ancient
Athens, with a highly restricted polis featuring communicative excel-
lence internally, but repression externally). Thus we need to look at the
‘who’ as well as the ‘what’ of democratic communication.
Who communicates
Resources can be found in deliberative democracy to answer this ques-

tion too. The theory of deliberative democracy arrived as an account of
democratic legitimacy. That is, a decision is regarded as legitimate to the
degree all affected by it have the right, opportunity, or capacity to par-
ticipate in deliberation about the decision in question (Manin 1987:
352; Cohen 1989: 22; Benhabib 1996: 69). While there can be problems
in operationalizing this ideal (Parkinson 2003), as a criterion it can be
applied as a matter of degree. We can think of networks as more or less
inclusive of those affected by a decision, as well as more or less delibera-
tive when it comes to the participation of those affected. So the key idea
here is a ‘democracy of the affected’ (Eckersley 2000). That is, the rele-
vant demos for any decision consists of those affected by it. From the
point of view of any one citizen, influence should be concentrated on
the issues that affect him or her most deeply. There would, then, be mul-
tiple demoi (in James Bohman’s terms), not one single all-purpose
demos. These demoi could exist at different levels, below, above, and
across the state.
We are now in a position to rethink democratic ideals of popular con-
trol and political equality. In a network, popular control can be opera-
tionalized in terms of responsiveness of collective decisions to
deliberation on the part of those affected. One complicating factor here
is that ‘the affected’ is not always a well-defined group, especially when
there are degrees of affectedness. Taking this difficulty into account,
political equality can be operationalized in terms of inclusion propor-
tionate to the degree an individual or group is affected by the relevant
collective decisions. While this operationalization might seem to allow
inequality – because those most affected have the most say – this
inequality is likely to cancel out across different networks or different
issues.
There is a further complication here when it comes to networks set up
by or in the shadow of the state. If such networks influence or determine
how public funds get spent, then ‘those affected’ encompasses everyone
who pays taxes. Public choice scholars would regard it as crucial that
networks not be allowed to solve their own problems – possibly in
exemplary democratic fashion when it comes to inclusion and voice of
all those affected – if the network as a whole is engaged in a collective
rent-seeking exercise at the expense of taxpayers in general. Or to put it
in the very different language of critical theory: interests that look ‘pub-
lic’ and ‘generalizable’ within the network may look very private and
partial from the point of view of the state.
Beyond models of democracy
These extensions notwithstanding, it would appear that there is an

uneasy fit between existing models of democracy (in all their variety)
and governance networks. Does this mean that we need a new model of
democracy for a networked world? Not necessarily. Perhaps such a world
makes models of democracy obsolete; not because they are democratic,
but because they are models. A model is a fully-specified ideal set of con-
ditions, a comprehensive whole with logical connections between its
parts. Advocates of particular models of democracy normally present
them as universally applicable to all situations. The very idea of a model
implies something that is fixed and static. In a world of multiple demoi
and multiple governance mechanisms, perhaps the time has come to
move beyond models of democracy, and toward thinking instead
about processes of democratization that can be applied in any context,
without reference to any ideal end state. ‘Democratization’ here means

not the spread of liberal democracy to more corners of the world – the
conventional usage in comparative politics. Instead, it means deepening
of the democratic qualities of any situation, structure or process. Such
deepening can be recognized along three dimensions (Dryzek 1996):
franchise, scope, and authenticity.
Franchise refers to the number of affected actors engaged in control of
a situation. Normally the term is used in connection with voting, but its
meaning can be extended to participation of any sort. Thus an increase
in franchise connotes democratization in any setting, whether or not
votes are taken (of course in networks they usually are not). Franchise
can be linked to both the ‘popular control’ and ‘political equality’ crite-
ria. Greater popular control on the part of more equal individuals means
an increase in the effective franchise. Of course it is not any individuals
who matter here. Rather, it is the population of individuals who are
affected by relevant collective decisions. Advance on the franchise
dimension should be sought only within this population.
Scope refers to the range of issues under collective control.
Correspondingly, an increase in scope means a reduction in peremptory
claims on behalf of particular actions or values. For example, claiming
that an economic policy of deregulation, welfare state reduction, and
free trade is demanded by globalization is a peremptory claim. An
increase in scope would subject such claims to critical scrutiny, and
bring areas such as national security as well as the essentials of economic
policy under collective control. A network no less than a state can be a
means of collective control.
Authenticity means the degree to which control is substantive rather
than symbolic, exercised by informed, reflective, and competent actors.
Authenticity relates very closely to the key concerns of deliberative
democrats about the character of communication. There is no problem
at all in applying the authenticity dimension to networks as opposed to
sovereign authorities.
In applying these three criteria, several additional considerations
should be borne in mind. First, advance on any one of the three dimen-
sions should never be bought at the expense of retreat on any other of
the three dimensions. This is because there is no common metric that
would enable us to tell whether any such advance was worth it. This
constraint would also guard against the worries of (say) liberals about
extension of public control into areas traditionally demarcated as pri-
vate, which liberal rights protect. Feminists have also pointed out that
this ‘private’ realm can also be a site of oppression that should be rectified
by public means. A democratic adjudication of scope here might proceed

in terms of the protection afforded by franchise and authenticity: that
is, the empowerment of those at the receiving end of ‘private’ oppres-
sion, and the protection of the individual autonomy necessary for
authentic democratic participation. Second, advance in the long term
should never be bought at the expense of retreat in the short term; given
uncertainty, it is never clear that the long term benefit of (say) suspend-
ing democratic procedures in a crisis will in fact eventuate.
The contribution of governance networks to democracy
While it takes some hard work and conceptual stretching to do so, it is,
then, possible to apply democratic theory to networked governance.
However, democratic theory itself is transformed in the encounter, hav-
ing to let go of some key tenets, and bend others. All this stretching and
bending is likely to make some democratic theorists uncomfortable,
which explains their resistance to letting go of the sovereign state,
operating under a constitution, governed by the rule of law, whose
legitimacy rests on responsiveness to a single and well-defined demos.
It is probably less troublesome to start from the opposite direction,
and instead of asking how to apply democratic theory to governance
networks, to ask instead how governance networks can contribute to
democracy – or democratization. In light of the theory of discursive
democracy, the important thing about networks is not that they consti-
tute a polity-substitute, ripe for the application of polity-derived criteria
for evaluating democracy such as popular control, political equality, and
liberal freedoms. Instead, any particular network can be seen as a
site where discourses can or should meet, as part of larger processes that
help constitute discursive democracy. My conception of discursive
democracy here is an insistently critical sub-category of deliberative
democracy. Discursive democracy emphasizes engagement across dis-
courses in the public sphere. The results of this engagement can influ-
ence more formal authority structures (such as governments). But the
engagement itself can be intrinsically valuable in its constitution and
reconstitution of social relationships (see Dryzek 2000 for more detail
on how discursive democracy in this sense differs from deliberative
democracy that is beholden to liberal constitutionalism).
Of course there are times when a particular network may be charac-
terized by a single discourse, as opposed to engagement across multiple
discourses. Indeed, this may be the normal case. For example, those
engaged in environmental networks often subscribe to a discourse of
sustainable development. Those involved in international economic

affairs are coordinated largely by the discourse of market liberalism, or
neo-liberalism. Certainly the presence of a shared discourse can produce
substantial benefits, not least on the grounds of deliberative economy: it
is not necessary for participants to continuously debate first principles.
Moreover, a shared discourse is a low-cost way of coordinating the
actions of members of a network. Thus when it comes to efficiency and
even problem-solving effectiveness, there may be much to be said for
stabilization of a network through a shared or hegemonic discourse.
However, from the point of view of discursive democracy, a multiplic-
ity of discourses existing in association with a network is much more
healthy. Such multiplicity provides grist for the engagement across dis-
courses that is at the heart of discursive democracy. It means that net-
works can play a role in the production of public opinion in the public
sphere, not simply act as an arm of government. Multiple discourses
mean multiple points of entry for different sorts of actors. For example,
in environmental affairs, sustainable development discourse can be
exclusionary, especially when it is bent in a business-friendly direction
(as it has been lately). Welcoming a discourse of environmental justice
means a place for advocates of those at the receiving end of environ-
mental risks and hazards. Admission of green radical discourse means
representation for the advocates of ecosystemic integrity and non-
human nature, which are normally ignored in sustainable development.
The simple existence of multiple discourses in the vicinity of a net-
work does not however guarantee that they are contributing to discur-
sive democracy (or democratization). Much turns on how discourses
engage one another. This might be done through manipulative means.
Consider for example the ‘war of ideas’ that accompanies the US-led
‘war on terror’. The ‘war of ideas’ is a metaphor used by US leaders to
describe their intention to destroy discourses that provide comfort for
Islamic radicalism–along with other anti-American dispositions. The
fact that this ‘war’ is prosecuted in inept and generally counter-
productive fashion should not obscure the fact that its intention is to
advance the standing of pro-American discourses by any means possible;
propaganda, public relations, advertising, planting stories in the press,
cultivating sympathetic reporters, funding compliant media, and so
forth. The intention is most certainly not to engage other discourses
(including for example European sceptics about the use of military
power), but to defeat them. Sometimes the idea of ‘soft power’ (Nye
2002) is proposed as a gentler alternative. Soft power involves being
solicitous of the interests of others in the international system in an
effort to induce others to share one’s values and viewpoints, and so gen-
erate supportive actions. (There is also a ‘cultural’ aspect to soft power,
making use of US domination of the worlds of popular culture and
higher education.) However, soft power is treated by its American pro-
ponents as something for the US to impose upon the world; it thinks of
the rest of the world as tabula rasa in discourse terms, not already alive
with its own discourses. The only actor seen as the legitimate purveyor
of soft power is the United States, and that right is backed by the very
hard power of military and economic dominance. Advocates of soft
power are not interested in any deliberative exchange with the rest of
the world, only the bending of the world to US interests.
What matters, then, is not the simple existence of multiple discourses
in the presence of a network, but how they are engaged. If this
engagement proceeds according to principles of respect, reciprocity, and
equality in the capacity to raise and challenge points, it can help con-
tribute to discursive democracy. If it proceeds in terms of propaganda,
manipulation, spin, and public relations, it can undermine democracy,
discursive or otherwise.
Conclusion
The fit between governance networks and existing models of democ-

racy, especially those of a statist disposition, is not going to be an easy
one. The fit between governance networks and ‘talk-centric’ delibera-
tive, communicative, and discursive approaches to democracy is much
easier, though much remains to be done on how this connection can be
worked out in practice. I have argued that a discursive approach empha-
sizing discourse contestation can most fruitfully illuminate both the
democratic potential and democratic problems of governance networks.
Democracy is about a conversation between theory and practice, not the
evaluation and design of practices in terms of models. The conversation
between democratic theory and governance networks has begun, and it
promises to be productive and fruitful.
Notes
1. This research was supported by Australian Research Council Discovery Grant
DP0342795.
2. The two principles of ‘political equality’ and ‘popular control’ are the starting
points for democratic audits, which began in the United Kingdom a decade
ago and have since been applied in New Zealand, Australia, Canada, and else-
where. For a justification of these principles see Beetham (1999).
16
Democratic Accountability and
Network Governance – Problems
and Potentials
Anders Esmark
Democratic network governance?
This chapter concerns the democratic problems and potentials of

network governance. Thus, the article is not about the theoretical
elaboration of the concept of network governance, which is dealt with
in great detail elsewhere in this volume; rather, the question is whether
the use of networks by state and public authorities for purposes of pol-
icy formulation, implementation and delivery can be considered an
instrument not only of governance, but also of democratic governance.
Democratic theory offers an abundance of democratic values, stan-
dards and norms against which actual political decision-making
processes can be measured. The democratic standard employed in this
article is that of democratic accountability. When assessing network
governance from an accountability perspective, one should obviously
take note of the fact that the notion of accountability is deeply rooted
in the liberal doctrine of representative government. Correspondingly,
assessments of network governance from the viewpoint of democratic
accountability have often yielded negative results, or, at the very least,
considerable concern as to the democratic value of network gover-
nance. The essence of such concerns is captured nicely by March &
Olsen:
Policies are made in complex networks of actors … because many

participants contribute in many different ways, it is difficult even in
principle to identify who is responsible for political outcomes and
thus to establish political accountability. In particular, limiting
accountability to officials seems myopic in modern political systems.
Political outcomes are the product of ‘many hands’ (1995: 158).
274
Democratic Accountability & Network Governance 275
Such concerns are certainly not unfounded. In many instances, network

governance can be conducted in a manner rendering it difficult to main-
tain the standards of political accountability. Nonetheless, the contention
of this article is that network governance and accountability are not nec-
essarily adverse: network governance can even be said to hold great
potential for strengthening accountability. In order for this potential to
be realized, however, the theoreticians and practitioners of network gov-
ernance must develop a clearer notion of networks as representative insti-
tutions and conduct network governance accordingly. The democratic
norm of accountability is fundamentally premised on the idea of repre-
sentation. The democratic promise invoked by the norm of accountabil-
ity is not one of direct democracy, strong participatory democracy or
radical democracy; instead, it is one of proper representation.
Consequently, the article does not deal with theories of direct, partic-
ipatory or radical democracy. Rather, the article deals with strands of
democratic thinking that can be labelled as ‘postliberal’, provided that
we take this to mean strands of thinking critically engaged in the revi-
sion and further development of the central tenets of liberalism rather
than something beyond or in fundamental opposition to liberalism; in
particular, drawing on theories of functional representation and theories
of representation through deliberation in order to develop a notion of
accountability that is not limited to the conventional doctrine of elec-
toral representation. Theories of functional representation largely coin-
cide with what Sørensen and Torfing identifies as competitive democracy
elsewhere in this volume. The term ‘functional’, however, is meant to
stress that I take the autonomy of functionally delimited groups – rather
than competition – to constitute the core of this strand of thinking.
Theories of representation through deliberation refer to a subset of the
larger field of deliberative democracy, which do not take the core delib-
erative notion of a well-working public sphere of sustained moral or rational
argument to be an alternative to representation, but rather a particular
way of sustaining representation.
The article specifies three dimensions of democratic accountability
that must be considered in this respect: inclusion, publicity and respon-
siveness. The question of inclusion relates both to representatives (or deci-
sion-makers) and those being represented. Accountability requires that
it is possible to discern who the decision-makers are in relation to a
given decision and who should be counted among those represented by
the decision-makers in question. Secondly, the relationship between
decision-makers and those affected must be one of sufficient publicity.
Accountability requires a public sphere in which decision-makers argue
openly and remain ready to explain and assume responsibility. In other

words, accountability requires a public sphere and what might be
referred to as a discourse of accountability, which is to say a discourse of
providing explanations and assuming responsibility. Furthermore, pub-
licity requires that those affected have sufficient access to relevant infor-
mation, which is also occasionally referred to as transparency or
openness. Finally, accountability requires sufficient responsiveness.
Without the possibility of those represented to correct the action of the
representatives, that is, to impose sanctions, accountability remains
without value. For the sanctions to remain legitimate, however, clear
mandates specifying the desired course of action must be in place.
The article proceeds through the following steps: first, the notion of
accountability is discussed in general terms and more specifically as a
democratic norm and the roots of this norm in democratic theory.
Secondly, the relationship between accountability and inclusion is elab-
orated, followed by an assessment of network governance in this regard.
Thirdly, accountability is discussed in terms of publicity followed by an
assessment of network governance. Fourthly and finally, the brief con-
clusion aside, accountability in terms of responsiveness is made the
background against which to evaluate network governance.
Accountability as a democratic norm
Accountability is not necessarily democratic, nor is it even necessarily

political. The term ‘accountability’ refers both to the subject and the
object of an account. The subject of an account is the person being held
accountable: the person ‘subject to giving an account’, either to oneself
or someone else. The object of the account is that which is accounted for
in providing the account, be it events, actions, feelings, physical objects
and so forth. At the most general level, then, accountability implies a
subject accounting for an object, which is to say providing an explana-
tion as well as assuming responsibility for the object. Thus, accountabil-
ity implies the existence of more or less institutionalized and formalized
rules of making explanations and giving and assuming responsibility. At
the most general level, accountability is simply a particular type of rela-
tion between subjects and objects, which forms a basic existential prem-
ise of modern rationalized culture (Power 1997).
Conceived more specifically as a question of democratic governance,
accountability implies that decision-makers can be called upon by those
whom they represent to explain and assume responsibility for their decisions.
As such, accountability is historically linked to representation and even

more specifically to electoral representation as conceived within the
‘aggregative’ or ‘liberal-protective’ strand of thinking that count
Hobbes, Madison, Bentham and James Mill among its founding fathers
(Macpherson 1977; Held 2000). Here, accountability emerges as the
raison d’être of the electoral institutions of representative government.
The core of this democratic strand of thinking is that:
… governors must be held accountable to the governed through

mechanisms (the secret ballot, regular voting and the competition
between political representatives, among other things”) which give
citizens satisfactory means for choosing, authorizing and control-
ling political decisions (Held 2000: 88; bold added).
According to this well-established liberal doctrine of electoral represen-

tation, accountability is basically to be secured through what has been
called the ‘remarkably stable structure’ of representative electoral insti-
tutions: 1. rulers, those who govern, are selected through elections;
2. while citizens are free to discuss, criticize and demand at all times, they
are not able to give legally binding instructions to the government; and
3. rulers are subject to periodic elections (Manin et al. 1999: 3). The fact
that accountability is first and foremost a democratic norm framed
within the liberal doctrine of electoral representation invites the ques-
tion of whether the evaluation of network governance from the view-
point of accountability is in fact a futile exercise. Assessments from such
a perspective tend to produce a swift and negative judgment regarding
network governance, as networks ‘undermine the institutions of repre-
sentative democracy and the autonomy of civil society’ (see Sørensen
and Torfing Chapter 13 in this volume). According to the conventional
liberal doctrine, networks are in conflict with the OPOV-principle (one
person, one vote – formerly the OMOV-principle), with the autonomy
of civil society, with the sovereign position of elected government, with
the orientation political towards the common good and so on.
More specifically in terms of accountability, the danger inherent in
network governance, as seen from the liberal perspective, is that net-
works imply negotiation and shared decision-making, thus turning
actors who cannot be called to explain or assume responsibility into
de facto decision-makers. Networks are thus associated with both
obscure decision-making and ‘buck-passing’. Another threat is that rela-
tions of accountability come to be established within networks rather
than between elected representatives and their constituency. Within

network governance, elected representatives are likely to form relations
of accountability with fellow network members rather than with their
constituency – a critique well known from earlier debates about corpo-
ratism, iron triangles and so forth. As seen from the liberal interpreta-
tion of accountability, then, the undemocratic nature of network
governance appears a foregone conclusion and any further discussion
about the democratic potentials of network governance unnecessary.
However, the story need not end here. If we turn to the different
strands of postliberal thinking and move beyond the narrow doctrine of
electoral representation, networks may emerge as a democratic potential
rather than as a threat. Following postliberal thinking, we can conceive
of representation as extending beyond electoral institutions. The
postliberal alternative, though, is by no means a radical alternative sug-
gesting the end of electoral institutions. The point is simply that electoral
institutions are a necessary but insufficient condition of accountability.
Even in cases where electoral institutions are working well, they cannot
produce sufficient accountability. This is where networks enter the
scene: networks may alleviate the accountability deficiency of electoral
institutions. For networks to fulfil such a role, however, it is essential to
recognize, first, that networks should a supplement rather than replace
electoral institutions and, secondly, that networks themselves must
work within a framework of representation. Drawing on theories of
functional representation and representation through deliberation, I
now turn to a more detailed elaboration of the three criteria of demo-
cratic accountability – inclusion, publicity and responsiveness –
together with an assessment of the democratic problems and potentials
with respect to each of these criteria.
Accountability and inclusion
The question of inclusion is probably the most fundamental of ques-

tions in democratic theory and practice: who should be included in a
political community (and by implication: who can legitimately be
excluded)? In terms of accountability, it should be emphasized that
inclusion does not imply direct participation in the decision-making of
a political community. Rather, being included means being represented
in decision-making. In terms of accountability, more specifically, proper
inclusion means being able to assume the role of an accountability
holder, that is, someone in a position to hold someone accountable,
which in turn can be designated (although not exactly in the Queen’s

English) as the accountability holdees:
Our system of accountability has two types of people: Either you are
an accountability holder or you are and accountability holdee. It’s
great to be an accountability holder. It’s not so much fun to be an
accountability holdee (Behn 2001: 2).
Being an accountability holdee – or simply being held accountable –

may not be much fun, but it is nonetheless something that those who
make decisions on behalf of others must live with in a system of demo-
cratic governance. The most basic problem involved in any discussion of
democratic accountability is determining who can legitimately assume
the role as accountability holders in relation to a certain group of
decision-makers, who can in turn be designated as accountability
holdees. According to the electoral doctrine, legitimate relations of
accountability are first and foremost electoral relations between elec-
torates and their elected politicians. There may be some variations as to
whom politicians are actually seen as accountable to: only to those who
voted for them, to their regional constituency, to all eligible voters or to
all citizens in the political community (including non-voters), neverthe-
less, accountability remains for many of the liberalist persuasion a
matter of proper and well-functioning electoral systems (Przeworski
et al. 1999).
However, any political–administrative system includes a second rela-
tion of accountability. Besides the political accountability of elected politi-
cians to their constituencies, the liberal doctrine also acknowledges the
administrative accountability of administrators and administrative units.
Following the liberal doctrine, administrators are accountable first and
foremost to the political institutions (executive, legislative and judici-
ary). Administrators ought to accept the role of accountability holdees
in relation to the political institutions due to the fact that the latter has
been delegated the role of accountability holder through the popular
mandate at election time. However, the liberal doctrine also acknowl-
edges that elected politicians are not the only legitimate accountability
holders in relation to the administration. As rule-bound bureaucrats,
civil servants should be accountable also to the law, to standards of
proper conduct as defined their professional peers, to citizens and to a
somewhat abstract but nonetheless important ‘public ethos’ (Peters
1995; du Gay 2000).
However, postliberal theories of democracy have voiced critique of the

notions of political as well as administrative accountability suggested by
the electoral doctrine. The basic proposition that sufficient representa-
tion ought to be fundamentally recast as a question of affectedness plays
a vital role in both theories of functional representation and deliberative
democracy. At the core of theories of functional representation is the
notion that the establishment of functionally defined and self-governing
‘associations’ or ‘organizations’ are in fact equally (if not more) impor-
tant as the territorially based democracy of citizenship and elections. In
the associative democracy of Paul Hirst, the argument in favour of such
functionally delimited and relatively autonomous associations is their
capacity to ensure that those affected by decisions – the stakeholders –
are included in decision-making. According to a functionalist principle
of inclusion, stakeholders – rather than citizens – are the legitimate
accountability holders. However, associations based on a functionalist
principle of inclusion are not to take the place of conventional repre-
sentative institutions, but rather to supplement the latter primarily at
the local level (Hirst 2000).
The principle of functionalist inclusion is also important to demo-
cratic elite-theory, which takes the autonomy of functionally defined
elites to be the crowning achievement of western democracy (see
Esmark 2003 for a more elaborate discussion on this point). Ultimately,
the relative autonomy of elites is important on the grounds that it pre-
vents a system of rule from forming into a concerted power structure. In
other words, elite autonomy is seen to give a system of rule a ‘progres-
sive potential’ (Etzioni-Halevy 1993: 199). The ‘concerted power struc-
ture’ that democratic elite theory seeks to avoid is defined by two
well-known trademarks from corporatism, that is, the subjugation of
elites to the state and collusion between elites from different functional
domains. The democratically viable alternative to subjugation and col-
lusion is called coordination (Etzioni-Halevy 1993: 109–21). For demo-
cratic elite theory, accountability holders are sub-elites and public in
relation to the elites (accountability holdees) within functionally
defined domains and organizations such as schools, universities, hospi-
tals and so on.
The most comprehensive reframing of representation and accountabil-
ity is probably found in theories of representation through deliberation.
Some versions of deliberative democracy have a strong participatory
flavour to them, emphasizing the direct participation of citizens in the
public sphere as the true aim of deliberative democracy, but other
proponents of deliberative democracy hold that representation is both

necessary and desirable (Young 1997, 2000). However, adequate repre-
sentation is to be sustained through a process of deliberation in the pub-
lic sphere rather than simply through elections. We can thus conceive of
the relation between accountability holdees and holders as a relation
between ‘a special class of political deliberators’ ‘deliberating for all’ and
those being deliberated for (Gutmann & Thompson 1996: 129). In other
words, the role of accountability holder is no longer determined by
inclusion in a particular electoral system; rather, it is inferred from the
considerably more complex notion of being affected by a decision that
is the subject of a deliberative process:
If we reject, as deliberative democracy suggests that we should, the

view that representatives are accountable only to those who can vote
for them, then we face a formidable challenge in determining who
should count as the constituents to whom they owe an account
(Gutmann & Thompson 1996: 131).
As is the case with theories of functional representation, the principle of

affectedness is at the core of this relation. Rather than a functionally
delimited organization or association, however, the relation of account-
ability is to be established in an open and pluralistic public sphere
of sustained moral or rational argument. In theories of representation
through deliberation, accountability holders are therefore not consid-
ered stakeholders in a narrow functional sense, but as members of a public,
or in the words of Gutmann & Thompson, a ‘moral constituency’.
Gutmann and Thompson suggest three types of moral constituents
not present in an electoral constituency: non-residents (territorially
excluded), groups (excluded by principles of universality) and future
generations (temporally excluded) (Gutmann & Thompson 1996: 145).
The notion of a moral constituency presents a more inclusive notion of
legitimate accountability holders, certainly compared to an electorate,
but also as compared to most notions of a functionally delimited set of
stakeholders.
The possible set of accountability holders is also widened. Assuming
the role of representatives in a deliberative process may well include
others than the elected officials. In principle, the notion of deliberative
accountability extends to anyone claiming the position of a representa-
tive, such as journalists, organizational spokespersons, experts and so
forth, or anyone else assuming the role of deliberator for some or all
in the public sphere. Insofar as actors take the position of representa-

tives, they do in fact by implication become accountability holdees, not
just to their readers, their organizational members or their peers,
but also to the moral constituency. In fact, widening the field of eligible
accountability holdees may be an equally important democratic
challenge as widening the field of accountability holders. As stated ear-
lier, however, it is more fun being an accountability holder than an
accountability holdee.
First challenge: finding the holders and holdees
From the perspective of the conventional liberal doctrine, networks can-

not conform either to the standards of political accountability or admin-
istrative accountability. When elections constitute the only source of
political accountability, the most notable characteristic of networks is of
course that they comprise few politicians and have no electorates. The
network is not a representative body in electoral terms and thus, its
members are not subject to direct relations of accountability. In terms of
administrative accountability, the case looks no better. Even when civil
servants do participate in networks, they do so at the risk of becoming
accountable to network members and particularistic interests rather
than politicians, citizens and a public ethos premised on the notion of a
public good.
However, we need not turn to conventional liberalism to perceive of
network governance as problematic. In terms of inclusion, accountabil-
ity is essentially about assigning a sufficient degree of clarity to the roles
of accountability holders and holdees. A viable democracy requires rela-
tions between clearly defined accountability holders and clearly defined
accountability holdees. Irrespective of how the roles of accountability
holders and holdees are perceived substantially, getting decision-makers
to assume the role of accountability holdees in relation to specified and
clearly defined accountability holders is the most basic and simple pre-
condition implied in democratic accountability. In general, the poten-
tial problem with network governance in this respect is that the network
structure itself tends to blur the clearly defined roles of accountability
holders and holdees in favour of a situation in which each actor is
equally an accountability holder and holdee.
Such blurring should come as no surprise, since all of the known
definitions of network governance stress symmetry, interdependence,
horizontality and negotiations as prominent trademarks. The relation
between accountability holder and holdee, by contrast, is essentially
asymmetrical, hierarchical and non-negotiable. The accountability

holder is in a superior position – based on popular sovereignty – to the
holdee and stands in judgement over the holdee. In this sense, the
worst-case scenario from any perspective is that there is a fundamental
mismatch between the structures required by democratic accountability
and the structure of networks. Lacking a clear distinction, neither the
role of accountability holder nor the role of accountability holdee will
be fulfilled through network governance. This is certainly not
an uncommon critique: ‘buck-passing’, a lack of clear responsibility,
decision-makers wihtout accountability holders and so forth are all
known critiques in relation to network governance.
However, acknowledging that there is in fact a worst-case scenario
does not imply dismissing network governance altogether. Turning
instead to the issue of what is required of network governance from the
perspective of democratic accountability, the basic rule involved is that
networks ought to first and foremost be considered representative forums of
accountability holdees. Whereas participatory strands of thinking tend to
associate the democratic potential of networks with the capacity to
include a large number (and potentially all) relevant stakeholders,
thinking about network governance in terms of accountability instead
requires that we recognize networks more clearly as representative
forums and strengthen their capacity to work as such. In other words,
the relationship between accountability holders and holdees should not
be seen as a relationship inside networks; but rather, as a relation
between accountability holders (stakeholders) outside of the network
and their accountability holdees within the network.
The organizational features of the network structure render networks
a poor choice for sustaining accountability internally, which is to say
that networks should not stretch across both sides of the distinction
between accountability holder and holdee, but remain on the latter side.
In other words, networks should refrain from attempting to include
accountability holders. Making network governance an asset for democ-
racy from the viewpoint of accountability thus implies acknowledging
that network members should always be considered accountability
holdees and never accountability holders. As simple as this may seem,
the rule is nonetheless in conflict with the common perception inspired
by more participatory strands of thought that networks can and should
include the accountability holders themselves.
Building on this premise, network governance can be considered to
have substantial democratic potential from a postliberal perspective.
First, networks are in many ways the organization of choice from the
perspective of theories of functional representation: networks very often

display the characteristics of functionally delimited and partly self-
governing organizations created or self-grown in relation to a specific
policy problem or policy area. Furthermore, networks are flexible in rela-
tion to the scale of the problem or policy in question (local, regional,
national, transnational). The most important democratic value added
by networks is their capacity to bring all of the relevant stakeholders
together in relation to a particular decision or problem. However, fol-
lowing the argument made above, we should take care not to think of
networks as being able to include all stakeholders, but rather, as poten-
tially being able to include representatives of all stakeholders. It follows
that the network should not be considered a unified collective with a
common set of stakeholders, but an array of network members, each
with their own set of stakeholders to whom they are accountable.
Following the same line of argument, networks could also be consid-
ered a democratic asset for identifying the moral constituency men-
tioned by theories of representation through deliberation. However, this
implies conceiving of networks as proper deliberative forums of ‘some
deliberating for all’ according to the rules of proper political communi-
cation. Or in short, the democratic potential of network governance
hinges on the capacity of networks to function as public spheres in
themselves. This capacity is certainly not a given, but from the perspec-
tive of democratic theory of the public sphere that has long since aban-
doned the notion of a single common national public sphere in favour
of several multi-layered and loosely coupled public spheres (Habermas
1996), the case for network governance may appear quite promising.
However, in order to further probe the potential of network governance
in relation to the deliberative perception of accountability, we must turn
to the notion of publicity.
Accountability and publicity
The relation between accountability holders and holdees is a relation of

informed consent. The institutional structures of accountability are
meant to guarantee that the decisions of the governors are always based
on the informed consent of the governed (March & Olsen 1995: 150).
Calling decision-makers to explain their decisions and assume responsi-
bility requires procedures through which the accountability holders
can demand and receive explanations by accountability holdees. Such
procedures can be designated as procedures of publicity, which is to say
procedures of openness, transparency and access to information. In a

more technocratic vein, we may refer to them as procedures of ‘moni-
toring’.
The different democratic strands of thinking remain in agreement on
some of the basics of publicity, such as of documenting decisions when
exercising governmental authority and the laws granting citizens the
possibility of access to these documents. Accurate records and journals
are a fundamental prerequisite of publicity. A vital meaning of account-
ability is ‘keeping account’ of decisions made. Accountability implies
‘… an obligation for keeping accurate records of property, documents, or
funds’ (Shafritz, cited in Behn 2001: 4). Although openness and access is
the minimal standard of publicity, it is certainly not employed without
exception. Any state or public authority retains the right to keep deci-
sions secret or confidential. Rules of secrecy and confidentially are legit-
imized with reference both to the national interest, efficiency of
administration and the privacy of citizens. Most theories of democracy,
however, recognize the need for exceptions to the rule of openness and
access: there is always a trade-off to be considered and no universal rule
to determine the optimal balance between publicity and secrecy (March
& Olsen 1995: 163).
As seen from the perspective of the liberal tradition, informed consent
first and foremost requires sufficient information to choose between
competing politicians at election time. By contrast, the deliberative strand
of thinking takes the notion of publicity to be the core principle of
democracy itself, making the notion of publicity more demanding and
more complex. In general, deliberative democracy takes informed consent
to presuppose, not simply to inform about decisions made, but a suffi-
cient deliberative process within a public sphere. In other words, the
publicity required to produce informed consent is equated with the exis-
tence of a public sphere sustaining a particular kind of political communi-
cation:
The responsibility of the representative is not simply to tell citizens
how she has enacted a mandate they authorized or served their inter-
ests, but as much to persuade them of the rightness of her judgement
(Young 2000: 131).
A proper public sphere not only implies procedures of making decisions
public, but also a particular way of justifying these decisions, which
in some way or another makes reference to universal rationality
or morality in the process of deliberation. The key figure here is
undoubtedly Jürgen Habermas. Habermas has presented variations on
this theme throughout his oeuvre, from his original thesis about ‘The
Transformation of The Public Sphere’ (Habermas 1962) past commu-
nicative action (Habermas 1995–97 [1981]) and discourse ethics
(Habermas 1991) to the current writings about postnationalism (and the
EU) (Habermas 1998). At the core of these writings, and deliberative
democracy more broadly, is an attempt to transform the ideal of scien-
tific enlightenment into practical standards of communication in the
public sphere. Such standards imply that political communication
ought to conform to an ideal of rational reflexivity. Emphasizing the debt
to Kantian moral philosophy, Gutmann & Thompson characterize the
difference between liberal (procedural and constitutional) democracy
and deliberative democracy in the following manner:
… deliberative democracy raises the stakes of democratic accounta-

bility. It requires more than do procedural and constitutional democ-
racy. A deliberative principle of accountability asks representatives to
do more than try to win re-election, and more than to respect consti-
tutional rights. In a deliberative democracy representatives are
expected to justify their actions in moral terms (Gutmann &
Thompson 1996: 129).
From the perspective of theories of representation through deliberation,

accountability requires that accountability holdees (and holders) con-
form to a certain standard of political communication. Theories of
representation through deliberation pose strict conditions on how
explanations are to be made and responsibility assumed, that is, on the
basis of rational reflexivity and moral consistency. These criteria
constituted what may be labelled a discourse of accountability, or, put
differently: accountability as a particular discursive genre or script
within the larger framework of proper political communication. A
discourse of accountability is basically a discourse based on the notion
of choices, causality and (personal) responsibility. Such a discourse
may be difficult to sustain in the face of causal complexity, intercon-
nectedness, historical forces and structural dynamics being outside of
human control (globalization being the most apparent example).
Nonetheless, accountability requires this particular form of discourse.
Second challenge: sufficient publicity
Perhaps the most persistent critique of network governance is that net-

works are closed, inaccessible and dominated by technocratic discourse.
In short, networks are supposedly the antithesis of the procedures of

publicity necessary to constitute the relation between the accountability
holder and accountability holdee as a relation of informed consent.
Proponents of a strong theory of the public sphere have always been
inclined to see decisions made in networks as decisions taken in ‘dark-
ness’ and secrecy, as opposed to the soothing light of public scrutiny.
Not least media researchers and journalists themselves have voiced such
a concern. Although sometimes based on somewhat metaphorical
concept of networks, there is clearly some truth to the proposition that
networks do not always fit smoothly with the procedures of publicity.
The question is to what extent the publicity offered by networks is suf-
ficient to ensure the informed consent of the accountability holders out-
side of the network on the decisions made within the network. As stated
above, the point is not that everyone must be included in the network –
which is the concern of strong notions of inclusion and participation –
but rather, to what extent networks institutionalize procedures of
publicly assuming responsibility and giving explanations according to
given standards of communication in relation to the stakeholders or
moral constituency outside of the network. In the deliberative vein, the
question is to what extent networks can be made to work as deliberative
forums based on an acceptance of a representational structure, which is
to say that some deliberate for all. There are at least three issues to con-
sider here: maintaining records and journals, the role of the media, and
the extent to which networks conform to a discourse of accountability.
1. Keeping strict records and journals of decision-making processes is a

trademark of both political and administrative institutions in most
western democracies. It is one of the cornerstones of the Weberian
bureaucracy and implemented almost without exception in legisla-
tive and judicial bodies; however, the executive is a different matter.
In most cases, the executive is entitled to a great deal of secrecy, even
in a parliamentary democracy where the legislative (and by implica-
tion, the people) can hold the executive accountable by calling the
executive to respond to questions raised. The problem in most tradi-
tional representative bodies, however, is not keeping the records, but
gaining access to them. Laws on access to documents vary a great
deal, and even access to documents that are not classified as secret
might be subject to various informal obstructions, such as delay,
insufficient documentation or deliberate overload. Nonetheless, we
do have the benefit of formal rules on the matter in the case of the
traditional representative bodies.
In the case of networks, a basic problem is that the position of net-

works within the framework of laws on documentation and access is
fuzzy at best. Networks do not enjoy the status of a ‘subject of law’.
However, networks may nevertheless keep sufficient records. First,
many participants are well trained in the conventional procedures of
keeping records of decision-making and will do so almost automati-
cally. Secondly, most networks have some form of secretariat. In
cases where such a secretariat is based in a branch in the public
administration (typically the case in Denmark), the secretariat will
keep documents subject to the same rules of access as any other part
of the administration. There are, however, also cases where there is
no secretariat or where the secretariat is based in the private sector.
Although a private secretariat may keep and grant access to docu-
ments, networks involved in the production of public policy and
service should preferably have a secretariat based within the public
administration. Public secretariats represent a relatively simple way
to bring the procedures of documentation and access up to speed
with the rules of the country in question – leaving aside for now the
issue of whether these are in themselves problematic.
2. Procedures of documentation and access to documents are of course
insufficient according to any notion of publicity. The notion of pub-
licity also involves more direct forms of monitoring. The most direct
form of monitoring is of course presence during the process of delib-
eration itself, as for example being present during parliamentary
debate. Some networks are closed and do not offer this possibility;
others are fairly open. Networks may even be more open to some
form of participation as opposed to simple monitoring, but our con-
cern here is primarily with networks as a representative forum, that
is, a forum of representatives subject to monitoring and sanctions.
Given the fact that decisions made in parliament are often pro forma
acceptance of decisions made in networks, opening up avenues for
direct monitoring is immensely important. Network participants
could doubtlessly do more on this point, but there is also a problem
to consider for which networks cannot be blamed: the fact that direct
monitoring appears to be of little interest.
Direct monitoring has more or less been substituted for mediated
monitoring. The increasing importance of media and journalists in
politics has been lamented for a long time. It was an important part
of Habermas’ original version of the fall of the public sphere and it
continues to carry the blame for both ‘commercialization’ and
‘conversationalization’ (Fairclough 1995). However, for better or for
worse, ‘public’ primarily means ‘in the media’ in contemporary poli-

tics. With respect to the issue of monitoring, the most important
implication is obviously that monitoring is not direct, but monitor-
ing by proxy. Citizens do not monitor directly, but have professional
monitors and auditors – most notably journalists – monitor and audit
for them: ‘The basic story of auditing is everywhere the same.
Auditors are agents of others who cannot be as well informed’ (March &
Olsen 1995: 164). The issue here, however, is not the extent to which
monitoring by proxy is in itself problematic. Rather, the question is,
given the circumstance that ‘mediated politics’ is a fact of current and
future democracies (Bennet & Entman 2001), how network gover-
nance fares under such circumstances.
A fundamental problem in this context is that journalists are still
lagging far behind in coming to terms with the reality of network
governance. Journalists remain predominantly oriented towards tra-
ditional representative – national and local – institutions (Cook).
Thus, we encounter a clash between the working routines and
news criteria of journalists and the complex reality of network gover-
nance. The problem sometimes assumes the form of a self-fulfilling
prophecy: the news criteria and working routines of political journal-
ists define network governance as lacking clear accountability hold-
ees, which clashes with the need for personal responsibility and
clearly defined sinners and thus the lack of accountability holdees
become the reality offered to citizens. Of course, journalists may be
right, as stated earlier. However, the role of journalists is to maintain
the discourse of accountability, even in cases where it is not easily
applied. Furthermore, journalists may be wrong in many cases: even
if networks do in fact consist of members who assume the role of
accountability holdees, institutionalize procedures of documentation
and access and offer reasonable openness, accountability holders
may lack sufficient means of monitoring due to the fact that journal-
ists do not assume their democratic role as the intermediary account-
ability holder.
3. The final issue to be considered here is the discourse of accountabil-
ity. The basic problem here is that many networks are deeply embed-
ded in modes of discourse that do not conform to the discourse of
accountability. Most networks are to some extent expert-networks or
technocratic networks meant to produce policy solutions or services
within specific policy areas with a tendency to talk in terms of func-
tional necessities rather than contingency, choices and personal
responsibility. In other words, the discourses employed by networks
very often tend to be elitist or technocratic. It may be that such dis-

courses are indeed ‘functional’ in relation to the policy problems at
hand, but this remains a matter of efficiency. From the perspective of
democracy, networks need to maintain the construction of accounts –
or the discourse of accountability – in order to be held accountable.
As always, sustaining a particular mode of communication is not
merely a matter of institutional reform: the challenge is to cultivate a
particular form of discourse within the network. Such cultivation
remains one of the most important democratic challenges to network
governance.
Accountability and responsiveness
Accountability implies that accountability holders have some level of

control over the actions of accountability holders (Peters 1995).
Specifying the roles of accountability holdees and holders, institutional-
izing procedures of publicity and maintaining a proper discourse of
accountability are essential, but accountability also requires the possi-
bility of imposing sanctions on the accountability holdees. Thus, March &
Olsen identify two sources of accountability: information (roughly
equating to what I have just called publicity) and sanctions: ‘A pre-
sumption of democratic accountability is that unfavourable informa-
tion leads to the imposition of sanctions on political actors’ (March &
Olsen 1995: 165). In short, the real touchstone of any system of
accountability is the level of responsiveness of the accountability holdees
to the accountability holders:
Accountability is, in this view, a property of institutional structures,

whereas responsiveness is a consequence of interaction within these
structures. Put another way, responsiveness is a measure of how
much accountability an institutional structure permits (Ferejohn
1999: 131).
Sanctions can generally be understood as the means of punishment. Or,

in other words, the relationship between accountability holder and
holdee is also a relationship between punisher and punishee (Behn
2001: 8). Legitimately administering punishment requires a negative
outcome of a comparison between a mandate already given and deci-
sions made by the accountability holder. Temporally speaking, sanc-
tions come after mandates. In terms of the policy process, sanctions
presuppose the existence of mandates, which can be compared with
policy outcome in order to judge whether the imposition of sanctions is

necessary. A positive evaluation does not result in sanctions, but (at least
in principle) a renewal of the mandate already given. A positive evalua-
tion means lack of change and business as usual, whereas a negative
evaluation means more or less radical change for the accountability
holdee:
They recognize that, if someone is holding them accountable, two

things can happen: When they do something good, nothing happens.
But when they screw up, all hell can break loose. Those whom we want
to hold accountable have a clear understanding of what accountability
means: Accountability means punishment (Behn 2001: 3).
Holding decision-makers accountable in order to secure sufficient

responsiveness, in other words, implies two questions: first, what is the
legitimate mandate through which the accountability holder can spec-
ify the desired course of action to the accountability holdee? Mandates
are an institutionalization of the capacity of accountability holders to
provide instructions to accountability holdees. In terms of accountabil-
ity, the importance of mandates lies in the fact that mandates specify
the desired course of action that the accountability holders are going to
hold the accountability holdee accountable for. In other words, man-
dates specify – more or less precisely – the decisions expected of the
holdee. The authorization of the accountability holdee through the
specification of some form of mandate is an integral aspect of any rela-
tionship between accountability holder and holdee.1
The second question is which sanctions – or forms of punishment –
may legitimately be employed according to the principle of democratic
accountability if the mandate given is not followed? The most straight-
forward answer is provided by the liberal tradition: in the case of politi-
cal accountability, sanctions mean being voted out of office. Since
legitimate mandates are regarded as given exclusively through the elec-
toral process, the only legitimate sanction offered is the revocation of
the mandate. In addition to ordinary elections, most electoral systems
employ extraordinary measures of ‘non-confidence votes’, recall and so
forth. In other words, the sanction is the negation of the mandate.
‘Throwing the bastards out’ when dissatisfied with their performance is
the basic form of sanction. Administrative accountability, on the other
hand, is associated with mandates in the form of job descriptions and
sanctions, such as warnings, being demoted, fired and the re-organization
of public administration units.
Not surprisingly, conceiving of mandates only as electoral mandates

and sanctions as the revoking of these mandates is considered overly
limited by the various postliberal positions. These positions generally
share the notion that mandates ought to be much more flexible and
re-negotiable than is possible through elections. Similarly, sanctions
should cover a much wider spectrum of means of corrections than the
crude yes/no offered by elections. A crucial argument in functional
democracy is that functionally delimited associations and organizations
make possible much more specific and flexible mandates based on
shared understandings and knowledge of problems and policy, espe-
cially in cases where such associations and organizations are constituted
on the local level of governance. Mandates can be given through hear-
ings, meetings, specialized channels of discussion and continuously
adjusted and re-negotiated through these very same channels. The issue
of sanctions is less clear, but the primary sanction is probably exit
instead of voice, that is, to withdraw from the association or organiza-
tion in question. Democratic elite theory conceives of sanctions as the
ability of sub-elites and the public to replace the elites at the head of the
functionally delimited domains or organizations. Such means of impos-
ing sanctions do not imply elections, but rather, a sufficient degree of
elite rotation within the functionally delimited spheres of society, or in
other words:
… the public standing in judgment over the elite’s performance and

its ability to weaken, demote or dismiss the elites when it is dissatis-
fied with its performance (Etzioni-Halevy 2001: 167).
Theories of representation through deliberation understand responsive-

ness on the basis of mandates and sanctions as an ongoing adjustment
in the public sphere. Electoral mandates and sanctions are considered a
completely insufficient means of ensuring responsiveness. What is
needed is rather a continuous renewal of the popular mandate through
proper deliberation. No sanctions are required in a perfect public sphere,
as the quality of political communication ensures that the mandate can
be constantly renegotiated and specified. Conversely, one of the more
important sanctions in the public sphere is ‘silencing’, that is, excluding
someone from the process of deliberation. In cases where accountability
holdees do not conform to proper standards of communication, such a
sanction may be considered legitimate.
Third challenge: adequate responsiveness
The liberal conception of political accountability based on electoral

mandates and sanctions, on the one hand, and administrative account-
ability based on orders from the political level and sanctions such as
being demoted of fired on the other, does not fare very well with the
reality of network governance. Network members are seldom politi-
cians. Even when they are, their participation in networks is not covered
by mandates or sanctions. In some instances, the participation of civil
servants in networks may be subject to orders and instructions from the
political level and to more or less clearly specified sanctions. But in
many cases they are not; and in any case, network members include
other actors in addition to than civil servants. In short, networks evade
political accountability and to some extent also administrative account-
ability, according to the liberal doctrine.
Once again, however, the postliberal strands of democratic thinking
may lead to less pessimistic conclusions. The postliberal notion that
democracy actually requires more flexible and less crude mandates and
sanctions than those offered by elections and conventional means of
politically controlling the administration may in fact point to networks
as an important supplement to conventional representative institutions.
However, our core question remains how to create and maintain the
network as a representative forum in itself. According to postliberal
thinking, democracy requires mandates that can be continuously
revised and re-negotiated, as well as less crude sanctions. At least two
themes must be considered in this regard: the plurality of possible man-
dates and sanctions and the mechanisms through which mandates and
sanctions can be imposed.
1. Following the proposition that the network ought to be considered a

forum of accountability holdees, accountable for their decisions to
different and perhaps partially overlapping sets of stakeholders or
moral constituencies, the issue of adequate responsiveness through
mandates and sanctions is primarily a question of the relation of each
network member to his or her own set of stakeholders. In other
words, there is no single type of mandate or sanctions that applies to
a network. The network should not be considered a collective identity
subject to the same mandate, but a forum for coordination between
actors each subject to their own mandates and sanctions, depending
on the set of stakeholders or the moral constituency that they
represent: ‘Deliberative democracy does not specify a single form of

representation. It searches for modes of representation that support
the give-and-take of serious and sustained moral argument’
(Gutmann & Thompson 1996: 131).
The important challenge, however, is to ensure that each network
member is in fact subject to clear mandates and sanctions from their
relevant stakeholders. This is no easy task. Many networks are consti-
tuted through contacts with ‘usual suspects’, meaning the usual
organizations – if not even the usual persons – within these organiza-
tions, without any further consideration as to whether they are actu-
ally proper representatives, that is, subject to mandates and
sanctions. The degree to which network members are in fact the
proper representatives – sufficiently responsible to a clearly defined
set of stakeholders – must be made an explicit condition of network
formation and durability. This is of course primarily an issue of the
initial selection of network members and possibilities of excluding
and including new members during the lifetime of the network. Such
considerations would also render it possible to map whether a net-
work constitutes a suitable set of mandates in relation to the policy
problem in question.
2. The next question to be considered is of course what can be consid-
ered legitimate mandates and sanctions in network governance. Most
network members are subject to organizational mandates, that is,
mandates from the branches of administration, organizations or
social movements that they represent. If the network is to remain
operative, the question of how to specify proper mandates must be
left to the organizations in question. Participation in network gover-
nance thus requires an effort by the organizations themselves. What is
required is more a question of ensuring that all of the relevant stake-
holders enjoy some form of representation. In this respect, stakehold-
ers who are not very well organized or not organized at all ought to be
of particular concern. Networks have a tendency to favour strong
organizational representatives for reasons of pragmatism and effi-
ciency, but for network governance to be democratic in terms of
accountability, non-organized stakeholders must be given the oppor-
tunity to specify mandates for appointed representatives. Thus, a con-
dition of democratic network governance will in some instances be
the creation of organizations or loosely coupled organizations that
can serve as a means of including non-organized stakeholders.
As regards sanctions, a possible course of action would be to develop
mechanisms of periodic review of the network as a collective actor
‘from above’ as well as ‘from below’ Democratically speaking, the

value of subjecting networks to review and sanctions from above in
this respect lies within its capacity to extend the already existing
chain of mandates and sanctions from the people to political institu-
tions and on to the network itself. It would be interesting, however,
to consider the issue of how accountability holders could govern net-
works directly, including the possibility of sanctions: of shutting the
network down. Such mechanisms could take the form of periodical
reviews or hearings with the added bonus of being able to vote the
network ‘out of office’.
Conclusion
While basically acknowledging that network governance is in fact

capable of producing results adverse to the norm of accountability, the
basic line of argument in this chapter has been that networks hold a
good deal of potential for strengthening democracy, not in terms of the
sort of participatory or direct democracy that often constitute the pre-
ferred strand of thinking when making a case for the democratic virtues
of democracy. Accountability remains a norm of primary concern for
liberal thought and postliberal thought engaged in the rethinking of
proper representation as the core of democracy. Correspondingly, my dis-
cussion of network governance in terms of accountability has emphasized
the need to think of networks as representative forums rather than
means of participation and direct democracy. My point in this respect is
certainly not that the participation and direct democracy are irrelevant
and thus to advocate a sort of ‘the end of history’-thesis. Alternatives to
liberalism remain relevant in general as well as in relation to the debate
about network governance. However, setting aside pragmatism and
the division of labour between this and other contributions in this vol-
ume (see chapters by Hansen 14 and Dreyzek 15), it is my contention
that the concept of the network as a representative forum subject to the
standards of accountability must play a more vital role in future
thought regarding the democratic potential and problems of network
governance.
Furthermore, it seems important to stress that increasing the represen-
tative capacity of networks need not be at the cost of other democratic
potentials. Democratic norms and standards, which are often perceived
as being in conflict or at least a matter of grave either/or-choices at the
level of normative theory and debate, are often combined in ways that
are less orderly and even more creative at the level of concrete decision-
making and governance. While not exactly keeping the purity and strict-
ness of coherent systems of thought and theoretical traditions, such
pragmatic mixing of democratic traditions in concrete decision-making
processes and institutions may often produce highly interesting and
legitimate results. Thus, rather than taking an increase in the representa-
tive potential of network governance to mean the end of other demo-
cratic potentials of networks, one should rather look for ways in which
increased attention to representation can be combined with democratic
norms and standards in the concrete practice of network governance,
Note
1. Mandates and responsiveness are sometimes regarded as an alternative
to accountability and sanctions. (Manin et al Przeworski & Stokes 1999a).
Such reasoning is in no small part inspired by Hanna Pitkin’s distinction
between authorization called the ‘mandate conception of representation’
and ‘accountability conception’ based on sanctions (Pitkin 1967, see also
Stokes 1999). As emphasized by Iris Young, however, Pitkin herself notes that
the idea that these modes of representation constitute distinct systems of rule
in themselves is misconstrued (Young 2000: 128).
The Second Generation of
Governance Network
Theory and Beyond
This book has aimed to deliver theoretically informed answers to new

and pressing questions about the dynamics of governance networks, the
conditions for their success and failure, the forms and functions of
metagovernance, and the democratic problems and potentials of
network governance. In this concluding chapter we aim to assess the
answers that the authors have given to these questions. We pinpoint
and discuss their theoretical points of departure, summarize and
compare their answers, and identify areas where further discussion and
research needs to be done in the future. Our assessment of the theoreti-
cally informed answers to the questions raised by the new second
generation of governance network theory will be followed by specula-
tions about where this kind of research might go next. However, first we
briefly restate the argument about the importance of focussing on
interactive governance through networks.
Governance networks are here to stay
Today governance networks proliferate to an astonishing extent within

different policy areas and levels of governance. The institutional forms,
the patterns of participation and the supporting rhetoric might differ
from country to country or from level to level, but the core feature in
terms of governance through negotiated interaction among a plurality
of relevant and affected policy actors tends to remain the same. The
political and societal background for the recent surge of network gover-
nance is that public governance is confronted by a huge dilemma which
an increasing reliance on network-based governance might help to
solve. On the one hand, there are high and ever increasing demands
that public policy and governance must be proactive, flexible, strategic,
297
targeted, knowledge-based and responsive. On the other hand, the

societal systems that are supposed to be governed according to these
high demands are becoming increasingly complex, fragmented,
multi-layered and dynamic. There has always been a discrepancy
between the ambitions and capacities of governance, but the gulf is
widening and becoming more and more visible. In this situation, the
mobilization of the knowledge, energy and resources of a plurality of
public and private actors within different kinds of governance networks
seems to offer promising way of realizing the growing ambitions of
modern governance in an increasingly differentiated polity. At least, this
explains why central decision-makers at the local, national and transna-
tional levels increasingly conceive governance networks as an effective
and legitimate mode of governance and an important supplement to
traditional forms of governance in terms of hierarchy and market. The
central decision makers’ celebration of the new forms of network gover-
nance is spurred by political pressures from an increasing number of
professional interest organizations, politically engaged citizen groups
and constructive and problem-driven ‘everyday makers’ who want to be
able to influence the output and outcomes of the political and adminis-
trative system in and through active participation in the processes of
agenda setting, policy formulation and policy implementation. The
pressures from ‘below’ are further sustained by pressures from ‘above’
where international and transnational organizations like the World Bank
and the European Union advocate the inclusion of stakeholders in pub-
lic policy making as a part of their quest for ‘good governance’. Last but
not least, new information and communication technologies facilitate a
rapid and easy identification of stakeholders, exchange of knowledge
and organization of meetings.
The increasing reliance on governance networks does not mean that
the use of traditional forms of governance in terms of hierarchy and mar-
ket is diminishing. On the contrary, there are many examples of how
hierarchical governance by politicians and executive administrators is
reinvigorated by the adoption of new steering and control techniques in
terms of management by objectives, performance measurement and
bench marking. Likewise, privatization, contracting out and the intro-
duction of quasi-markets within the public sector are still on the top of
the political agenda in most Western countries. But in many cases, the
proliferation of governance networks seems to run parallel to the rein-
forcement of hierarchy and market. In fact, governance networks often
help to sustain the new forms of hierarchical governance by providing
valuable inputs to the formulation of the overall policy objectives and
by coordinating the actor’s attempt to reach their targets. Governance
The Second Generation of Governance Network Theory and Beyond 299
networks also help to counter the competitive logic of the market by the
generation of trust and mutual learning. However, governance networks
are not merely an appendix to hierarchy and market, they also provide
an alternative mode of governance that might be preferred in cases
where the policy problems, objectives and solutions are unclear and
ambiguous and where there are many stakeholders with potentially
conflicting interests.
No matter whether governance networks are supplementing or replac-
ing the traditional forms of governance, they tend to be constrained by
the operation of the market forces and to be influenced by the hierar-
chical forms of government. Hence, the expansion of private markets
and public quasi-markets tend to create a ‘no-go area’ for governance
networks as the political negotiation of goals and solutions among inter-
dependent network actors easily come into conflict with the autonomy
of the independent market operators. At the same time, governance
networks are operating in the ‘shadow of hierarchy’ as the formation,
functioning and development of governance networks are shaped and
reshaped by the strategic interventions of government officials at different
levels (Scharpf, 1994).
At present, it seems that governance networks are here to stay. It is
becoming more and more difficult to govern society from a privileged
centre and interactive coordination of objectives, funding and policy
actions among a host of relevant and affected policy actors seem to be a
prerequisite for effective governance. Therefore, it is increasingly impor-
tant to address the questions raised by the second generation of gover-
nance network research. It is no longer enough to explain why
governance networks are formed, to show how they differ from hierarchy
and market, and to account for their contribution to effective and proac-
tive governance in different countries and policy fields. We must learn to
live with governance networks and develop a critical and theoretically
informed understanding of their dynamic development, the conditions
for their success and failure, the different forms of metagovernance, and
how to assess and improve their democratic performance. Empirical
studies based on standard and non-standard methods are also important,
but these should be based on a theoretical understanding of the key
questions of the second generation of governance network research.
A multi-theoretical approach to network governance
The different theories of governance networks that have been developed

from the early 1970s onwards provide the basic theoretical framework
for answering our four basic research questions. The answers to the first
question about the formation, functioning and development of

governance networks also draw on the new developments within insti-
tutional theory and the answers to the last question about the
democratic problems and potentials of governance networks primarily
draw on various forms of democratic theory. In order to emphasize the
analytical diversity of governance network theory, institutional theory
and democratic theory, we have aimed to map the rival approaches
within the three theoretical fields according to their paradigmatic
conceptions of social action and the societal systems of governance.
Hence, the rival theories and approaches within each of the three
theoretical fields have been divided into a four fold table that combines
calculation based and culture based views of social action with conflict
based and consensus based conceptions of societal governance. We are
perfectly aware that ‘hell is the fury of a misplaced specimen’, but we
believe that a heuristic mapping of different approaches and theories
will help to keep the theoretical fields open and to provide different,
complementary and even conflicting answers to the four basic research
questions raised by the second generation of governance network
research.
The authors, who have contributed to this volume, have been asked to
clarify their theoretical point of departure for answering the questions
that we had posed to them. However, they have been free either to
associate themselves with one of the theoretical positions in the
four-fold tables, to combine them in any sort of way, or even to
transcend the analytical distinctions altogether by drawing on other the-
oretical resources that are not easily captured by the four fold tables that
aim to map the theoretical fields invoked in this volume. Not all the
authors agreed with our way of mapping the three theoretical fields, but
they ended up positioning themselves in relation to the three
theoretical maps.
The authors, whose chapters appear in Part I, have been struggling to
answer the fundamental question about the dynamics of governance
networks. How are governance networks formed? How are they are
functioning? How do they develop and change over time? Niels
Hertting’s argument draws on game theory to explain the formation of
governance networks. However, he claims that rational choice institu-
tionalism should be supplemented by other institutionalist approaches
that are better suited to explaining the struggles and conflicts that
determine the actors’ perception of their mutual dependencies. Guy Peters
initially discusses the contributions of historical institutionalism and
rational choice institutionalism to understanding the institutionalization
and deinstitutionalization of governance networks, but his main argu-

ment builds on an analytical model based on sociological organization
theory and normative institutionalism. Finally, Mark Bevir and
R.A.W. Rhodes aim to develop a decentred theory of the change of
governance networks in response to success and failure. Their theory
emphasizes the social construction of network relations by situated
social actors whose beliefs and actions are shaped by traditions and story
telling. The hermeneutic aspect of their theoretical account of network
change is in line with normative institutionalism and their emphasis on
decentred structures and actors involved in discursive story telling
brings them close to poststructuralist institutionalism. However, their
strong emphasis on the radical contingency of social practices and the
absence of a theory of discursive power struggles makes it difficult to
assimilate them into these theoretical positions. This is hardly surprising
as they deliberately seek to trancend the analytical framework in the
search for a pragmatist theory inspired by the post-analytical philoso-
phy of Wittgenstein.
The authors in Part II have aimed to answer the crucial question about
the conditions for success and failure of governance networks. Linze
Schaap deals with the dilemma between openness and closure and works
his way from interdependency theory via governability theory to a more
rigorous system theoretical position inspired by the works of Luhmann.
The latter clearly trancends our initial mapping of governance network
theory. Joop Koppenjan focuses on the dilemma between consensus and
conflict and his analysis mainly draws on interdependency theory
although there are also references to key insights from governability
theory. Finally, Tanja Börzel and Diana Panke’s discussion of the trade-off
between the effectiveness and legitimacy of governance networks is
firmly based on governability theory.
The authors contributing to Part III discuss how different aspects of
network-based policy processes can be metagoverned. Peter Triantafillou
focuses on how governance networks are formed through the construc-
tion of autonomous and interdependent actors and clearly bases his
argument on Foucault’s theory of governmentality. Erik-Hans Klijn and
Jurian Edelenbos analyse the metagovernance of governance networks in
terms of process management and institutional design and their analysis
is firmly grounded in interdependency theory. Finally, Laurence O’Toole
draws on governability theory in his discussion of the metagoverning of
network-based policy outputs and policy outcomes.
The authors in Part IV have aimed to assess the democratic problems
and potentials of network governance. Allan Dreyer Hansen takes issue
with the communitarian variants of participatory democracy and

discusses the question of democratic participation in governance
networks on the basis of theories of agonistic democracy. John Dryzek
focuses on the democratic demands for equality and freedom in relation
to network-based deliberation processes. He emphasizes the commu-
nicative aspect of democracy and develops a theory of discursive
democracy which is founded on a critical engagement with a Habermasian
version of communitarian democracy. Finally, Anders Esmark discusses the
democratic accountability of governance networks on the basis of theories
of competitive democracy and a ‘thin’ version of communitarian democ-
racy that emphasizes the importance of public deliberation.
On the whole, the authors have located themselves within the
theoretical terrain that has been mapped by the three four-fold tables.
Most of the authors associate themselves with one of the theoretical
positions (Peters, Koppenjan, Börzel and Panke, Triantafillou, Klijn and
Edelenbos, O’Toole, Dryzek). There are a few examples of cross-fertilization
(Hertting, Hansen, Esmark), but only two cases where the authors delib-
erately attempt to trancend the theoretical positions of the four-fold
tables and the analytical distinctions on which they build (Bevir and
Rhodes, Schaap). The fact that most of the contributions fall within the
theoretical territory mapped by the three four-fold tables does not in
any way prove the authority or validity of our theoretical mapping and
the analytical distinctions on which it builds. The authors were asked to
position themselves in relation to these particular theoretical landmarks
and they might have had great difficulties in doing so. However, at least,
the four-fold tables have proved their value as heuristic tools for
distinguishing between different theoretical approaches. Other ways of
cutting the cake can easily be imagined, but this way of doing it helps to
spot important theoretical differences, combine different approaches,
and trancend the established positions in search for new ones.
When looking at how the authors place themselves in the theoretical
landscape, it is clear that the theoretical positions are not equally
represented in the contributions to this volume. We have selected the
authors both in order to cover particular research questions and in order to
represent particular theoretical approaches, but with a small number of
authors it has been impossible to ensure an even distribution of the
contributions in terms of their theoretical approach. Hence, it comes as
no surprise that the most well-established theories and approaches are
better represented than the novel ones. Another observation is that the
theoretical points of departure of the different authors are closely linked
to the theoretical problem that they are addressing. This might be seen
as an evidence of the problem-driven research approach that seems to

dominate in the study of policy, governance and democracy.
Contributions to our understanding of

governance networks
It is time to assess the answers provided by the contributing authors.

The four parts of the book have addressed crucial questions about: 1. the
dynamics of governance networks; 2. the reasons for governance network
failure; 3. how to metagovern governance networks; and 4. the demo-
cratic problems and potentials of governance networks. Each of the four
parts consisted of four chapters: a review chapter presenting different
theoretical approaches to the issue at hand, and three thematic chapters
attacking the particular research question from different angles. In this
section, we summarize the contributions of the thematic chapters, and
identify loose ends and issues that must be dealt with by the second
generation of governance network theory.
Understanding network dynamics

The first part of the book takes on the difficult task of producing new
theoretical insights into the dynamics of governance networks. In
Chapter 2 Nils Hertting shows how governance networks differ from
other institutional settings by not being formed once and for all, but
again and again. The particular game structures inherent in governance
networks give rise to free riding problems, assurance problems and
generosity problems, and for that reason governance networks are per-
manently on the brink of falling apart. Accordingly, their survival
demands constant efforts to stabilize them in and though ongoing
attempts to ensure policy coordination. This emphasis on the vulnera-
bility and instability of governance networks increases our understanding
of the process of network formation, but it also calls for further reflection
on the reasons why actors choose to form networks in the face of the var-
ious problems relating to collective action. In addition, it raises a broader
set of questions that need to be addressed by the second generation of
governance network research: How can the sources of instability be
reduced through various forms of institutionalization? How are the
preferences and motivations of the actors shaped by the institutional
conditions and traditions? Is the formation of governance networks
path-dependent?
In Chapter 3 Guy Peters theorizes further on the inherent vulnerabil-
ity and instability of governance networks, and the specific institutional
dynamics that cause it. In order to be effective, he argues, governance

networks must reach some level of stability that is achieved through insti-
tutionalization and deinstitutionalization. Effective network governance
requires the presence of committed network actors (individual as well as
collective) who are capable of making decisions and producing desired
outcomes. It also requires that governance networks are capable of inter-
acting with and relating to outside actors from other networks and/or
institutions. Further considerations on how the actors’ commitment to
interaction, their capacity for joint action and their ability to relate to out-
side actors are initiated and consolidated in and through different institu-
tional structures would be very interesting in this respect. Among other
things, it is central to the second generation of governance network theo-
ries to address the question of how governance networks function in an
institutional context dominated by institutional features associated with
the rule of the state, the market, or civil society.
In Chapter 4 Mark Bevir and R.A.W. Rhodes stress the central role of
beliefs and traditions for the functioning and transformation of gover-
nance networks. They point to the efforts of the individual network
actors to cope with contradictory demands deriving from competing
narratives on how governance is produced as one of the main sources of
network transformation. The recognition of the importance of individ-
ual coping strategies directed towards the balancing of competing
demands seems crucial for our understanding of network transforma-
tion. It does, however, remain somewhat unclear as to what extent
existing beliefs and traditions are to be perceived as a source of such
competing demands, and thus whether an integrated part of network
transformation is changes of such beliefs and traditions. If so, the
uncovering of institutionalized beliefs and traditions is crucial for our
efforts to understand how and why governance networks are being
transformed. This interplay between network transformation and the
transformation of institutionalized beliefs and traditions needs to be
further analysed. However, a better understanding of network transfor-
mation also calls for an analysis of how governance networks and the
narratives that ensure their internal cohesion are affected by changes in
the broader political, cultural and socioeconomic context. While trans-
formations of networks might not be easy to impose from the outside,
contextual changes are likely to produce dilemmas that, whether
intended or not, transform governance networks in one way or another.
Reasons for governance network failure

Part II has aimed to enhance our theoretical understanding of the
reasons for governance network failure and the conditions for success.
In Chapter 6, Linze Schaap considers the extent to which network clo-

sure is likely to cause network failure. He identifies different kinds of
social and cognitive closure on the part of the actors or the network as a
whole and he explains these different forms of closure as a result of:
1. the accommodation of the veto powers of the interdependent
network actors; 2. the filtering effects of the common frame of reference;
and 3. the networks’ imbeddedness in a particular policy specific com-
munication system. While Schaap provides important insights into the
impact of different forms of network closure, he does not touch upon
the closure invoked by different forms of self-exclusion, whereby rele-
vant and affected policy actors either construct the governance network
as an enemy, or believe that participation in it will somehow compromise
their identity. This issue needs further attention by the second generation
of network theorists, and opens a more general debate about the rela-
tionship between exclusion, network closure and governance failure.
Will exclusion necessarily produce closure? Will closure necessarily
result in governance network failure, or is some degree of closure
necessary in order to govern successfully?
In Chapter 7 Joop Koppenjan argues that a certain balance between
conflict and consensus is needed in order for governance networks to
function properly. Conflicts do not necessarily produce failure.
Of course, the presence of long, strong and irreconcilable conflicts can
be damaging, but the absence of conflicts is equally problematic. In fact,
a certain level of conflict will help to prevent closure, enhance plurality,
promote transparency, mobilize actors, and provide new information,
higher quality and more innovation. However, as Koppenjan argues, a
certain degree of consensus is also important in order to avoid network
failure as consensus helps to provide certainty, establish enduring rela-
tions, simplify problem solving and reduce transaction costs. While the
recognition of the importance of balancing conflict and consensus is
important for understanding why networks sometimes fail, it remains
somewhat unclear how consensus and conflict are related to each other,
and how the two concepts are defined. Is consensus the absence of con-
flict, or is consensus a result of conflicts and power struggles and itself
conditioned on the exclusion of a constitutive outside of enemy forces?
Such considerations call for a general debate about the nature of politics,
negotiated governance and conflict resolution.
In Chapter 8 Tanja Börzel and Diana Panke analyse the relationship
between effectiveness and legitimacy. They argue that the relationship
between the two is not necessarily, as it is often argued, that of a trade
off. In fact, governance networks might help to transform the relation
between effectiveness and legitimacy into a plus-sum game where the
two crucial objectives compliment and reinforce each other. However,

the plus-sum game between effectiveness and legitimacy is not ensured
merely by means of enhancing the capacity for effective problem
solving through the inclusion of active, knowledgeable and responsible
stakeholders. It is not enough to increase output legitimacy through the
production of more effective policy outcomes. Governance networks
should also enhance democratic input legitimacy by increasing participa-
tion, deliberation and accountability. This intriguing claim for the need
to increase the input legitimacy of governance networks raises the
important question, which is only marginally considered by Börzel and
Panke, about how governance networks are to be held accountable
through public scrutiny and political sanctions. In addition, we need to
study how the demand for increased democratic legitimacy can be
realized in different institutional settings and at different levels of
governance.
Metagoverning governance networks

The third part of the book has focussed on the question of how to
metagovern governance networks. In Chapter 10 Peter Triantafillou looks
at how governance networks are formed and mobilized through metagov-
ernance. He shows that the metagovernance of governance networks is
not, as most governance network theorists tend to argue, first and fore-
most a matter of affecting the structures of interdependency between
rational, self-interested network actors. Metagovernance is primarily exer-
cised through a combination of technologies of agency that shape the
identities, norms and interests of the network actors and technologies of
performance that regulate the calculations and conduct of the network
actors. By focussing on the institutional devices, methods and technolo-
gies of metagovernance, rather than on the exercise of political authority
over the network, it becomes possible to see the relation between power
and freedom as a positive-sum rather than a zero-sum game. The empha-
sis on institutional technologies that shape the identity of the network
actors and construct the normative framework for their interaction pro-
vides a new insight into the practice of metagovernance in advanced lib-
eral societies. However, it remains a little unclear who the metagovernors
are, how their identity as metagovernors is shaped and what guides the
regulatory practices. The technologies of metagovernance are a part of a
certain governmentality that tends to change over time, but it is not clear
what drives the historical changes.
In Chapter 11 Erik-Hans Klijn and Julian Edelenbos ask how gover-
nance networks can be metagoverned through the exercise of network
management that combines process management and institutional

design. Process management can be exercised through the activation of
particular actors, the creation of organizational arrangements, the guid-
ance of network interactions, formation of goal-achieving strategies,
joint knowledge production, and the creation of trust. Institutional
design is carried out through transformations of the game structures
that regulate the patterns of interaction between the network actors.
Klijn and Edelenbos argue that the complexity of network management
means that the designated network managers must possess a broad
range of skills including analytical skills, substantive knowledge, design
competences, capacities to assess, choose and implement workable net-
work strategies, and the ability to access network conditions. The
authors make an important contribution to understanding how
complex networks can be managed, but we need further reflection on
the effect of network management on governance networks. In particu-
lar, we need further discussion of how governance networks can be
metagoverned without ruining their capacity for self-regulation. How
can process management and institutional design help to facilitate
self-regulation and how do we avoid network management resulting in
the disempowerment of the network actors?
In Chapter 12, Laurence O’Toole follows up on the question of how to
regulate governance networks in order to improve their functioning. He
focuses on the interaction between public authorities and governance
networks in order to investigate how public authorities can shape the
outputs and outcomes of governance networks through various forms of
metagovernance. O’Toole shows how public authorities are neither in
total control nor completely impotent vis-à-vis governance networks.
The specific status, expectations and formal authority of government
agencies permit them to operate at a meta-level of governance con-
cerned with policy formation and constitutional regulation and not just
at the operational level of concrete governance processes. That allows
them to influence the games that structure the interaction within
network governance. This identification of the specific capacity that
public authorities have for exercising metagovernance is important and
convincing, but further studies are needed in order to assess how the
capacity to exercise metagovernance depends on the location of the
public authorities at particular levels of governance (local, national,
transnational) and on the specific character of the governance network
and the actors it inhabits. Do local authorities have the same capacity to
govern networks as national and transnational authorities, and are
public authorities capable of metagoverning governance networks
consisting of strong business corporations in the same ways and to the

same degree as governance networks consisting of local citizens?
Democratic problems and potentials

The fourth and final part of the book has dealt with the crucial question
about the democratic implications of governance networks. In Chapter 14
Allan Dreyer Hansen looks at the question of the democratic problems
and potentials of governance networks through the lens of a reformu-
lated theory of participatory democracy. He concludes that governance
networks are likely to have positive as well as negative effects on demo-
cracy. On the one hand, governance networks might strengthen demo-
cracy through the facilitation of democratic learning and empowerment
of the involved actors and through the promotion of responsive delib-
eration based on agonistic respect. On the other hand, these democratic
potentials are likely to be countered by democratic problems in terms of
unequal patterns of participation and limited publicity and trans-
parency. Accordingly, we need to consider what democratic norms
count the most in order to be able to assess the aggregated effects of
governance networks on democracy. Hansen’s identification of the
ambiguous democratic effects of governance networks is convincing,
but it does seem rather unclear why governance networks necessarily
provide a more unequal access, and are more focussed on particularistic
interests than territorial forms of participation and governance such as
city councils, neighbourhood councils, etc. A clarification of these issues
calls for a general discussion about the relationship between territorially
and functionally demarcated forms of governance, and for an analysis of
the way these forms of governance regulate democratic participation.
In Chapter 15, John Dryzek discusses how governance networks can
possibly realize the democratic ideals of political equality, popular
control and political freedom. The question he raises is whether these
democratic ideals, which constitute the backbone of the territorially
demarcated forms of liberal parliamentary democracy, can be redeemed
in the context of the much more issue-specific, informal and sometimes
transnational governance networks. Dryzek claims that it is possible to
apply the demand for political equality, popular control and political
freedom to governance networks, but it requires a rethinking of the
content of these values, and of the liberal theories of democracy as such,
along the lines of a discursive theory of democracy. Seen from the
perspective of discursive democracy, political equality must be reinter-
preted as a demand for inclusion of all those who are affected by the
decisions of a particular governance network. Popular control must be
reinterpreted as a question of the scope of issues that the affected should

be able to control and as a question of the how much control the
affected should have with a particular issue. Finally, political freedom
must be understood as access to contestation through all the available
forms of communication. The claim for the necessity to reinterpret
liberal theories of democracy and its basic norms is compelling, but it
raises the central question as to how the reinterpreted democratic values
are to be institutionalized: How do we regulate the processes that lead to
defining who should be considered as affected? How should we decide
who is to control what? How can we ensure that this control is substan-
tial and not merely symbolic, and how can we maintain the highest
possible level of contestation? Governance networks might contribute
to promoting norms associated with a discursive democracy, but it is
important to consider how governance networks can be institutional-
ized and articulated with other institutionalized forms of democracy.
In Chapter 16, Anders Esmark addresses the pressing question about
the democratic accountability of governance networks. Starting from a
post-liberal democratic line of thinking he focuses on three central
aspects of accountability: inclusion, publicity and responsiveness.
Inspired by functionalist theories of representation Esmark claims that
the members of governance networks should not be seen as accounta-
bility holders, but rather as accountability holdees. Hence, a governance
network should be seen as a representative forum that must seek
accountability among a broader environment of affected stakeholders.
Seen from this perspective, he argues that governance networks have the
potential to enhance democracy. The enhancement of democracy is not,
as some might argue, a result of the introduction of some kind of direct
participation. Rather, it is the result of the introduction of a new form of
functional representation that contributes to ensuring the democratic
accountability of the political system as a whole by supplementing the
democratic accountability produced by parliamentary democracy. The
claim that network participants should first and foremost be seen as
accountability holdees points to the fact that governance networks in
most cases consist of political sub-elites and, therefore, should not be
interpreted and evaluated as a form of direct citizen participation.
However, it remains somewhat unclear whether it is possible to distin-
guish between the role of governance networks as an accountability
holdee in relation to the stakeholders and as an accountability holdee in
relation to the political authorities or other governance networks. This
issue needs further consideration by the second generation of gover-
nance network research. But even more importantly, the new insights
into the democratic accountability of governance networks calls for a

general discussion about how a democratic relationship between gover-
nance networks as accountability holdees and the broader set of affected
stakeholders is to be institutionalized and maintained.
Where next?
We hope that this book will help to renew and enlarge the research
agenda in the field of governance network theory. Hence, to the questions
raised by the first generation of governance network research we have
added some new and intriguing questions that aim to refocus the schol-
arly debate about the new interactive forms of negotiated governance.
The authors have through their excellent contributions provided some
initial answers to the questions we have raised, but there is some way to
go before we have fully explored the terrain that has been opened by the
new questions and answers, and there are crucial challenges ahead.
The theoretical challenge is to find new ways of sharpening reformu-
lating, comparing, combining and even transgressing the theoretical
approaches discussed in this volume and the analytical distinctions on
which they build. The theoretical approaches all have particular
strengths and weaknesses in answering the questions of the second
generation of governance network research. Therefore, we must find
ways of exploiting the advantages of the different approaches, while
eliminating, or compensating for, their inherent limitations. This must
be done through a reformulation of the competing theoretical frame-
works or the construction of new frameworks out of the old ones. It is
tempting to try to incorporate arguments from other theories into a
particular promising theoretical framework, or to try to fuse different
theories into a new framework, but in both cases it is important to avoid
an eclectic combination of theoretical arguments that are informed by
different analytical assumptions and ontologies.
At the empirical level, we need individual and comparative case
studies based on clear operationalizations of the theoretical concepts in
order to refine the theoretical concepts and arguments and test how the
extent to which they can capture the empirical richness of the concrete
networked policy processes. As such, we need empirical studies that aim
to uncover the institutional conditions for the formation and transfor-
mation of governance networks, the context dependent dilemmas
conditioning their functioning, the deployment and effects of different
forms of metagovernance, and the extent to which governance
networks are democratically anchored (Sørensen and Torfing, 2005a).
The methodological challenge is to expand the methodological bag of

tools to include other than the standard qualitative social science
methods in terms of research interviews and document studies. Other
non-standard methods such as focus group interviews, observation
studies, diary writing and social network analysis can help us to reveal
the form and character of the formal and informal relations between
relevant and affected policy actors. The application of a broader range of
social scientific methods will help to provide a richer data base that will
facilitate the generation of plausible research results on the basis of tri-
angulation. Another methodological challenge is to develop a more
interactionist research practice based on a continued dialogue with the
knowledgeable and reflexive policy actors inhabiting the empirical field
that we are studying (Zouridis, 2003). Interaction with the actors who are
involved in particular governance networks can help us to sharpen our
research questions by focussing on relevant policy problems, identify
and gain access important data sources and qualify our interpretation of
data and test the plausibility of our conclusions.
The research agenda in the field of governance network theory will be
transformed and expanded several times in the years to come, both in
response to new theoretical impulses and re-orientations and in
response to empirical developments in a politico-administrative envi-
ronment where the institutionalized forms of network governance are
constantly problematized by political, administrative and lay actors.
A couple of interesting questions are already calling for our attention.
One question relates to the conceptions, reasons and rationales informing
the political choice between different combinations of hierarchy,
market and networks in relation to particular policy problems. Another
question concerns the positive and negative effects that hierarchy and
market may have on governance networks and vice versa. A third
question is about the impact of political cultures and traditions on the
form and functioning of governance networks in different countries and
at different levels. A fourth question concerns the role of governance
networks in linking actors from different countries, levels and spheres in
transnational, multi-level and multi-cultural networks in which the
rules and norms guiding negotiated policy interaction are far from clear
in the outset. Several other important questions could be mentioned,
but only time will tell whether any of them will make their way to the
top of the research agenda.
However, the future of governance network research should not be
allowed to drift ahead without deliberate and reasoned attempts to
point the large community of governance researchers in new directions.
Several contributions to this book seem to suggest that the future

analysis of governance networks will benefit from a more careful
analysis of different kinds of decentred structures. As such, we need to
focus more on: the discursive structures that help unify different and
conflicting actors around common objectives and joint policy making;
the structures of interaction that assign and distribute different roles and
functions to the network actors; and the power structures that define the
political conflicts and cleavages between the network actors and
determine the output and outcome of their negotiated interaction.
Network analysis often has great difficulties accounting for the
internal cohesion and external boundaries of governance networks,
conceived as loosely-defined coalitions of policy actors (Knox et al.,
2006: 121). The network actors might have conflicting preferences and
interests, might come from different cultures and might use different
linguistic and behavioural codes, but still engage in stable policy negoti-
ations with each other. This is commonly explained by the actors’
mutual need for policy coordination and exchange of resources. But
what is it that should be coordinated? How is the object of coordination
and joint regulation defined and dealt with by the actors? Network-
based policy processes dealing with a particular policy problem rarely
include all the relevant and affected actors. We might, therefore, ask
how the in-group and out-group are defined and how the inclusion and
exclusion of policy actors is regulated. In order to answer these pertinent
questions, the analysis of governance networks will benefit from a closer
analysis of the emerging discursive structures and story lines that construct
a hegemonic conception of policy problems, political values and feasible
solutions within a particular governance network and define the limits
and boundaries of the network by positing a threatening other of
enemies and policy disasters that, at once, negate and stabilize the
policy discourse of the governance network (Hajer, 1993, 1995;
Howarth, 2000; Torfing, 1999).
Governance networks are held together by the policy discourse that
the networks actors enact and re-enact in the course of interaction. The
structuring of the policy discourse around tendentially empty signifiers
like ‘modernization’, ‘efficiency’, and ‘social justice’ permits a large
number of policy actors to subscribe themselves into the networked
policy making, but as the dominant actors succeed to define the content
of the empty signifiers exclusions are inevitable. The policy discourse
defines a common room for negotiation and the actors will play differ-
ent roles and have different functions in these negotiations. Despite the
seemingly horizontal articulation of policy actors within governance
networks and their mutual dependency on each other’s resources, there

are often huge differences among the network actors in terms of
the intensity, direction and density of their contacts and in terms of
their position in relation to different actors and cliques. A careful analy-
sis of the structures of interaction will help to establish the overall shape
of the network and the character and centrality of the different network
actors. This will in turn facilitate the determination of the different bro-
kerage roles of the network actors who might perform important roles as
coordinators, gatekeepers, representatives, etc. The quantitative tech-
niques associated with Social Network Analysis (Scott, 1991) are helpful in
analysing the interaction structures within governance networks, but since
this type of analysis is blind to the content of the network interactions, it
should be firmly anchored in political and social anthropological
network theory (Knox et al., 2006).
The analysis of the structures of interaction makes it possible to under-
stand the different positions and roles of the network actors. The struc-
tural position of the network actors is an important source of influence
for the individual actors and it is often sustained by key resources in terms
of formal authority, knowledge, money, organizational capacities, etc.
However, the distribution of resources among the network actors does
not tell the full story about the power relations within governance
networks. Of course, the deployment of different kinds of resources are
important for the exercise of power, but the value and significance of
the actors’ resources is always contextually determined and, in the last
instance, the exercise of power rests on the power strategies of the actors
and the way these strategies are framed by the discourse that constructs
the cohesion and boundaries of the governance network. The analysis of
the power structures of governance networks involve careful analysis of
how the different network actors succeed to influence the concrete
decision-making, the political agenda of the governance network, and
the commonly accepted values, objectives and beliefs. But the analysis
of the ‘three faces of power’ (Lukes, 1974) must be imbedded in an
analysis of how the network actors draw upon, enact and transform the
policy discourse that supports and constrains the governance network
by invoking a particular set of inclusions and exclusions of meaning,
actions and metaphors (Torfing, 1999).
The emphasis on power, conflict and antagonism will help to develop
a more critical perspective on the hegemonic conception of the modern
governance. Although one should be careful not to generalize, there is a
strong tendency in the literature on governance networks and among
policy makers to deny the political character of network governance by
emphasizing the rational, consensus-oriented deliberation that is

supposed to result in pragmatic problem solving through a plus-sum
game without any losers. The official partnership rhetoric that supports
the formation and development of governance networks at different
levels is often informed by the same belief in a non-adversarial, mana-
gerial policy making. According to Chantal Mouffe (2005), the implicit
idea of a network-based policy making without politics is a part of a dan-
gerous post-political vision that aims to eliminate conflict and antago-
nism in the name of a rationalist consensus. Contrary to the optimistic
belief that the elimination of political conflict and social antagonisms,
and the political passions they are invoking, will lead to a strong and
harmonious democracy, the post-political elimination of strife, passion
and social antagonism from the political and democratic domain will
tend to spur the formation of moral antagonisms based on increasingly
essentialist and fundamentalist values that cannot be subjected to
democratic deliberation and therefore constitute a threat to democracy.
In other words, ‘a healthy democratic process calls for a vibrant clash of
political positions and an open conflict of interests’ (Mouffe, 1993: 6).
Of course, this does not mean that governance networks should be
turned into battlegrounds for irreconcilable conflicts between sworn
enemies. Social antagonisms are an intrinsic part of public governance
and they must find ways of expressing themselves at the level of
democratic policy making, but they must be tamed by the development
of a democratic ethos and a grammar of democratic conduct that turn
‘enemies’ into ‘adversaries’ and replace ‘antagonistic clashes’ with
‘agonistic respect’ based on the recognition of the contingency of all
political interest, beliefs and values.
Our speculations about the future agenda of governance network
research, and the need to focus more explicitly on the structures of
discourse, interaction and power, should not blind us to the fact that
recent trends in the debate on governance tend to turn the study of
governance networks into a subset of a broader research program
focussing on governance through ‘regulated self-regulation’ (Cruikshank,
1999; Dean, 1999; Rose, 1999). According to this argument, govern-
ments at all levels are facing mounting pressures to solve complex policy
problems through flexible, targeted, timely and responsive interven-
tions, while, at the same time, deploying a minimum of political force
and fiscal resources. A way out of this impasse is that the governments
seek to rely more on practices of ‘regulated self-regulation’, whereby
individuals, target groups, private enterprises, public institutions,
governance networks, etc. are called upon to regulate themselves in
accordance with broadly defined goals, norms, values and rationalities.

Today, governance at a distance, through the regulation of self-regulation,
is not restricted to the formation and activation of governance
networks. Increasingly, citizens are expected to regulate themselves in
relation to norms defining a healthy living. Ethnic groups are expected
to organize and regulate themselves in order to enhance their social and
cultural integration into mainstream society. Private firms are hailed by
the discourse of Corporate Social Responsibility to developed new forms
of sheltered employment and to facilitate integration of different groups
of unemployed. Public institutions are called upon to translate general
political objectives into relevant local targets to be reached through the
construction of partnerships with private actors. In short, a whole range
of societal actors are expected to mobilize their energies, resources and
knowledge in practices of self-regulation that are metagoverned by
political authorities at various levels. In this perspective governance
networks are revealed as one among many examples of regulated
self-regulation and the notion of metagovernance is expanded to other
studies of governance, which in turn might inspire and feed back into the
study of how to develop and sustain democratic network governance.
Bibliography
Abbott, K. W., Keohane, R. O., Moravcsik, A., Slaughter, A. M., Adcock, R., Bevir, M.
and Stimson, S. (2006) ‘Historicizing the New Institutionalism(s)’, in R. Adcock,
M. Bevir, and S. Stimson (eds), Modern Political Science: Anglo-American
Exchanges Since 1880 (Princeton, Princeton University Press). Forthcoming.
Adcock, R., Bevir, M. and Stimson, S. (eds) (2006). Modern Political Science: Anglo-
American Exchanges since 1880 (Princeton, Princeton University Press).
Adler, E. (1992) ‘The Emergence of Cooperation: National Epistemic
Communities and the International Evolution of the Idea of Arms Control’,
International Organization, 46 (1), pp.101–46.
Agranoff, R. I. (1986) Intergovernmental Management. Human Services Problem-Solving
in Six Metropolitan Areas (Albany, New York, State University of New York Press).
Agranoff, R. I. (2003) Leveraging Networks: A Guide for Public Managers Working
Across Organizations (Arlington, IBM Endowment for the Business of
Government).
Agranoff, R. and McGuire, M. (2001) ‘Big Questions in Public Network
Management Research’, Journal of Public Administration Research and Theory, 11
(3), pp. 295–326.
Agranoff, R. and McGuire, M. (2003) Collaborative Public Management; New Strategies
for Local Governments (Washington, DC, Georgetown University Press).
Aldrich, H. E. (1979) Organizations and Environments (Englewood Cliffs, NJ,
Prentice-Hall).
Aldrich, H. A. and Whetten, H. D. A. (1981) ‘Organization-sets, Action-sets
and Networks: Making the Most Out of Simplicity’, in P. C. Nystrom, and
W. H. Starbuck (eds), Handbook of Organizational Design (NY, Oxford University
Press), pp. 385–408.
Allison, G. T. (1971) Essence of Decision (Boston, Little, Brown and Company)
[second edition 1999, co-authored by Ph. Zelikow (New York, Longman)].
Almond, G. et al. (eds) (1973) Crisis, Choice and Change, Historical Sudies of Political
Development (Boston).
Alter, C. and Hage, J. (1993) Organizations working together (Newbury Park, Sage).
Aristoteles (1993) Der Staat der Athener (Stuttgart, Reclam).
Axelrod, R. A. (1984) The Evolution of Cooperation (New York, Basic Books).
Axelrod, R. A. (1986) ‘An Evolutionary Approach to Norms’, American Political
Science Review, 80 (4), pp. 1095–111.
Bache, I. (2000) ‘Government Without Governance: Network Steering in
Yorkshire and the Humber’, Public Administration, 78 (3), pp. 575–92.
Bachrach, P. and Baratz, M. S. (1962) ‘Two Faces of Power’, American Political
Science Review, 56 (4), pp. 947–52.
Bang, H. (ed.) (2003) Governance as Social and Political Communication
(Manchester, Manchester University Press).
Bang, H. P. and Hansen, A. P. (2005) Kvarterløft: Borgerinddragelse gennem
kulturstyring i Kgs. Enghave’, in H. P. Bang, T. B. Dyrberg, and J. Hoff (eds),
316
Bibliography 317
Magtens nye ansigt. Netværkspolitik, kulturstyring og ny elitisme (Copenhagen,

Jurist- og Økonomforbundets Forlag) pp. 123–67.
Bang, H. P., Hansen, A. D. and Hoff, J. (eds) (2000) Demokrati fra neden: Casestudier
fra en dansk kommune (Copenhagen, Jurist og Økonomforbundets Forlag).
Barber, B. (1984) Strong Democracy. Participatory Politics for a New Age (Los Angeles,
University of California Press).
Bardach, E. (1999) Getting Agencies to Work Together (Washington, DC, The
Brookings Institution).
Barns, M. (1997) ‘Challenges to the Practice and Theory of Public Administration
in Europe’, Journal of Theoretical Politics, 9 (3), pp. 389–415.
Bates, R. H., Figueiredo, R. J. P. and Weingast, B. R. (1998) ‘The Politics of
Interpretation: Rationality, Culture, and Transformation’, Politics & Society,
26 (4), pp. 603–42.
Beetham, D. (1999) Democracy and Human Rights (Cambridge, Polity).
Behn, R. D. (2001) Rethinking Democratic Accountability (Washington, DC, The
Brookings Institution).
Benhabib, S. (1996) ‘Toward a Deliberative Model of Democratic Legitimacy’, in
S. Benhabib (ed.), Democracy and Difference: Contesting the Boundaries of the
Political (Princeton, Princeton University Press), pp. 67–94.
Bennet, W. L. and Entmann, R. (eds) (2001) Mediated Politics – Communication in
the Future of Democracy (Cambridge, Cambridge University Press).
Benson, J. K. (1978) ‘The interorganizational network as a political economy’, in
L. Karpik (ed.), Organization and Environment (London, Sage), pp. 69–102.
Benson, J. K. (1982) ‘A Framework for Policy Analysis’, in D. L. Rogers, and
D. A. Whetten (eds), Interorganizational Coordination: Theory, Research, and
Implementation (Ames, Iowa State University Press), pp. 137–76.
Benz, A. (1995) ‘Politiknetzwerke in der horizontalen Politikverflechtung’, in
D. Jansen, and K. Schubert (eds), Netzwerke und Politikproduktion. Konzepte,
Methoden, Perspektiven (Marburg, Schüren).
Bevir, M. (2004) ‘Governance and Interpretation: What Are the Implications of
Post-foundationalism?’, Public Administration, 82, pp. 605–25.
Bevir, M. (2005) New Labour: A Critique (London, Routledge).
Bevir, M. and Rhodes, R. A. W. (2003) Interpreting British Governance (London,
Routledge).
Bevir, M. and Rhodes, R. A. W. (2006) Governance Stories (London, Routledge).
Bevir, M., Rhodes, R. A. W. and Weller, P. (2003) ‘Comparative Governance:
Prospects and Lessons’, Public Administration, 81, pp. 191–210.
Blom-Hansen, J. (1997) ‘A New Institutional Perspective on Policy Networks’,
Public Administration, 75 (winter), pp. 669–93.
Blumberg, B. F. (2001) ‘Cooperation Contracts Between Embedded Firms’,
Organization Studies, 22 (5), pp. 825–52.
Bogason, P. (1991) ‘Control for whom? Recent Advances in Research on
Governmental Guidance and Control’, European Journal of Political Research,
20 (2), pp. 189–208.
Bogason, P. (2000) Public Policy and Local Governance – Institutions in Postmodern
Society (Cheltenham, Edward Elgar).
Bogason, P. and Zølner, M. (eds) (2006) Analysing Democratic Network Governance:
Methodological issues (Basingstoke, Palgrave Macmillan).
318 Bibliography
Bohman, J. (1998) ‘The Coming of Age of Deliberative Democracy’, Journal of

Political Philosophy, 6 (4), pp. 399–423.
Bohman, J. (2005) ‘From Demos to Demoi: Democracy across Borders’, Ratio Juris,
18 (3), pp. 293–314.
Boli, J. and Thomas, G. M. (1999) Constructing World Culture. International
Nongovernmental Organizations Since 1875 (Stanford, CA, Stanford University
Press).
Börzel, T. A. (1998) ‘Organizing Babylon – On the Different Conceptions of Policy
Networks’, Public Administration, 76, pp. 253–73.
Börzel, T. A. and Hösli, M. (2003) ‘Brussels between Bern and Berlin. Comparative
Federalism Meets the European Union’, Governance, 16 (2), pp. 179–202.
Börzel, T. A. and Risse, T. (2005) ‘Public-Private Partnerships: Effective and
Legitimate Tools of Transnational Governance?’, in E. Grande, and L. W. Pauly
(eds), Complex Sovereignty: On the Reconstitution of Political Authority in the
21st. Century (Toronto, University of Toronto Press).
Bovens, M. A. P., Hart, P. T. and Peters, B. G. (2000) Success and Failure in Public
Governance (Cheltenham, Edward Elgar).
Bressers, H. (1983) Beleidseffectiviteit en waterkwaliteitsbeleid (Enschede,
Universiteit Twente).
Bressers, H. (2004) ‘Understanding the Implementation of Instruments: How
To Know What Works, Where, When and How’, in W. Lafferty (ed.), Governance
for Sustainable Development: The Challenge of Adapting Form to Function
(Cheltenham, Edward Elgar).
Bressers, H. and Klok, P. J. (1988) ‘Fundamentals for a Theory of Policy
Instruments’, International Journal of Social Economics, 15, pp. 22–41.
Bressers, H. and O’Toole, L. J. Jr. (1999) ‘The Selection of Policy Instruments:
A Network-Based Perspective’, Journal of Public Policy, 18, pp. 213–39.
Bressers, H. and O’Toole, L. J. Jr. (2005) ‘Instrument Selection and Implementation
in a Network Context’, in P. Eliadis, M. Hill, and M. Howlett (eds), From
Instrument Choice to Governance: Future Directions for the Choice of Governing
Instrument (Montreal, McGill Queens University Press), pp. 132–53.
Bressers, H., Klok, P. J. and O’Toole, L. J. Jr. (2000) ‘Explaining Policy Action:
A Deductive but Realistic Theory’, presented at the IPSA World Congress,
Quebec City.
Brock, L. (1998) ‘Die Grenzen der Demokratie: Selbstbestimmung im Kontext
des globalen Strukturwandels und des sich wandelnden Verhältnisses von
Staat und Markt’, Politische Vierteljahresschrift, 39, PVS Sonderheft 29,
pp. 271–92.
Brown, L. D. (1983) Managing Conflict at Organizational Interfaces (Reading, Mass.,
Addison-Wesley).
Brühl, T., Debiel, T., Hamm, B., Hummel, H. and Martens, J. (eds) (2001) Die
Privatisierung der Weltpolitik: Entstaatlichung und Kommerzialisierung im
Globalisierungsprozess (Bonn, Dietz).
Bruijn, J. A de, Heuvelhof, E. F. ten and Veld, R. J. In ‘t (1998)
Procesmanagement: Over procesontwerp en besluitvorming (Schoonhoven,
Academic Services).
Bruijn, J. A., Heuvelhof, de E. F. ten and Veld, R. J. in ‘t (2002) Process Management:
Why Project Management Fails in Complex Decision Making Processes (Dordrecht,
Kluwer Academic Publishers).
Bibliography 319
Brunsson, N. and Olsen, J. P. (1993) The Reforming Organization (London, Routledge).

Bueren, E. van and Edelenbos, J. (2004) ‘Conflicting Knowledge: Why
Is Knowledge Production Such a Problem’, Science & Public Policy, 31 (4),
pp. 289–99.
Bueren, E. van, Klijn, E. H. and Koppenjan, J. F. M. (2003) ‘Dealing with Wicked
Problems in Networks: Analysing an Environmental Debate From a Network
Perspective’, Journal of Public Administration Research and Theory, 13,
pp. 193–212.
Burchell, G. (1996) ‘Liberal Government and Techniques of the Self’, in A. Barry,
T. Osborne, and N. Rose (eds), Foucault and Political Reason (London, UCL Press).
Burns, T. R. and Flam, H. (1987) The Shaping of Social Organisation: Social Rule
System Theory with Application (London, Sage).
Burton, J. W. (1996) Conflict Resolution: Its Language and Processes (London).
Butler, C. T. (1987) On Conflict and Consensus (Takoma Park, Maryland, FNB
Publishing).
Calvert, R. L. (1995) ‘The Rational Choice Theory of Institutions: Implications for
Design’, in D. Weimer (ed.), Institutional Design (Dordrecht, Kluwer).
Campbell, J. L. (1994) Recent Trends in Institutional Analysis (Copenhagen, COS
Research Report 5).
Campbell, J. L. (1997) ‘Mechanisms of Evolutionary Change in Economic
Governance: Interaction, Interpretation and Bricolage’, in L. Magnusson, and
J. Ottosson (eds), Evolutionary Economics and Path-Dependency (Cheltenham,
Edward Elgar), pp. 10–32.
Campbell, J. L. and Pedersen, O. K. (eds) (2001) The Rise of Neoliberalism and
Institutional Analysis (Princeton, Princeton University Press).
Carlsson, L. (2000) ‘Policy Network as Collective Action’, Policy Studies Journal,
28 (3) 502–20.
Castells, M. (1996) The Rise of the Network Society (Cambridge, MA, Blackwell
Publishers).
Castells, M. (2000, 1st edn 1996) The Rise of the Network Society: Economy, Society
and Culture (Cambridge, Blackwell Publishers).
Cawson, A. (1986) Corporatism and Political Theory (Oxford, Blackwell).
Chalmers, J. and Davis, G. (2001) ‘Rediscovering implementation: Public sector
contracting and human services’, Australian Journal of Public Administration,
60 (2), pp. 74–85.
Chambers, S. (2003) ‘Deliberative Democratic Theory’, Annual Review of Political
Science, 6, pp. 307–26.
Chan, J. (1997) Changing Police Culture: Policing in a Multicultural Society
(Cambridge, Cambridge University Press).
Chayes, A. and Chayes, A. H. (1993) ‘On Compliance’, International Organization,
47 (2), pp. 175–205.
Chayes, A., Chayes, A. H. and Mitchell, R. B. (1998) ‘Managing Compliance:
A Comparative Perspective’, in E. B. Weiss, and H. K. Jacobsen (eds), Engaging
Countries: Strengthening Compliance with International Environmental Accords
(Cambridge, Mass., MIT Press).
Checkel, J. T. (2001) ‘Why Comply?! Social Learning and European Identity
Change’, International Organization, 55 (39), pp. 553–88.
Chiles, T. H. and McMackin, J. (1996) ‘Integrating Variable Risk Preferences, Trust,
and Transaction Costs Economics’, Academy of Management Review, 3, pp. 73–99.
320 Bibliography
Clark, P. B. and Wilson J. Q. (1961) ‘Incentive Systems: A Theory of Organizations’,

Adminstrative Science Quarterly, 6 (2), pp. 129–66.
Clegg, S. R. (1990) Modern Organizations: Organization Studies in the Postmodern
World (London, Sage).
Cobb, R. W. and Elder, C. D. (1983, 1st edn 1972) Participation in American Politics:
The Dynamics of Agenda-building (Baltimore, John Hopkins University Press).
Cohen, J. (1989) ‘Deliberation and Democratic Legitimacy’, in A. Hamlin, and
P. Pettit (eds), The Good Polity: Normative Analysis of the State (Oxford, Basil
Blackwell), pp. 17–34.
Cohen, M. D., March, J. G. and Olsen, J. P. (1972) ‘A Garbage Can Model of
Organizational Choice’, Administrative Science Quarterly, 17 (1), pp. 1–25.
Coleman, J. S. (1986) ‘Norms as Social Capital’, in G. Radnitzky, and P. Bernholz
(eds), Economic Imperialism. The Economic Method Applied Outside the Field of
Economics (New York, Paragon House Publishers).
Coleman, J. S. (1990) Foundations of Social Theory (Cambridge/London, Harvard
University Press).
Connolly, W. E. (1991) Identity/Difference. Democratic Negotiations of Political
Paradox (Ithaca, Cornell University Press).
Connolly, W. E. (1993) ‘Democracy and Contingency’, in J. H. Carens (ed.),
Democracy and Possessive Individualism. The Intellectual Legacy of C. B. Macpherson
(New York, State University of New York Press).
Connolly, W. E. (1995) The Ethos of Pluralization (Minneapolis, University of
Minnesota Press).
Coser, L. A. (1956) The Functions of Social Conflict (New York, Free Press).
Covaleski, M. A. and Dinsmith, M. W. (1988) ‘An Institutional Perspective on the
Rise, Social Transformation and Fall of a University Budget Category’,
Administrative Science Quarterly, 33 (4), pp. 562–87.
Crozier, M. and Friedberg, E. (1980) Actors and Systems: The Politics of Collective
Action (Chicago, University of Chicago Press).
Cruikshank, B. (1999) The Will to Empower (Ithaca, Cornell University Press).
Cunningham, F. (2002) Theories of Democracy: A Critical Introduction (London/
New York, Routledge).
Cutler, C. A., Haufler, V. and Porter, T. (eds) (1999) Private Authority and International
Affairs (Albany, State University of New York).
Damgaard, B. (2002) ‘Building local governance in Denmark: National
legislated local partnerships’, SFI Working Paper, 35: 2002. (Available at:
http://www.sfi.dk/graphics/SFI/Pdf/Working_papers/WP-200235-BD-
pdf.prn.pdf).
Davies, A. and Thomas, R. (2003) ‘Talking COP: Discourses of Change and Police
Identities’, Public Administration, 81, pp. 681–99.
Deakin, S. and Michie, J. (eds) (1997) Contract, Co-operation, and Competition:
Studies in Economics, Management and Law (Oxford, Oxford University Press).
Deakin, S. and Wilkinson, F. (1998) ‘Contract Law and the Economics of
Interorganisational Trust’, in C. Lane, and R. Bachman (eds), Trust Within and
Between Organizations: Conceptual Issues and Empirical Applications (Oxford,
University Press, Oxford).
Dean, M. (1999) Governmentality: Power and Rule in Modern Society (London, Sage).
Deitelhoff, N. (2003) Taking States into Persuasion. Norm-Entrepreneurs and the
Manipulative Nature of Communicative Settings in Norm-Construction. (Available
at: http://www.ibdiskurse.de/ipc/papers/deitelhoff.doc) (Frankfurt am Main).
Bibliography 321
Deleuze, G. and Guattari, F. (1987) A Thousand Plateaus: Capitalism and

Schizophrenia (London, Continuum).
Denters, B. O., Huisman, J. V. H. and Klok, P. J. (2003) The Rise of Interactive
Governance and Quasi-Markets (The Hague, Kluwer).
Dery, D. (1984) Problem Definition in Policy Analysis (Kansas, University Press of
Kansas).
Dimaggio, P. and Powell, W. W. (1983) ‘The Iron Cage Revisited: Institutional
Isomorphism and Collective Rationality in Organizational Fields’, American
Sociological Review, 48, pp. 147–60.
Donzelot, J. (1988) ‘The Promotion of the Social’, Economy and Society, 17 (3),
pp. 395–427.
Dowding, K. (1995) ‘Model or Metaphor: A Critical Review of the Policy Network
Approach’, Political Studies, 43, pp. 136–58.
Downs, G. W., Rocke, D. M. and Barsoom, P. N. (1996) ‘Is the Good News About
Compliance Good News About Cooperation?’, International Organization,
50 (3), pp. 379–406.
Dryzek, J. S. (1990) Discursive Democracy: Politics, Policy and Political Science
(New York, Cambridge University Press).
Dryzek, J. S. (1996) Democracy in Capitalist Times: Ideals, Limits and Struggles
(New York, Oxford University Press).
Dryzek, J. S. (2000) Deliberative Democracy and Beyond: Liberals, Critics, Contestations
(Oxford, Oxford University Press).
Dunleavy, P. (1991) Democracy, Bureaucracy and Public Choice: Economic
Explanations in Political Science (Hertfordshire, Harvester Wheatsheaf).
Duran, P. (2000) Penser l’action sociale (Paris, L. G. D.J.).
Duverger, M. (1959) Political Parties: Their Organization and Activity in the Modern
State (London, Methuen).
Dyrberg, T. B. and Torfing, J. (1992) ‘Politik og institutioner’, Statsvetenskaplig
Tidskrift, 95 (2), pp. 59–71.
Easton, D. (1957) ‘An Approach to the Analysis of Political Systems’, World
Politics, 9, (3), pp. 383–400.
Easton, D. (1965) A Systems Analysis of Political Life (New York, Wiley & Sons).
Eckersely, R. (2000) ‘Deliberative Democracy, Ecological Representation and Risk:
Towards a democracy of the Affected’, in M. Saward, (ed.), Democratic
Innovation: Deliberation, Association, and Representation (London, Routledge),
pp. 117–32.
Edelenbos, J. (2000) Proces in vorm; procesbegeleiding van interactieve beleidsvorming
over locale ruimtelijke projecten (Utrecht, Lemma).
Edelenbos, J. (2005) ‘Institutional Implications of Interactive Governance:
Insights from Dutch Practice’, Governance, 18 (1), pp. 111–34.
Edelenbos, J. and Klijn, E. H. (2006a) ‘Managing Stakeholder Involvement
in Decision-making: A Comparative Analysis of Six Interactive Processes in
The Netherlands’, Journal of Public Administration Research and Theory.
Forthcoming.
Edelenbos, J., and Klijn, E. H. (2006b) ‘Trust in Complex Networks: A Theoretical
and Empirical Exploration’, Administration and Society.
Eeten, M. van (1999) Dialogues of the Deaf: Defining New Agendas for Environmental
Deadlocks (Delft, Eburon). Forthcoming.
Ehrmann, J. R. and Stinson, B. L. (1999) ‘Joint fact-finding and the use of technical
experts’, in L. Süsskind, S. McKearnan, and J. Thomas-Larmer (eds), The
322 Bibliography
Consensus building Handbook: A Comprehensive Guide to Reaching Agreement

(London, Sage), pp. 375–400.
Eijk, C. van der, and Kok, W. J. P. (1975) ‘Nondecisions Reconsidered’, Acta
Politica, 10 (3), pp. 277–301.
Ellwein, T., Hesse, J. J., Mayntz, R. and Scharpf, F. W. (eds) (1987) Jahrbuch zur
Staats- und Verwaltungswissenschaft, 1 (Baden-Baden, Nomos).
Elster, J. (1983) ‘Offentlighet og deltakelse. To teorier om deltakerdemokratiet’, in
T. Bergh (ed.), Deltakerdemokratiet. Teori og praksis (Oslo, Universsitetsforlaget).
Elster, J. (1985) Sour Grapes: Studies in the Subversion of Rationality (Cambridge,
Cambridge University Press).
Elster, J. (1989) Nuts and Bolts for the Social Sciences (Cambridge, Cambridge
University Press).
Elster, J. (1990) Vetenskapliga förklaringar (Göteborg, Korpen).
Elster, J. (ed.) (1998) Deliberative Democracy (Cambridge, Cambridge University
Press).
Eriksen, E. O. and Weigaard, J. (2003) Kommunikativt Demokrati – Jürgen Habermas’
teori om politik og samfund (Copenhagen, Hans Reitzel).
Esmark, A. (2003) ‘Network Governance Between Public and Elites’, Unpublished
paper (Durban, the IPSA-conference ‘Democracy, Tolerance, Justice: Challenges
for political change’, June/July 2003).
Esmark, A. and Triantafillou, P. (2006) ‘Document Analysis of Network Topography
and Network Programmes’, in P. Bogason, and M. Zølner (eds), Analyzing
Democratic Network Governance: Methodological issues (Palgrave MacMillan,
Basingstoke). Forthcoming.
Esping, H. (1994) Ramlagar i förvaltningspolitiken (Stockholm, SNS Förlag).
Etzioni-Halevy, E. (1964) Modern Organizations (Englewood Cliffs, NJ,
Prentice-Hall).
Etzioni-Halevy, E. (1993) The Elite Connection: Problems and Potential of Western
Democracy (Cambridge, Polity Press).
Etzioni-Halevy, E. (2003) ‘Network Governance as a Challenge to Democratic
Elite Theory’, Paper presented at the Conference on Democratic Network Governance,
Centre for Democratic Network Governance, Roskilde University.
EU (2001a) European Governance: A White Paper (Brussels, The Commission of the
European Communities).
EU (2001b) Report of Working Group ‘Networking People for a Good Governance
in Europe’. White Paper on Governance (Work Area no. 4, Coherence and
Cooperation in a Networked Europe).
Evan, W. M. (1976) Interorganizational Relations (Harmondsworth, Penguin Books).
Fairclough, N. (1995) Media Discourse (London, Arnold).
Ferejohn, J. (1991) ‘Rationality and Interpretation: Parliamentary Elections in
Early Stuart England’, in K. R. Monroe (ed.), The Economic Approach to Politics:
A Critical Reassessment of the Theory of Rational Action (New York, Harper Collins
Publishers).
Finer, H. (1954) The Theory and Practice of Modern Government (London, Methuen).
Finlayson, A. et al. (2004) ‘The Interpretive Approach in Political Science:
A Symposium’, British Journal of Politics and International Relations, 6, pp. 1–36.
Fischer, F. and Forester, J. (eds) (1993) The Argumentative Turn in Policy Analysis and
Planning (Duke Durham, NC, University Press).
Bibliography 323
Fisher, R. and Ury, W. (1981) Getting to Yes: Negotiating Agreement Without Giving
in (Boston, Houghton Mifflin).
Fleming, J. (2006) ‘Networks and Police Organisational Change’, in J. Fleming,
and J. Wood (eds), Policing, Security and Democracy in a Networked Society
(Sydney, UNSW Press). Forthcoming.
Fleming, J. and Rhodes, R. A. W. (2005) ‘Bureaucracy, Contracts and Networks:
The Unholy Trinity and the Police’, Australian and New Zealand Journal of
Criminology, 38, pp. 192–205.
Florini, A. (2000a) ‘Lessons Learned’, in A. Florini (ed.), The Third Force: The Rise
of Transnational Civil Society (Tokyo, Washington, DC, Japan Center for
International Exchange and Carnegie Endowment for International Peace).
Florini, A. (ed.) (2000b) The Third Force: The Rise of Transnational Civil Society
(Tokyo/Washington, DC. Japan Centre for International Exchange/Carnegie
Endowment for International Peace).
Forester, J. (1989) Planning in the Phase of Power (Berkeley, University of California
Press).
Foucault, M (1976) The History of Sexuality (Harmondsworth, Penguin Books).
Foucault, M. (1978) Sécurité, territorie et population. Cours au Collége de France.
Unpublished. Available at Bibliothéque IMEC, Rue Bleue, Paris.
Foucault, M. (1979) Discipline and Punish: The Birth of the Prison (Harmondsworth,
Penguin).
Foucault, M. (1980) ‘Truth and power’, in C. Gordon (ed.), Power/Knowledge:
Selected Interviews and Other Writings 1972–1977 (New York, Pantheon Books),
pp. 109–33.
Foucault, M. (1982) ‘The Subject and Power’, in H. L. Dreyfus, and P. Rabinow,
Michel Foucault: Beyond Structuralism and Hermeneutics (Chicago, University of
Chicago Press).
Foucault, M. (1986) ‘The subject of power’, in C. Gordon (ed.), Power/Knowledge:
Selected Interviews and Other Writings 1972–1977 (New York, Pantheon Books).
Foucault, M. (1988) ‘Politics and Reason’, in L. D. Kritzman (ed.), Michel Foucault:
Politics, Philosophy, Culture (London, Routledge), pp. 57–85.
Foucault, M. (1990) The History of Sexuality, 1 (Harmondsworth, Pelican).
Foucault, M. (1991a) ‘Governmentality’, in G. Burchell, C. Gordon, and P. Miller
(eds), The Foucault Effect (Hertfordshire, Harvester Wheatsheaf), pp. 87–104.
Foucault, M. (1991b) ‘Questions of method’, in G. Burchell, C. Gordon, and P. Miller
(eds), The Foucault Effect (Hertfordshire, Harvester Wheatsheaf), pp. 73–86.
Frances, J. (1991) ‘Introduction’, in G. Thompson et al. (eds), Markets Hierarchies
and Networks: The Co-ordination of Social Life (London, Sage).
Freeman, J. L. (1965) The Political Process (New York, Random House).
Friedland, R. and Alford, R. R. (1991) ‘Bringing Society Back in: Symbols, prac-
tices, and institutional contradiction’, in W. W. Powell, and P. J. DiMaggio
(eds), The New Institutionalism in Organizational Analysis (Chicago, University of
Chicago Press), pp. 332–66.
Friend, J. K., Power, J. M. and Yewlett, C. J. L. (1974) Public Planning: The Inter-
corporate Dimension (London, Tavistock).
Fung, A. and Wright, E. O. (eds) (2003) ‘Deepening Democracy: Institutional
Innovations’, Empowered Participatory Governance, Real Utopias Project Series,
IV (London, Verso).
324 Bibliography
Gage, R. W. and Mandell, M. P. (eds) (1990) Strategies for Managing Intergovernmental

Policies and Networks (New York/London, Preager).
Galtung, F. (2000) ‘A Global Network to Curb Corruption: The Experience of
Transparency International’, in A. Florini (ed.), The Third Force: The Rise of
Transnational Civil Society (Tokyo/Washington, DC, Japan Center for
International Exchange and Carnegie Endowment for International Peace).
Gay, P. du (2000) In Praise of Bureacracy (London, Sage).
Geddes, B. (2003) Paradigms and Sand Castles: Theory Building and Research Design
in Comparative Politics (Ann Arbor, University of Michigan Press).
Geertz, C. (1973) The Interpretation of Cultures (New York, Basic Books).
Gibbons, M., Limonges, C., Nowotny, H., Scott, S. and Trow, M. (1994) The New
Production of Knowledge: The Dynamics of Science and Research in Contemporary
Societies (London, Sage).
Giddens, A. (1984) The Constitution of Society: Outline of the Theory of Structuration
(London/Berkeley/Los Angeles, Macmillan).
Glasbergen, P. (ed.) (1995) Managing Environmental Disputes, Network Management
as an Alternative (Dordrecht, Kluwer).
Gordon, C. (1991) ‘Governmental rationality: An introduction’, in G. Burchell,
C. Gordon and P. Miller (eds), The Foucault Effect (Hertfordshire, Harvester
Wheatsheaf), pp. 1–52.
Gordon C. and Miller, P. (eds) (1991) The Foucault Effect (Hertfordshire, Harvester
Wheatsheaf), pp. 1–52.
Goss, S. (2001) Making Local Governance Work (Houndmills, Basingstoke, Palgrave).
Gutmann, A. and Thompson, D. (1996) Democracy and Disagreement (Cambridge,
The Belknap Press).
Gutmann, A. and Thompson, D. (2003) ‘Deliberative Democracy Beyond
Process’, in J. S. Fishkin, and P. Laslett (eds), Debating Deliberative Democracy
(London, Blackwell Publishing), pp. 31–54.
Haahr, J. H. (2004) ‘Open co-ordination as advanced liberal government’, Journal
of European Public Policy, 11 (2), pp. 209–30.
Habermas, J. (1962) Struturwandel der Öffentlichkeit – Untersuchungen zu einer
Kategorie der bürgerlichen Gesellschaft (Frankfurt, Hermann Luchterland
Verlag).
Habermas, J. (1976) ‘Legitimationsprobleme im modernen Staat’, in
P. G. Kielmansegg (ed.), Legitimationsprobleme politischer Systeme (Opladen,
Westdeutscher Verlag).
Habermas, J. (1991) Erläuterungen zur Diskursethik (Frankfurt an Main, Suhrkamp).
Habermas, J. (1995–97 [1981]) The Theory of Communicative Action (Cambridge,
Polity Press).
Habermas, J. (1996) Between Facts and Norms: Contributions to a Discourse Theory of
Law and Democracy (Cambridge, MA, MIT Press).
Habermas, J. (1998) Die Postnationale Konstellation – Politische Essays (Frankfurt
am Main, Suhrkamp).
Hajer, M. (1993) ‘Discourse Coalitions and the Institutionalization of Practice:
The Case of Acid Rain in Britain’, in F. Fischer, and J. Forester (eds), The
Argumentative Turn in Policy Analysis and Planning (Durham, Duke University
Press), pp. 43–76.
Hajer, M. (1995) The Politics of Environmental Discourse: Ecological Modernization
and the Policy Process (Oxford, Clarendon Press).
Bibliography 325
Hajer, M. (2003) ‘A Frame in the Fields: Policymaking and Reinvention of

Politics’, in M. Hajer (ed.), Deliberative Policy Analysis. Understanding Governance
in the Network Society (Cambridge, Cambridge University Press).
Hajer, M. and Versteeg, W. (2005) ‘Performing Governance through Networks’,
European Political Studies, 4 (3), 340–7.
Hajer, M. and Wagenaar, H. (eds) (2003) Deliberative Policy Analysis; Understanding
Governance in the Network Society (Cambridge, Cambridge University Press).
Hall, P. A. (1986) Governing the Economy: The Politics of State Intervention in Britain
and France (Cambridge, Polity Press).
Hall, P. A. (1989) ‘Introduction’, in P. A. Hall (ed.), The Political Power of Economics
Ideas: Keynesianism across Nations (Princeton, Princeton University Press),
pp. 3–26.
Hall, P. A. (1993) ‘Policy Paradigms, Social Learning, and the State: The Case of
Economic Policymaking in Britain’, Comparative Politics, 25 (3), pp. 275–96.
Hall, P. A. and Taylor, R. C. R. (1996) ‘Political Science and the Three New
Institutionalisms’, Political Studies, 44 (4), pp. 936–57.
Hall, T. E. and O’Toole, L. J. Jr. (2000) ‘Structures for Policy Implementation: An
Analysis of National Legislation, 1965–66 and 1993–94’, Administration and
Society, 31, pp. 667–86.
Hall, T. E. and O’Toole, L. J. Jr. (2004) ‘Shaping Formal Networks Through the
Regulatory Process’, Administration and Society, 36, pp. 186–207.
Hammersley, M. (1990) Reading Ethnographic Research. A Critical Guide (Harlow,
Essex, Longman).
Hanf, K. and O’Toole, L. J. (1992) ‘Revisiting Old Friends: Networks,
Implementation Structures and the Management of Inter-organizational
Relations’, European Journal of Political Research, 21, pp. 163–80.
Hanf, K. and Scharpf, F. W. (eds) (1978) Interorganizational Policy-making: Limits to
Central Coordination and Control (London, Sage).
Hanf, K. and Toonen, Th. A. J. (1985) Policy Implementation in Federal and Unitary
Systems (Dordrecht, Martinus Nijhoff Publishers).
Hardin, G. (1986) ‘The Tragedy of the Commons’, Science, 162, pp. 1243–8.
Hardin, R. (1982) Collective Action (Baltimore, MD, Johns Hopkins University Press).
Harding, A. (1998) ‘Regime Formation in Manchester and Edinburgh’, in G. Stoker
(ed.), The New Politics of British Local Governance (London, Macmillan).
Hart, P. ‘t (1990) Groupthink in Government: A Study of Small Groups and Policy
Failure (Amsterdam, Swetz en Zeitlinger).
Häusler, J., Hohn, H. W. and Lütz, S. (1993) ‘The Architecture of a R&D
Collaboration’, in F. W. Scharpf (ed.), Games in Hierarchies and Networks:
Analytical and Empirical Approaches to the Study of Governance Institutions
(Frankfurt am Main/Colorado, Campus Verlag/Westview Press).
Hay, C. (2002) Political Analysis (Houndmills, Basingstoke, Palgrave).
Hay, C. (2004) ‘Theory, Stylized Heuristic or Self-fulfilling Prophecy?: The Status
of Rational Choice Theory in Public Administration’, Public Administration, 82
(4) pp. 39–63.
Hay, C. and Richards, D. (2000) ‘The Tangled Webs of Westminister and
Whitehall: The Discourse, Strategy and Practice of Networking Within British
Core Executive’, Public Administration, 78 (1), pp. 1–28.
Healey, P. (1997) Collaborative Planning: Shaping Places in Fragmented Societies
(London, MacMillan).
326 Bibliography
Heclo, H. (1978) ‘Issue Networks and the Executive Establishment’, in A. King

(ed.), The New American Political System (Washington, DC, American Enterprise
Institute for Public Policy Research, pp. 87–124.
Heclo, H. and Wildavsky, A. (1974) The Private Government of Public Money
(London, Macmillan).
Hedström, P. and Swedberg, R. (1996) ‘Social Mechanisms’, Acta Sociologica, 38,
pp. 280–307.
Heffen, O. V., Kickert, W. J. M. and Thomassen, J. A. (2000) Governance in Modern
Society: Effects, Change and Formation of Government Institutions (Dordrecht,
Kluwer Academic Publishers).
Heisler, M. O. (1974) ‘Patterns of European Politics: The European ‘Polity’ Model’,
in M. O. Heisler (ed.), Politics in Europe: Structures and Processes in Some Post-
industrial Democracies (New York, David McKay).
Held, D. (1987) Models of Democracy (Cambridge, Polity Press).
Held, D. (1995) Democracy and the Global Order: From the Nation State to
Cosmopolitan Governance (Cambridge, Polity).
Held, D. (2000) Models of Democracy (Cambridge, Polity Press).
Héritier, A. (2002) ‘New Modes of Governance in the European Union: Policy-
Making Without Legislation?’, in A. Héretier (ed.), Common Goods: Reinventing
European and International Governance (Lanham, Rowman and Littlefield).
Héritier, A. (2003) ‘New Modes of Governance in Europe: Increasing Political
Capacity and Policy Effectiveness?’, in T. A. Börzel, and R. Cichowski (eds), The
State of the European Union, 6 – Law, Politics, and Society (Oxford, Oxford
University Press).
Hermansson, J. (1990) Spelteorins nytta (Uppsala, Acta).
Hermansson, J. (2003) Politik på upplysningens grund: Om den demokratiska
reformismens möjligheter och problem (Stockholm, Liber).
Hernes, G. (1975) Makt og avmakt (Bergen, Universitetsförlaget).
Hertting, N. (2003) Samverkan på spel: Rationalitet och frustration i nätverksstyrning
och svensk stadsdelsförnyelse (Stockholm, Egalité).
Hindmoor, A. (1998) ‘The importance of being trusted: transaction costs and
policy network theory’, Public Administration, 76 (spring), pp. 25–43.
Hirst, P. (1994) Associative Democracy. New Forms of Economic and Social Governance
(Cambridge, Polity Press).
Hirst, P. (2000) ‘Democracy and Governance’, in J. Pierre (ed.), Debating Governance
Hjern, B. and Hull, C. (1982a) ‘Implementation Research as Empirical
Constitutionalism’, European Journal of Political Research, 10, pp. 105–16.
Hjern, B. and Hull, C. (1982b) ‘Implementation Research – the Link Gone
Missing’, Journal of Public Policy, 2 (3), pp. 301–8.
Hjern, B. and Hull C. (1984) ‘Going Interorganizational: Weber Meets Durkheim’,
Scandinavian Political Studies, 7 (3), pp. 197–212.
Hjern, B. and Porter, D. O. (1981) ‘Implementation Structures: A New Unit for
Administrative Analysis’, Organizational Studies, 3, pp. 211–37.
Hodgson, G. M. (1994) ‘Lock-in and Chreodic Development’, in G. M. Hodgson,
W. J. Samuels and M. R. Tool (eds), The Elgar Companion to Institutional and
Evolutionary Economics L–Z (Aldershot, Edward Elgar), pp. 15–19.
Hodson, D. and Maher, I. (2001) ‘The Open Method as a New Mode of Governance:
The Case of Soft Economic Policy Co-ordination’, Journal of Common Market
Studies, 39 (4), pp. 719–46.
Bibliography 327
Holden, B. (1988) Understanding Liberal Democracy (Oxford, Philip Allan).

Holden, B. (1993) Understanding Liberal Democracy (New York, Harvester).
Holzinger, K. (2004) ‘Bargaining Through Arguing: An Empirical Analysis Based
on Speech Act Theory’, Political Communication, 21, pp. 195–222.
Hoppe, R. (1999) ‘Policy Analysis, Science and Politics, from “Speaking Truth to
Power” to “Making Sense Together” ’, Science and Public Policy, 26, (3), pp. 201–10.
Howarth, D. (2000) Discourse (Buckingham, Open University Press).
Huntington, S. P. (1968) Political Order in Changing Societies (New Haven, CT, Yale
University Press).
Hurd, I. (1999) ‘Legitimacy and Authority in International Politics’, International
Organization, 53 (2), pp. 379–408.
Jachtenfuchs, M. (2003) ‘Regieren jenseits der Staatlichkeit’, in G. Hellmann,
K. D. Wolf and M. Zürn (eds), Die neuen internationalen Beziehungen (Baden-
Baden, Nomos).
Janis, I. L. (1982) Groupthink. Psychological Studies of Policy Decisions and Fiascoes
(Boston, Houghtin Mifflin).
Jessop, B. (1998) ‘The Rise of Governance and the Risk of Failure: The Case
of Economic Development’, International Social Science Journal, 50 (155), pp. 29–45.
Jessop, B. (2000a) ‘Governance Failure’, in G. Stoker (ed.), The New Politics of
British Local Governance (Houndmills, Basingstoke, Macmillan).
Jessop, B. (2000b) ‘The Network Society, New Forms of Governance, and Democratic
Renewal’, unpublished paper (Copenhagen, Center for offentlig organisation og
styring, Copenhagen Business School).
Jessop, B. (2002) The Future of the Capitalist State (Cambridge, Polity Press).
Johansson, R. and Borell, K. (1999) ‘Central Steering and Local Networks: Old-age
Care in Sweden’, Public Administration, 77 (3), pp. 585–98.
John, P. (2001) Local Governance (London, Sage).
Johnson, J. D. (1991) ‘Rational Choice as a Reconstructive Theory’, in K. R. Monroe
(ed.), The Economic Approach to Politics: A Critical Reassessment of the Theory of
Rational Action (New York, Harper Collins Publishers).
Johnson, N. (1973) Government in the Federal Republic of Germany: The Executive at
Work (Oxford, Pergamon Press).
Jong, M. de (1999) Institutional Transplantation: How to Adopt Good Transport
Infrastructure Decision-making Ideas from Other Countries? (Delft, Eburon
Publishers).
Jordan, G. (1990) ‘Sub-Governments, Policy Communities and Networks:
Refilling the Old Bottles?’, Journal of Theoretical Politics, 2 (3), pp. 319–38.
Kaiser, K. (1971) ‘Transnational Relations as a Threat to the Democratic Process’,
in R. O. Keohane and J. S. Jr. Nye (eds), Transnational Relations in World Politics
(Cambridge, MA., Harvard University Press).
Katz, D. and Kahn, R. L. (1978) The Social Psychology of Organizations (New York,
Wiley).
Kaufmann, F. X., Majone, G. and Ostrom, V. (eds) (1986) Guidance, Control and
Evaluation in the Public Sector: The Bielefeld Interdisciplinary Project (Berlin, Walter
de Gruyter).
Keck, M. and Sikkink, K. (1998) Activists Beyond Borders: Advocacy Networks in
International Politics (Ithaca, NY, Cornell University Press).
Kenis, P., and Schneider, V. (1991) ‘Policy Networks and Policy Analysis. Scrutinizing
a New Analytical Toolbox’, in B. Marin, and R. Mayntz (eds), Policy Networks:
Empirical Evidence and Theoretical Considerations (Campus Verlag), pp. 25–59.
328 Bibliography
Keohane, R. O. and Nye, J. S. Jr. (2001) ‘Globalization: What’s New? What’s Not
(And so What?)’, Foreign Policy, 118, pp. 104–99.
Keohane, R. O., Moravcsik, A. and Slaughter, A. M. (2000) ‘Legalized Dispute
Resolution: Interstate and Transnational’, International Organization, 54 (3),
pp. 457–88.
Kersbergen, K. V. and Waarden, F. V. (2004) ‘Governance’ as a Bridge Between
Disciplines: Cross-disciplinary Inspiration Regarding Shifts in Governance and
problems of governability, accountability and legitimacy’, European Journal of
Political Research, 43 (2), pp. 143–71.
Kickert, W. J. M. (1993) ‘Autopoiesis and the Science of (Public) Administration:
Essence, Sense and Nonsense’, Organization Studies, 14, (2), pp. 261–78.
Kickert, W. J. M. and Koppenjan, J. F. M. (1997) ‘Public Management and Network
Management: An Overview’, in W. J. M. Kickert, E. H. Klijn and J. F. M. Koppenjan
(eds), Managing Complex Networks: Strategies for the Public Sector (London, Sage),
pp. 35–62.
Kickert, W. J. M., Klijn, E. H. and Koppenjan, J. F. M. (eds) (1997) Managing
Complex Networks (London, Sage).
King, L. A. (2003) ‘Deliberation, Legitimacy, and Multilateral Democracy’,
Governance, 16 (1), pp. 23–50.
Kingdon, J. W. (1984) Agendas, Alternatives and Public Policies (Boston, Little,
Brown & Company) [2nd edn 1995, New York: Harper Collins College
Publishers].
Kiser, L. L. and Ostrom, E. (1982) ‘The Three Worlds of Action: A Metatheoretical
Synthesis of Institutional Approaches’, in Ostrom, E. (ed.), Strategies of Political
Inquiry (Beverly Hills, Sage), pp. 179–222.
Kiss, G. (1990) Grundzüge und Entwicklung der Luhmannschen Systemtheorie
(Stuttgart, F. Enkeverlag).
Klijn, E. H. (1996) ‘Analysing and Managing Policy Processes in Complex
Networks; A Theoretical Examination of the Concept Policy Network and its
Problems’, Administration and Society, 28 (1), pp. 90–119.
Klijn, E. H. (1997) ‘Policy networks: An Overview’, in W. J. M. Kickert, E. H. Klijn,
and J. F. M. Koppenjan (eds), Managing Complex Networks: Strategies for the Public
Sector (London, Sage), pp. 14–34.
Klijn, E. H. (2001) ‘Rules as Institutional Context for Decision-making in
Networks: The Approach to Post-war Housing Districts in Two Cities’,
Administration and Society, 33 (3) (may), pp. 133–64.
Klijn, E. H. and Edelenbos, J. (2006) ‘Network Management as Meta-Governance’,
Chapter 11, this volume.
Klijn, E. H. and Koppenjan, J. F. M. (2000a) ‘Public Management and Policy
Networks: Foundations of a Network Approach to Governance’, Public
Management, 2, (2), pp. 135–58.
Klijn, E. H. and Koppenjan, J. F. M. (2000b) ‘Interactive Decision Making and
Representative Democracy: Institutional Collisions and Solutions’, in O. V.
Heffen, W. J. M. Kickert, and J. A. Thomassen (eds), Governance in Modern
Society: Effects, Change and Formation of Government Institutions (Dordrecht,
Kluwer Academic Publishers), pp. 114–24.
Klijn, E. H. and Koppenjan, J. F. M. (2004) Managing Uncertainties in Networks: A
Network Approach to Problem Solving and Decision-making (London, Routledge).
Bibliography 329
Klijn, E. H., Koppenjan, J. F. M. and Termeer, C. J. A. M. (1995) ‘Managing

Networks in the Public Sector: A Theoretical Study of Management Strategies in
Policy Networks’, Public Administration, 73 (3), pp. 437–54.
Kneer, G. and Nassehi, A. (1991) ‘Verstehen des Vestehens. Eine systemtheoretische
Revision der Hermeneutik’, Zeitschrift der Soziologie, 19, pp. 341–56.
Knox, H., Savage, M. and Harvey, P. (2006) ‘Social Networks and the Study of
Relations: Networks as Method, Metaphor and Form’, Economy and Society,
vol. 35, no. 1, pp. 113–40.
Koch, H. (1981 [1946]) Hvad er demokrati? (Copenhagen, Gyldendal).
Kohler-Koch, B. (1997) Interactive Governance: Regions in the Network of European
Politics (Mannheim, Universität Mannheim).
Kohler-Koch, B. (1999) ‘The Evolution and Transformation of European
Governance’, in B. Kohler-Koch, and R. Eising (eds), The Transformation of
Governance in the European Union (London, Routledge).
Kohler-Koch, B. and Eising, R. (eds) (1999) The Transformation of Governance in
Europe (London, Routledge).
Kooiman, J. (ed.) (1993) Modern Governance: New Government–Society Interactions
(London, Sage).
Kooiman, J. (2000) ‘Societal Governance: Level, Modes and Orders of
Social–Political Interaction’, in J. Pierre (ed.), Debating Governance: Authority,
Steering and Democracy (Oxford, Oxford University Press), pp. 138–64.
Kooiman, J. (2003) Governing as Governance (London, Sage).
Koppenjan, J. (2001) ‘Design and Management of Local Interactive Processes’, in
J. Edelenbos, and R. Monnikhof (eds), Local Interactive Policy Development
(Utrecht, Lemma) (in Dutch), pp. 143–74.
Koppenjan, J. F. M., Ringeling, A. B. and Velde, R. H. A. te (eds) (1987) Beleidsvorming
in Nederland (‘s-Gravenhage, Vuga).
Krasner, S. D. (1984) ‘Approaches to the State. Alternative Conceptions and
Historical Dynamics’, Comparative Politics, 16 (3), pp. 223–46.
Krasner, S. D. (1988) ‘Sovereignty. An Institutional Perspective’, Comparative
Political Studies, 21 (1), pp. 66–94.
Laclau, E. (1990) New Reflections on the Revolution of Our Time (London, Verso).
Laclau, E. (1996a) Emancipation(s) (London, Verso).
Laclau, E. (1996b) ‘Why do Empty Signifiers Matter to Politics?’, in E. Laclau (ed.),
Emancipation(s) (London, Verso), pp. 36–46.
Laclau, E. (2001) ‘Democracy and the Question of Power’, Constellations, 8 (1),
pp. 3–14.
Lane, C. and Bachman, R. (1998) Trust Within and Between Organizations:
Conceptual Issues and Empirical Applications (Oxford, Oxford University Press).
Laumann, E. O. and Knoke, D. (1987) The Organizational State: Social Choice in
National Policy Domains (Wisconsin University of Wisconsin Press).
Lawrence, T. B., Hardy, C. and Phillips, N. (2002) ‘Institutional Effects of
Interorganizational Collaboration: The Emergence of Proto-Institutions’,
Academy of Management Journal, 45, pp. 281–90.
Le Galès, P. and Thatcher, M. (eds) (1995) Les réseaux de politique publique: Débat
autour des policy networks (Paris, L’Harmattan).
Levi, M. (1990) ‘A logic of institutional change’, in K. S. Cook, and M. Levi (eds),
Levine, S., and White, P. E. (1961) ‘Exchange as a Conceptual Framework for
330 Bibliography
the Study of Interorganizational Relationships’, Administrative Science Quarterly,

5, pp. 583–601.
Levy, A. and Merry, U. (1986) Organizational Transformation: Approaches, Strategies,
Theories (New York, Praeger).
Lindblom, C. E. (1977) Politics and Markets (New York, Basic Books).
Lindblom, C. E. and Cohen, D. K. (1979) Usable Knowledge: Social Science and
Social Problem Solving (New Haven/London, Yale University Press).
Lipsky, M. (1980) Street-level Bureaucracy. Dilemmas of the Individual in Public
Services (New York, Russell Sage Foundation).
Lowi, T. J. (1963) ‘American Business, Public Policy, Case Studies and Political
Theory’, World Politics, 16, pp. 677–715.
Lowi, T. J. (1972) ‘Four Systems of Policy, Politics and Choice’, Public
Administration Review, 32 (4), pp. 298–310.
Lowi, T. J. (1979) The End of Liberalism, 2nd edition (New York, Norton).
Lowi, T. J. (1999) ‘Frontyard Propaganda’, Boston Review. (Available at:
www.bostonreview.net/BR24.5/lowi.html).
Luhmann, N. (1975) Legitimation durch Verfahren (Darmstadt, Luchterhand).
Luhmann, N. (1981) Politische Theorie im Wohlfahrtsstaat (München, Olzog).
Luhmann, N. (1982) The Differentiation of Society (New York, Colombia University
Press).
Luhmann, N. (1984a) Soziale Systeme. Grundriss einer allgemeinen Theorie
(Frankfurt am Main, Suhrkamp).
Luhmann, N. (1984b) ‘Staat und Politik. Zur Semantik der Selbstbeschreibung
politischer Systeme’, in U. Bermbach (eds), Politische Theoriengeschichte,
Probleme einer Teildisziplin der Politischen Wissenschaft (Sonderheft) (Opladen,
Westdeutscher Verlag), pp. 99–125.
Luhmann, N. (1989) Die Wirtschaft der Gesellschaft (Frankfurt am Main, Suhrkamp).
Luhmann, N. (1990a) Essays on Self-Reference (New York, Columbia University
Press).
Luhmann, N. (1990b) Political Theory in the Welfare State (Berlin, De Gruyter).
Luhmann, N. (1990c) Die Wissenschaft der Gesellschaft (Frankfurt am Main,
Suhrkamp.
Luhmann, N. (1996) Soziale Systeme: Grundriß einer allgemeinen Theorie (Frankfurt,
Suhrkamp-Taschenbuch Wissenschaft).
Lukes, S. (1974) Power: A Radical View (Basingstoke, Macmillan).
Lynn, L. E. (1981) Managing the Public’s Business, The Job of the Government
Executive (Basic Books, New York).
Lynn, L. E. Jr. (1993) ‘Policy Achievement as a Collective Good: A Strategic
Perspective on Managing Social Programs’, in B. Bozeman (ed.), Public
Management: The State of the Art (San Francisco, Jossey-Bass), pp. 108–33.
Macpherson, C. B. (1977) The Life and Time of Liberal Democracy (Oxford, Oxford
University Press).
Mandell, M. P. (1990) Network Management: Strategic Behavior in the Public Sector, in
R. W. Gage, and M. P. Mandell (eds), Strategies for Managing Intergovernmental
Policies and Networks (New York, Praeger), pp. 20–53.
Mandell, M. P. (ed.) (2001) Getting Results Through Collaboration; Networks and
Network Structures for Public Policy and Management (Westport, Quorum Books).
Marin, B. (1987) ‘On Legitimacy and Political Deliberation’, Political Theory,
15 (3), pp. 338–68.
Bibliography 331
March, J. G. and Olsen, J. P. (1995) Democratic Governance (New York, The Free
Press).
March, J. G. and Olsen, J. P. (1989) Rediscovering Institutions. The Organizational
Basis of Politics (New York, The Free Press).
Marcussen, M. and Torfing, J. (eds) (2006) Democratic Network Governance in
Europe (Basingstoke, Palgrave Macmillan). Forthcoming.
Marin, B. and Mayntz, R. (eds) (1991) Policy Networks: Empirical Evidence and
Theoretical Considerations (Frankfurt-am-Main, Campus Verlag).
Marinetto, M. (2003) ‘Governing Beyond The Centre: A Critique of the Anglo-
Governance School’, Political Studies, 51, pp. 592–608.
Marsh, D. (1998) ‘The Utility and Future of Policy Network Analysis’, in
D. Marsh (ed.), Comparing Policy Networks (Buckingham, Open University Press),
pp. 185–97.
Marsh, D. and Rhodes, R. A. W. (eds) (1992) Policy Networks in British Government
(Oxford, Oxford University Press/Clarendon Press).
Marsh, D. and Smith, M. (2000) ‘Understanding Policy Networks: Towards a
Dialectical Approach’, Political Studies, 48, pp. 4–21.
Mastenbroek, W. F. (1987) Conflict Types and Organizational Development
(Chichester, New York, Wiley).
Mayntz, R. (1987) ‘Politische Steuerung und gesellschaftliche Steuerungsprobleme.
Anmerkungen zu einem theoretischen Paradigma’, in T. Ellwein, J. J. Hesse,
R. Mayntz, and F. W. Scharpf (eds) Jahrbuch zur Staats- und Verwaltungswissenschaft,
1 (Baden-Baden, Nomos), pp. 89–110.
Mayntz, R. (1989) ‘Political control and societal control problems: Notes on a
theoretical paradigm’, in T. Ellwein, et al. (eds), Yearbook on Government and
Public Administration, 1987/88 (Baden-Baden, Nomos), pp. 81–98.
Mayntz, R. (1993a) ‘Governing Failure and the Problem of Governability: Some
Comments on a Theoretical Paradigm’, in J. Kooiman (ed.), Modern Governance
(London, Sage), pp. 9–20.
Mayntz, R. (1993b) ‘Modernization and the Logic of Interorganizational
Networks’, pp. 3–18, in J. Child, M. Crozier, and R. Mayntz (eds), Societal change
Between Markets and Organization (Aldershot, Avebury).
Mayntz, R. (1997) Soziale Dynamik und politische Steuerung: Theoretische und
methodologische Überlegungen (Frankfurt und New York, Campus).
Mayntz, R. (1998) ‘New Challenges to Governance Theory’, Jean Monnet Chair
Paper Series, European University Institute, Florence, RSC 98/50.
Mayntz, R. (1999) New Challenges to Governance Theory, Paper (Köln, Max Planck
Institut für Gesellschaftsforschung, European Centre for Social Welfare Policy
and Research).
Mayntz, R. (2002) ‘Common Goods and Governance’, in A. Heritier (ed.),
Common Goods, Reinventing European and International Governance (Langham,
MD, Rowman & Little).
Mayntz, R. (2003) ‘New Challenges to Governance Theory’, in H. Bang (ed.),
Governance as Social and Political Communication (Manchester, Manchester
University Press).
Mayntz, R. and Marin, B. (eds) (1991) Policy Networks. Empirical Evidence and
Theoretical Considerations (Frankfurt-am-Main, Campus Verlag).
Mazmanian, D. and Sabatier, P. (1989) Implementation and Public Policy (rev. edn)
(Washington, DC, University Press of America).
332 Bibliography
McLaverty, P. (2002) Public Participation and Innovation in Community Governance

(London, Routledge).
Meier, K. J. and O’Toole, L. J. Jr. (2001) ‘Managerial Strategies and Behaviour in
Networks: A Model With Evidence from U.S. Public Education’, Journal of Public
Administration and Theory, 11 (3), pp. 271–93.
Meier, K. J. and O’Toole, L. J. Jr. (2004) ‘Conceptual Issues in Modeling and
Measuring Management and Its Impacts on Performance’, in L. E. Lynn Jr., and
P. W. Ingraham (eds), Effective Government: How Organization and Management
Affect Performance (Washington, DC, Georgetown University Press),
pp. 195–223.
Mill, J. S. (1991 [1861]) Considerations on Representative Government (New York,
Prometheus Books).
Miller, G. (2000) ‘Rational Choice and Dysfunctional Institutions’, Governance,
13 (4), pp. 534–48.
Milward, H. B. and Provan, K. G. (1998) ‘Principles for Controlling Agents: The
Political Economy of Network Structure’, Journal of Public Administration
Research and Theory, 8 (2), pp. 203–21.
Milward, H. B and Provan, K. G. (2000a) ‘Governing the Hollow State’, Journal of
Public Administration Research and Theory, 10 (2), pp. 359–79.
Milward, H. B. and Provan, K. G. (2000b) ‘How Networks are Governed’, in
C. J. Heinrich, and L. E. Lynn (eds), Governance and Performance:
New Perspectives (Washington, DC, Georgetown University Press), pp. 238–62.
Milward, H. B. and Wamsley, G. L. (1985) ‘Policy Subsystems, Networks and the
Tools of Public Management’, in K. I. Hanf, and Th. A. J. Toonen (eds), Policy
Implementation in Federal and Unitary Systems (Dordrecht, Martinus Publishers),
pp. 105–30.
Mintzberg, H. (1979) The Structuring of Organizations: A Synthesis of the Research
(Englewood Cliffs, NJ, Prentice-Hall).
Moe, T. (1990) ‘Political Institutions: The Neglected Side of the Story’, Journal of
Law, Economics and Organization, 6 (special issue), pp. 215–33.
Mouffe, C. (1992) ‘Democratic Citizenship and the Political Community’, in
C. Mouffe (ed.), Dimensions of Radical Democracy (London, Verso), pp. 225–39.
Mouffe, C. (1993) The Return of the Political (London, Verso).
Mouffe, C. (2000a) ‘Deliberative Democracy or Agonistic Pluralism’, Political
Science Series, 72, pp. 1–17.
Mouffe, C. (2000b) The Democratic Paradox (New York, Verso).
Mouffe, C. (2005) On the Political (London, Routledge).
Neyer, J. (2004) ‘Explaining the Unexpected: Efficiency and Effectiveness in
European Decision-making’, Journal of European Public Policy, 11 (1), pp. 19–38.
Nielsen, K. and Pedersen, O. K. (1988) ‘The Negotiated Economy: Ideal and
History’, Scandinavian Political Studies, 11 (2), pp. 79–101.
Nooteboom, B. (2002) Trust: Forms, Foundations, Functions, Failures and Figures
(Cheltenham, Edward Elgar Publishing).
Nowothy, H. P. S. and Gibbons, M. (2001) Rethinking Science: Knowledge and the
Public in an Age of Uncertainty (Cambridge, Polity Press).
Nye, J. S. Jr. (2002) The Paradox of American Power: Why the World’s Only Superpower
Cannot Go It Alone (New York, Oxford University Press).
Oliver, C. (1992) ‘The Antecedents of Deinstitutionalization’, Organization
Studies, 13, pp. 563–89.
Bibliography 333
Olsen, J. P. (1972) ‘Public Policy Making and Theories of Organizational Choice’,

Scandinavian Political Studies, 7, pp. 45–62.
Olson, M. (1965) The Logic of Collective Action: Public Goods and the Theory of
Groups (Cambridge, Mass., Harvard University Press).
Ostrom, E. (1986) ‘A Method of Institutional Analysis’, in F. X. Kaufman,
G. Majone, and V. Ostrom (eds), Guidance and Control in the Public Sector: The
Bielefeld Interdisciplinary project (Berlin, De Gruyter), pp. 459–75.
Ostrom, E. (1990) Governing the Commons: The Evolution of Institutions for Collective
Action (Cambridge, Cambridge University Press).
Ostrom, E. (1991) ‘Rational Choice Theory and Institutional Analysis: Toward
Complementarity’, American Political Science Review, 85 (1), pp. 237–43.
Ostrom, E. (1999) ‘Institutional Rational Choice: An Assessment of the
Institutional Analysis and Development Framework’, in P. A. Sabatier (ed.),
Theories of the Policy Process (Boulder, Colorado, Westview), pp. 35–71.
O’Toole, L. J. Jr (1988) ‘Strategies for Intergovernmental Management:
Implementing Programs in Interorganizational Networks’, International Journal
of Public Administration, 11 (4), pp. 417–41.
O’Toole, L. J. Jr (1993) ‘Multiorganizational Policy Implementation: Some
Limitations and Possibilities for Rational Choice Contributions’, in F. W. Scharpf
(ed.), Games in Hierarchies and Networks: Analytical and Empirical Approaches to
the Study of Governance Institutions (Frankfurt am Main, Boulder, Campus
Verlag/Westview Press).
O’Toole, L. J. Jr. (1996) ‘Rational Choice and the Public Management of
Interorganizational Networks’, in D. F. Kettl, and H. B. Milward (eds), The State
of Public Management (Baltimore, Johns Hopkins University Press), pp. 241–63.
O’Toole, L. J. Jr (1997) ‘Treating Networks Seriously: Practical and Research-
Based Agendas in Public Administration’, Public Administration Review, 57,
pp. 45–52.
O’Toole, L. J. Jr. (2004) ‘The Theory-Practice Issue in Policy Implementation
Research’, Public Administration, 82, pp. 309–29.
O’Toole, L. J. Jr and Meier, K. J. (1999) ‘Modeling the Impact of Public
Management: Implications of Structural Context’, Journal of Public Administration
Research and Theory, 9, pp. 505–26.
O’Toole, L. J. Jr and Meier, K. J. (2004) ‘Desperately Seeking Selznick: Cooptation
and the Dark Side of Public Management in Networks’, Public Administration
Review, 64, pp. 681–93.
O’Toole, L. J. Jr., Hanf, K. and Hupe, P. (1997) ‘Managing Implementation
Processes in Networks’, in W. J. M. Kickert, et al. (eds), Managing Complex
Networks: Strategies for the Public Sector (London, Sage).
O’Toole, L. J. Jr. and Meier, K. J. (2000) ‘Networks, Hierarchies and Public
Management: Modeling the Nonlinearities’, in C. J. Heinrich, and L. E. Lynn Jr.
(eds), Governance and Performance: New Perspectives (Washington, DC,
Georgetown University Press), pp. 263–91.
Ouchi, W. G. (1980) ‘Market, Bureaucracies, and Clans’, Administrative Science
Quarterly, 25, pp. 129–41.
Parkinson, J. (2003) ‘Legitimacy Problems in Deliberative Democracy’, Political
Studies, 51 (1), pp. 180–96.
Parson, T. (1960) Structure and Process in Modern Societies (Glencoe (ILL), Free
Press).
334 Bibliography
Parson, T. (1976) Zur Theorie sozialer Systeme (Opladen, Westdeutscher Verlag).

Pateman, C. (1970) Participation and Democratic Theory (Cambridge, Cambridge
University Press).
Pedersen, O. K. et al. (1994) Demokratiets lette tilstand (Copenhagen, Spektrum).
Perri 6, Leat, D., Seltzer, K. and Stoker, G. (2002) Towards Holistic Governance: The
New Reform Agenda (Houndmills, Basingstoke, Palgrave).
Peter, B. G., King, D. S. and Pierre, J. (2005) ‘Beyond Path-dependency: Political
Conflict in Historical Institutionalism’, Journal of Politics, 67, pp. 1275–300.
Peters, B. G. (1995) The Politics of Bureacracy (White Plains, Longman).
Peters, B. G. (1999) Institutional Theory in Political Science: The New Institutionalism
(New York, Pinter).
Peters, B. G. (2005) Institutional Theory in Political Science: The New Institutionalism,
2nd edn (London, Continuum).
Pettit, P. (1997) Republicanism: A Theory of Freedom and Government (Oxford,
Oxford University Press).
Pettit, P. (2003) ‘Deliberative Democracy, the Discursive Dilemma and Republican
Theory’, in J. S. Fishkin, S. James, and P. Laslett (eds), Debating Deliberative
Democracy (London, Blackwell Publishing), pp. 138–63.
Pharr, S. J. and Putnam, R. D. (2000) Disaffected Democracies: What’s Troubling the
Trilateral Countries (Princeton, Princeton University Press).
Pierre, J. (2000) Debating Governance. Authority, Steering, and Democracy (Oxford,
Pierre, J. and Peters, B. G. (2000) Governance, Politics and the State (Houndmills,
MacMillan Press Ltd/New York, St. Martins Press).
Pierre, J. and Peters, B. G. (2005) Governing Complex Societies: Trajectories and
Scenarios (London, Palgrave Macmillan).
Pierson, P. (1994) Dismantling the Welfare State?: Reagan, Thatcher and the Politics of
Retrenchment (Cambridge, Cambridge University Press).
Pierson, P. (1998) Not Just What, But When. Issues of Timing and Sequence in
Comparative Politics (Boston, Centre for European Studies, Harvard University).
Pierson, P. (2000a) ‘Increasing Returns, Path Dependence and the Study of
Politics’, American Political Science Review, 94 (2), pp. 251–67.
Pierson, P. (2000b) ‘The Limits of Design: Explaining Institutional Origins and
Change’, Governance, 13 (4), 475–99.
Pitkin, H. (1971 [1967]) The Concept of Representation (Berkeley, University of
California Press).
Plato (1973) Der Staat (Zürich; München, Artemis-Verlag).
Polsby, N. (1968) ‘The Institutionalization of the U.S. House of Representatives’,
American Political Science Review, 62, pp. 144–68.
Popper, K. (1985 [1967]) ‘The Rationality Principle’, in D. Miller (ed.), Popper
Selections (Princeton, Princeton University Press).
Powell, W. W. (1990) ‘Neither Market Nor Hierarchy: Network Forms of
Organization’, Research in Organizational Behavior, 12, pp. 295–336.
Powell, W. W. (1991a) ‘Neither Market nor Hierarchy: Network Forms of
Organization’, in G. Thomson et al. (eds), Markets Hierarchies and Networks: The
Coordination of Social Life (London, Sage), pp. 265–76.
Powell, W. W. (1991b) ‘Expanding the Scope of Institutional Analysis’ in
W. W. Powell, and J. DiMaggio (eds), The New Institutionalism in Organizational
Analysis (Chicago, University of Chicago Press), pp. 183–203.
Bibliography 335
Powell, W. W. and DiMaggio, J. (1983) ‘The Iron Cage Revisited: Institutional

Isomorphism and Collective Rationality in Organizational Fields’, American
Sociological Review, 48 (2), pp. 147–60.
Powell, W. W. and DiMaggio, J. (1991) The New Institutionalism in Organizational
Analysis (Chicago, University of Chicago Press).
Power, M. (1997) The Audit Society (Oxford, Oxford University Press).
Pratchett, L. (2002) ‘Local Government: From Modernisation to Consolidation’,
Parliamentary Affairs, 55 (2), pp. 331–46.
Pressman, J. L. and Wildavsky, A. (1973) Implementation: How Great Expectations in
Washington Are Dashed in Oakland (Berkeley, University of California Press).
Przeworski, A., Stokes, S. C. and Manin, B. (eds) (1999) Democracy, Accountability
and Representation (Cambridge, Cambridge University Press).
Rawls, J. (1997) ‘The Idea of Public Reason Revisited’, University of Chicago Law
Review, 94, pp. 765–807.
Rawls, J. (1999) The Law of Peoples (Cambridge, MA, Harvard University Press).
Rein, M. and Schön, D. A. (1986) ‘Frame-Reflective Policy Discourse’,
Beleidsanalyse, 15 (4), pp. 4–18.
Rein, M. and Schön, D. A. (1992) ‘Reframing Policy Discourse’, in F. Fischer, and
J. Forester (eds), The Argumentative Turn in Policy Analysis and Planning
(Durham, NC, Duke University Press), pp. 145–66.
Reinicke, W. H. (1998) Global Public Policy. Governing without Government?
(Washington, DC, Brookings).
Reinicke, W. H. and Deng, F. (2000) Critical Choices: The United Nations, Networks,
and the Future of Global Governance (Ottawa, International Development Center).
Rhodes, R. A. W. (1988) Beyond Westminster and Whitehall: The Subsectoral
Governments of Britain (London, Unwin Hyman).
Rhodes, R. A. W. (1990) ‘Policy Networks: A British Perspective’, Journal of
Theoretical Politics, 2 (3), pp. 293–317.
Rhodes, R. A. W. (1994) ‘The Hollowing Out of the State’, Political Quarterly, 65,
pp. 138–51.
Rhodes, R. A. W. (1995) ‘The Institutional Approach’, in D. Marsh and
G. Stoker (eds), Theory and Methods in Political Science (Basingstoke, Macmillan),
pp. 42–57.
Rhodes, R. A. W. (1996) ‘The New Governance: Governing without Government’,
Political Studies Association, 44 (4), pp. 651–67.
Rhodes, R. A. W. (1997) Understanding Governance: Policy Networks, Governance,
Reflexivity and Accountability (Buckingham, Open University Press).
Rhodes, R. A. W. (1997a) Understanding Governance: Policy Networks, Governance,
Reflexivity and Accountability (Buckingham/Philadelphia, Open University Press).
Rhodes, R. A. W. (1997b), ‘Forword’, in W. J. M. Kickert, E. H. Klijn and
J. F. M. Koppenjan (eds), Managing Complex Networks: Strategies for the Public
Sector (London, Sage), pp. 11–15.
Rhodes, R. A. W. (1997c) ‘It’s the Mix that Matters: From Marketization to
Diplomacy’, Australian Journal of Public Administration, 56, pp. 40–53.
Rhodes, R. A. W. (2000a) ‘Governance and Public Administration’, in J. Pierre (ed.),
Debating Governance: Authority, Steering, and Democracy (Oxford, Oxford
University Press), pp. 54–90.
Rhodes, R. A. W. (2000b) ‘The Governance Narrative: Key Findings and Lessons
from the ESRC’s Whitehall Programme’, Public Administration, 78 (2), pp. 345–64.
336 Bibliography
Rhodes, R. A. W. (ed.) (2000c) Transforming British Governance, 2 vols. (London,

Macmillan).
Rhodes, R. A. W. (2006a) ‘Policy Network Analysis’, in M. Moran, M. Rein and
R. E. Goodin (eds), The Oxford Handbook of Public Policy (Oxford, Oxford
University Press).
Rhodes, R. A. W. (2006b) ‘Networks and the Unholy Trinity’, in J. Fleming and
J. Wood (eds), Policing, Security and Democracy in a Networked Society (Sydney,
UNSW Press). Forthcoming.
Rhodes, R. A. W. (2006c) ‘Old Institutionalisms’, in R. A. W. Rhodes, S. Binder and
B. Rockman (eds), The Oxford Handbook of Political Institutions (Oxford, Oxford
University Press). Forthcoming.
Rhodes, R. A. W. and March, D. (1992) ‘New Directions in the Study of Policy
Networks’, European Journal of Political Research, 21 (1–2), pp.181–205.
Richards, D. and Smith, M. J. (2002) Governance and Public Policy in the UK
Richards, D. and Smith, M. J. (2004) ‘Interpreting the World of Political Elites’,
Public Administration, 82, pp. 777–800.
Richardson, J. (2000) ‘Government, Interest Groups and Policy Change’, Political
Studies, 48, pp. 1006–25.
Riker, W. H. and Ordeshook, P. C. (1973) An Introduction to Positive Political Theory
(Englewood Cliffs, NJ, Prentice-Hall).
Ripley, R. B. and Franklin, G. (1987) Congress, the Bureaucracy and Public Policy
(Dorsey, Homewood Ill).
Risse, T. (2000) ‘Let’s Argue. Communicative Action in International Relations’,
International Organization, 54 (1), pp. 1–39.
Risse, T. (2003) ‘Konstruktivismus, Rationalismus und die Theorie Internationaler
Beziehungen – warum empirisch nichts so heiß gegessen wird, wie es theoretisch
gekocht wurde’, in G. Hellmann, K. D. Wolf and M. Zürn (eds), Forschungsstand
und Perspektiven der Internationalen Beziehungen in Deutschland (Baden-Baden,
Nomos Verlagsgesellschaft).
Risse, T., Ropp, S. C. and Sikkink, K. (eds) (1999) The Power of Human Rights:
International Norms and Domestic Change (Cambridge, Cambridge University Press).
Rogers, D. L. and Whetten, D. A. (eds) (1982) Interorganizational Coordination:
Theory, Research, and Implementation (Ames, Iowa State University Press).
Rose, N. (1996) ‘Governing “Advanced” Liberal Democracies’, in A. Barry, T.
Osborne and N. Rose (eds), Foucault and Political Reason: Liberalism, Neo-
liberalism and Rationalities of Government (Chicago, University of Chicago
Press), pp. 37–64.
Rose, N. (1999) Powers of Freedom: Reframing Political Thought (Cambridge,
Cambridge University Press).
Rose, N. and Miller, P. (1992) ‘Political Power Beyond the State: Problematics of
Government’, British Journal of Sociology, 43 (2), pp. 172–205.
Rosenau, J. N. (1969) Linkage Politics – Essays on the Convergence of National and
International Systems (New York, Free Press).
Rosenau, J. N. (1992) ‘Governance, Order, and Change in World Politics’, in
J. N. Rosenau, and Ernst-Otto Czempiel (eds), Governance Without Government:
Order and Change in World Politics (New York, Cambridge University Press),
pp. 1–29.
Rosenau, J. N. and Czempiel, E. O. (eds) (1992) Governance Without Government:
Order and Change in World Politics (Cambridge, Cambridge University Press).
Bibliography 337
Rothstein, B. (1992) ‘Labour Market Institutions and Working Class Strength’, in

S. Steinmo, K. Thelen and F. Longstreth (eds), Structuring Politics: Historical
Institutionalism in Comparative Analysis (New York, Cambridge University Press),
pp. 35–56.
Rousseau, D., Sitkin, S. B., Burt, R. S. and Camerer, C. (1998) ‘Not So Different
After All: a Cross Discipline View of Trust’, Academy of Management Review, 23
(3), pp. 393–404.
Rousseau, J. J. (1996) Vom Gesellschaftsvertrag oder Grundlagen des politischen Rechts
(Frankfurt, Insel-Verlag).
Rowe, G. and Frewer, L. J. (2005) ‘A Typology of Public Engagement Mechanisms’,
Science, Technology & Human Values, 30 (2), pp. 251–90.
Sabatier, P. A. (1986) ‘What Can We Learn Form Implementation Research?’, in
F. X. Kaufmann, G. Majone and V. Ostrom (eds), Guidance, Control, and
Evaluation in the Public Sector (Berlin, Walter de Gruyter), pp. 313–25.
Sabatier, P. A. (1988) ‘An Advocacy Coalition Framework of Policy Change and
the Role of Policy-oriented Learning therein’, Policy Sciences, 21, pp. 129–68.
Sabatier, P. A. and Jenkins-Smith, H. C. (1993) Policy Change and Learning: An
Advocacy Coalition Approach (Boulder, Colorado, Westview Press).
Sabel, C., Fung, A. and Karkkainen, B. (1999) ‘Beyond Backyard Environmentalism:
How Communities are Quietly Refashioning Environmental Regulation’,
Boston Review. (Available at: www.bostonreview.mit.edu/BR24.5/sabel.html).
Sako, M. (1998) ‘Does Trust Improve Business Performance?, in C. Lane and R.
Bachman (eds), Trust Within and Between Organizations: Conceptual Issues and
Empirical Applications (Oxford University Press, Oxford).
Salamon, L. M. (ed.) (2002) The Tools of Government: A Guide to the New Governance
Sandel, M. J. (1996) Democracy’s Discontent – America in Search of a Public
Philosophy (Cambridge, Harvard University Press).
Saward, M. (1997) ‘In Search of the Hollow Crown’, in P. Weller, H. Bakvis and
R. A. W. Rhodes (eds), The Hollow Crown (London, MacMillan).
Sbragia, A. (2000) ‘The European Union as Coxswain: Governance by Steering’, in J.
Pierre (ed.), Debating Governance (Oxford, Oxford University Press), pp. 219–40.
Schaal, G. S. and Strecker, D. (1999) ‘Die politische Theorie der Deliberation:
Jürgen Habermas’, in A. Borodocz and G. S. Schaal (eds), Politische Theorie der
Gegenwart. Eine Einführung (Opladen, Leske Budrich).
Schaap, L. (1997) Op zoek naar prikkelende overheidssturing: Over autopoiese, zelfstur-
ing en stadsprovincie (The search for Inciting Governance. On autopoiesis,
selfgovernance and city-province) (Delft, Eburon).
Schaap, L. (2003) ‘Government or Governance in the Rotterdam Region’, in
J. Magone (ed.), Regional Institutions and Governance in the European Union
(Westport, Conn., Praeger), pp. 153–73.
Schaap, L. and Twist, M. J. W. van (1997) ‘The Dynamics of Closedness in Networks’,
in W. J. M. Kickert, E. H. Klijn and J. F. M. Koppenjan (eds), Managing Complex
Networks: Strategies for the Public Sector (London, Sage), pp. 62–78.
Scharpf, F. W. (1978) ‘Interorganizational Policy Studies: Issues, Concepts and
Perspectives’, in K. I. Hanf and F. W. Scharpf (eds), Interorganizational Policy
Making: Limits to Coordination and Central Control (London/Beverly Hills, Sage),
pp. 345–70.
Scharpf, F. W. (1988) ‘Verhandlungssysteme, Verteiligungskonflikte und
Pathologien der politische Steuerung’, in M. G. Schmidt (ed.), Staatstätigkeit.
338 Bibliography
International und historisch vergleichende Analysen (Opladen, Westdeutscher

Verlag), pp. 61–87.
Scharpf, F. W. (1992) ‘Koordination durch Verhandlungssysteme: Analytische
Konzepte und institutionelle Lösungen’, in A. Benz, F. W. Scharpf and R. Zintl
(eds), Horizontale Politikverflechtung. Zur Theorie von Verhandlungssystemen
(Frankfurt am Main/New York, Campus).
Scharpf, F. W. (1993) ‘Versuch über die Demokratie im verhandelnden Staat’, in
R. Czada and M. G. Schmidt (eds), Verhandlungsdemokratie, Interessen-
vermittlung, Regierbarket. Festschrift für Gerhard Lehmbruch (Opladen,
Westdeutscher Verlag).
Scharpf, F. W. (ed.) (1993) Games in Hierarchies and Networks: Analytical and
Empirical Approaches to the Study of Governmental Institutions (Boulder, Westview
Press//Frankfurt am Main, Campus-Verlag).
Scharpf, F. W. (1994) ‘Games Real Actors Could Play: Positive and Negative
Coordination in Embedded Negotiations’, Journal of Theoretical Politics, 6 (1),
pp. 27–53.
Scharpf, F. W. (1997) Games Real Actors Play: Actor-centered Institutionalism in Policy
Research (Boulder, Westview Press).
Scharpf, F. W. (1999) Governing in Europe – Effective and Democratic? (Oxford,
Schattschneider, E. E. (1960) The Semisovereign People: A Realist’s View of Democracy
in America (New York, Holt, Rinehart and Winston).
Scheuerman, W. E. (2006) ‘Critical Theory Beyond Habermas’, in J. S. Dryzek,
B. Honig and A. Phillips (eds), Oxford Handbook of Political Theory (Oxford,
Schlosberg, D. (1999) Environmental Justice and the New Pluralism: The Challenge of
Difference for Environmentalism (Oxford, Oxford University Press).
Schmitter, P. C. and Lehmbruch, G. (eds) (1979) Trends Toward Corporatist
Intermediation (London, Sage).
Schneider, V. (1993) ‘Networks and games in Large Technical Systems’, in
F. W. Scharpf (ed.), Games in Hierarchies and Networks: Analytical and Empirical
Approaches to the Study of Governance Institutions. (Frankfurt am Main, Campus
Verlag//Boulder, Westview Press).
Schön, D. A. and Rein, M. (1994) Frame Reflection: Toward the Resolution of
Intractable Policy Controversies (New York, Basis Books).
Schon, D. A. and Rein, M. (1996) Frame Reflection: Solving Intractable Policy
Disputes (Cambridge, MA, MIT Press).
Schumpeter, J. A. (1976 [1943]) Capitalism, Socialism and Democracy (London,
George Allen & Unwin).
Scott, J. (1991) Social Network Analysis: A Handbook (London, Sage).
Scott, W. R. (1987) Organizations: Rational, Natural and Open Systems (Englewood
Cliffs, NJ, Prentice-Hall).
Scott, W. R. (1995) Institutions and Organizations (London, Sage).
Seligson, M. A. (2002) ‘The Impact of Corruption on Regime Legitimacy:
A Comparative Study of Four Latin American Countries’, Journal of Politics, 64,
pp. 408–31.
Selznick, P. (1954) TVA and the Grass Roots (New York, Harper and Row).
Shepsle, K. (1989) ‘Studying Institutions: Lessons from the Rational Choice
Approach’, Journal of Theoretical Politics, 1 (2), pp. 131–47.
Bibliography 339
Sherlock, K. L., Kirk, E. A. and Reeves, A. D. (2004) ‘Just the Usual Suspects?:
Partnerships and Environmental Regulation’, Environment and Planning
C – Government and Policy, 22 (5), pp. 651–66.
Skelcher, C. (2004) ‘Jurisdictional Integrity, Polycentrism, and the Design of
Democratic Governance’, Governance, 18 (1), pp. 89–110.
Snidal, D. (2000) ‘The Concept of Legalization’, International Organization, 54 (3),
pp. 401–19.
Sørensen, E. (2006) ‘Metagovernance: The Changing Role of Politicians in
Processes of Democratic Governance’, American Review of Public Administration,
36, pp. 79–97.
Sørensen, E. and Torfing, J. (2000) Skanderborg på landkortet – et studie af lokale
styringsnetvœrk og politisk handlekraft (Copenhagen, Jurist- og Økonomforbundets
Forlag).
Sørensen, E. and Torfing, J. (2003) ‘Network Politics, Political Capital, and
Democracy’, International Journal of Public Administration, 26 (6), pp. 609–34.
Sørensen, E. and Torfing, J. (2005a). ‘Democratic Anchorage of Government
Networks,’ Scandinavian Political Studies, vol. 28, no. 3, pp. 195–218.
Sørensen, E. and Torfing, J. (2005b) Netvœrksstyring – fra government til governance
(Frederiksberg, Roskilde Universitetsforlag).
Sørensen, E. and Torfing, J. (2005c) ‘Network Governance and Post-liberal
Democracy’, Administrative Theory and Praxis, 27 (2), pp. 197–237.
Spencer-Brown, G. (1972) Laws of Form (New York, Juliana Press).
Sternberger, D. (1968) ‘Legitimacy’, in D. L. Sills (ed.), International Encyclopedia of
the Social Sciences (Macmillan & Free Press).
Sternberger, D. (1986) Grund und Abgrund der Macht: Über Legitimität von
Regierungen (Frankfurt, Insel verlag).
Stoker, G. (2002) ‘New Labour and Local Governance in Britain’, in J. Caulfield
and H. O. Larsen (eds), Local Government at the Millenium (Opladen, Leske
Budrich), pp. 27–45.
Stoker, G. (2004) Transforming Local Governance. (Basingtoke, Palgrave-Macmillan).
Stoker, G. (ed.) (1999) The New Management of British Local Governance (London,
Macmillan).
Streeck, W. and Schmitter, P. C. (1985) ‘Community, Market, State – and
Associations?: The Prospective Contribution of Interest Governance to Social
Order’, in W. Streeck and P. C. Schmitter (eds), Private Interest Government.
Beyond Market and State (London, Sage).
Sullivan, H. (2003) ‘Local Government Reform in Great Britain’, in N. Kersting
and A. Vetter (eds), Reforming Local Government in Europe: Closing the Gap
Between Democracy and Efficiency (Opladen, Leske Budrich), pp. 39–63.
Susskind, L. and Cruikshank, J. (1987) Breaking the Impasse: Consensual Approaches
Resolving Public Disputes (New York, Basic Books).
Teisman, G. R. (1997) Sturen via creatieve concurrentie (Nijmegen, Katholieke
Universiteit Nijmegen).
Teisman, G. R. (2000) ‘Models for Research into Decision-making Processes:
On Phases, Streams and Decision-making Rounds’, Public Administration, 78,
pp. 937–56.
Termeer, C. J. A. M. (1993) Dynamiek en Inertie rondom Mestbeleid: Een Studie
naar Veranderingsprocessen in het Varkenshouderijnetwerk (Den Haag, Vuga,
’s-Gravenhage).
340 Bibliography
Termeer, C. J. A. M. and Koppenjan, J. F. M. (1997) ‘Managing Perceptions in

Networks’, in W. J. M. Kickert, E. H. Klijn and J. F. M. Koppenjan (eds),
Managing Complex Networks, Strategies for the Public Sector (London, Sage),
pp. 79–98.
Teubner, G. C. M. (1989) Recht als autopoietisches System (Frankfurt am Main,
Suhrkamp).
Teubner, G. C. M. (1991) ‘Autopoiese and steering: How politics profit from the
normative surplus of capital’, in R. J. in t Veld, L. Schaap, C. J. A. M. Termeer
and M. J. W. van Twist (eds), Autopoiesis and Configuration Theory: New
Approaches to Societal Steering (Dordrecht/Boston/London, Kluwer Academic
Publishers), pp. 127–43.
Teubner, G. C. M. and Willke, H. (1984) ‘Kontext und Autonomie:
Gesellschaftliche Selbststeuerung durch reflexives Recht’, Zeitschrift für
Rechtssoziologie, 4 (6), pp. 4–35.
Thelen, K. and Steinmo, S. (1992) ‘Historical institutionalism in comparative
politics’, in S. Steinmo, K. Thelen and F. Longstreth (eds), Structuring
Politics: Historical Institutionalism in Comparative Analysis (New York, Cambridge
University Press), pp. 1–32.
Toke, D. (2000) ‘Policy Network Creation’, Public Administration, 78, (4), pp. 839–54.
Torfing, J. (1999) New Theories of Discourse (Oxford, Blackwell).
Torfing, J. (2001) ‘Path-dependent Danish Welfare Reforms: The Contributions of
the New Institutionalisms in Understanding Evolutionary Change’,
Scandinavian Political Studies, 24 (4), pp. 277–309.
Triantafillou, P. (2004) ‘Addressing Network Governance Through the Concepts
of Governmentality and Normalization’, Administrative Theory & Praxis, 26 (4),
pp. 489–508.
Tsebelis, G. (1990) Nested Games (Berkley, University of Carlifornia Press).
Urbinati, N. (2000) ‘Representation as Advocacy: A Study of Democratic
Deliberation’, Political Theory, 28 (6), pp.758–86.
Veld, R. J. in ‘t (ed.) (2000) Willingly and Knowingly: The Role of Knowledge About
Nature and the Environment in Policy Processes (Utrecht, Lemma).
Veld, R. J. in ‘t, Schaap, L., Termeer, C. J. A. M. and Twist, M. J. W. van (eds),
(1991) Autopoiesis and Configuration Theory: New Approaches to Societal Steering
(Dordrecht/Boston/London, Kluwer Academic Publishers).
Verweij, M. and Josling, T. E. (2003) ‘Special Issue: Deliberately Democratizing
Multilateral Organization’, Governance, 16 (1), pp. 1–21.
Villadsen, S. (2000) ‘Fra lokal statsforvaltning til kommunalt demokrati’, in
P. Bogason (ed.), Stat, forvaltning og samfund efter 1950 (Copenhagen,
DJØF-Forlag), pp. 151–92.
Walzer, M. (1989) ‘Citizenship’, in T. Ball, J. Farr and R. L. Hanson (eds), Political
Innovation and Conceptual Change (Cambridge, Cambridge University Press).
Weber, M. (1946) From Max Weber: Essays in Sociology. Trans. and (eds), H. H. Gerth
and C. W. Mills (Oxford, Oxford University Press).
Weber, M. (1966) ‘Die drei reinen Typen der legitimen Herrschaft’, in
J. Winckelmann (ed.), Staatssoziologie (Berlin).
Weber, M. (1976) Wirtschaft und Gesellschaft. Grundriß der verstehenden Soziologie
(Tübingen, Mohr).
Weick, K. E. (1979) The Social Psychology of Organising (2nd edn) (New York).
Weimer, D. L. (1995) ‘Institutional Design. Overview’, in D. L. Weimer (ed.),
Institutional Design (Boston/Dordrecht/London, Kluwer Academic Publishers).
Bibliography 341
Whitehouse, L. (2006) ‘Corporate Social Responsibility: Views From the

Frontline’, Journal of Business Ethics, 63 (3), pp. 279–96.
Wildavsky, A. (1979) Speaking Truth to Power. The Art and Craft of Policy Analysis
(Boston, Little, Brown).
Wildavsky, A. (1987) ‘Choosing Preferences by Constructing Institutions:
A Cultural Theory of Preference Formation’, American Political Science Review, 81
(1), pp. 3–31.
Wilks, S. and Wright, M. (1987) Comparative Government Industry Relations
(Oxford, Clarendon Press).
Williamson, O. E. (1979) ‘Transaction Cost Economics: The Governance of
Contractual Relations’, Journal of Law and Economics, 22, pp. 233–61.
Williamson, O. E. (1996) The Mechanisms of Governance (New York/Oxford,
Willke, H. (1987a) ‘Entzauberung des Staates. Grundlinien einer
Systemtheoretischen Argumentation’, in T. Ellwein, J. J. Hesse, R. Mayntz and
F. W. Scharpf (eds), Jahrbuch zur Staats- und Verwaltungswissenschaft, 1
(Baden-Baden, Nomos), pp. 285–308.
Willke, H. (1987b) ‘Kontextsteuerung durch Recht? Zur Steuerungsfunktion des
Rechts in polyzentrischer Gesellschaft’, in M. Glagow and H. Willke (eds),
Dezentrale Gesellschaftssteuerung. Probleme der Integration polyzentrischer
Gesellschaft (Pfaffenweiler, Centaurus-Verlagsgesellschaft), pp. 3–27.
Willke, H. (1991) ‘Regieren als die Kunst systemischer Intervention’, in
H. H. Hartwich and G. Wewer (eds), Regieren in der Bundesrepublik 3.
Systemsteuerung und ‘Staatskunst’ (Opladen, Leske Budrich), pp. 35–52.
Willke, H. (1997) Supervision des Staates (Frankfurt am Main, Suhrkamp).
Willke, H. (1998) Systemtheorie III – Steuerungstheorie (Stuttgart, Opladen, Lucius &
Lucius).
Winter, S. (2004) ‘The Implementation Perspective’, in B. G. Peters and J. Pierre,
(eds), Handbook of Public Administration (London, Sage).
Wittgenstein, L. (1972) Philosophical Investigations. Translated by G. Anscombe
(Oxford: Basil Blackwell).
Wolf, K. D. (2000) Die neue Staatsräson – Zwischenstaatliche Kooperation als
Demokratieproblem in der Weltgesellschaft (Baden-Baden, Nomos).
Wolf, K. D. (2002) ‘Contextualizing Normative Standards for Legitimate
Governance Beyond the State’, in J. R. Grote and B. Gbikpi (eds), Participatory
Governance (Lseke Budrich, Opladen).
Wynne, B. (1989) ‘Frameworks of Rationality in Risk Management. Towards the
Testing of Naive Sociology’, in J. Brown (ed.), Environmental Threats: Perception
Analysis, and Management (London, Belhaven), pp. 33–47.
Young, I. M. (2000) Inclusion and Democracy (New York/Oxford, Oxford University
Press).
Young, O. R. (1994) International Governance: Protecting the Environment in a
Stateless Society (Ithaca, NY, Cornell University Press).
Zafonte, M. and Sabatier, P. A. (1998) ‘Shared Beliefs and Imposed
Interdependencies as Determinants of Ally Networks in Overlapping
Subsystems’, Journal of Theoretical Politics, 10 (4), pp. 473–505.
Zintl, R. (1993) ‘Comment on O’Toole and Scharpf: Networks as a Challenge for
Theory’, in F. W. Scharpf (ed.), Games in Hierarchies and Networks: Analytical and
Empirical Approaches to the Study of Governance Institutions (Frankfurt am Main,
Campus Verlag/Colorado, Westview Press).
342 Bibliography
Zouridis, S. (2003) ‘A Quest for Practical Theory: Theory and Interaction Research
in a Dutch City’, Administrative Theory & Praxis, vol. 25, no. 3, pp. 351–70.
Zucker, L. G. (1988) ‘Where do Institutional Patterns Come From?: Organizations
as Actors in Social Systems’, in L. G. Zucker (ed.), Institutional Patterns and
Organizations: Culture and Environment (Cambridge, Ballinger).
Zürn, M. (2000) ‘Democratic Governance Beyond the Nation-State’, in
M. Th. Greven and L. Pauly (eds), Democracy Beyond the State: The European
Dilemma and the Emergning Global Order (Toronto, University of Toronto Press).
Index
access, to networks, 255–7 asymmetrical allocations, 9

accountability attenuated influence, 217
administrative, 279 authenticity, 270
as a democratic norm, 276–8 autonomy
holders and holdees, 282–4 of functionally delimited
and inclusion, 278–82 groups, 275
and information, 290 governing through formation of,
political, 279 190–4
and publicity, 284–6 autopoietic social systems, 111, 114,
and responsiveness, 290–2 115, 122
and sanctions, 290–1 autopoietic systems theory, 113–17
activation, networks, 204 Axelrod, R.A., 29, 159, 160, 206
activism, 188
actor calculation mechanism, 50–1 Bache, I., 192
adaptation/adaptiveness, 105–6 Bachman, R., 206
adaptive process management, 202 Bachrach, P., 136, 142
Adler, E., 72 ‘backward mapping’, 5
administrative accountability, Bang, H., 3
279, 291 Baratz, M.S., 136, 142
advanced liberal government, 6, 107, Barber, B., 247, 248, 249
184–5 Bardach, E., 67
definition of, 186 bargaining, 155
and governmentality, 185–7 Barns, M., 133, 134
advocacy coalitions, 144 Bates, R.H. et al., 45, 46, 49
affectedness, 280, 281 Behn, R.D., 279, 284, 285, 291
‘agentification’, 31, 43 Bengtsson, B., 45
agonistic democracy, 243–5, 266–8 Benhabib, S., 268
Agranoff, R.I., 84, 133, 146, 147, 199 Bennet, W.L., 289
Aldrich, H.E., 4, 121, 139 Benson, J.K., 4, 21, 48, 139, 140
Alford, R.R., 36 Benz, A., 153
Allison, G.T., 144 Bevir, M., 67, 77, 80, 82, 90
Alter, C., 212 Bevir, M. et al., 85
anarchy, 157 bias, mobilization of, 136
Anglo-governance school, 77, 79 Blom-Hansen, J., 44, 138
anthropological network theory, 313 Blumberg, B.F., 51
apathy, 248 Bogason, P., 5, 61, 112
arenas, 144, 145, 148 Bohman, J., 265
arguing, 156 Boli, J., 162
Aristoteles, 157 Borell, K., 191
Aristotle, 242 Börzel, T.A., 54, 138, 153, 154, 155,
associations, 237–8 158, 163
associative democracy, 237–8, 280 bottom-up network formation, 36, 45
assurance problem, 53–4 bounded rationality, 28
343
344 Index
Bressers, H. et al., 221, 222, 223, 224 cognitive closedness, 99

brokerage, 126, 128 cognitive closure, 118, 119
Brown, L.D., 137, 147, 149, 150 Cohen, D.K., 205
Brühl, W. et al., 153, 161 Cohen, J., 268
Bruijn, J.A. de et al., 201, 202 Cohen, M.D., 5
Bueren, E. van, 205, 213 Coleman, J.S., 46, 48, 57
Burchell, G., 187 collaboration, 227
Burns, T.R., 147, 207 collective action, 201
Butler, C.T., 135, 137 collective action problems, 34
collective behaviour, 242
Calvert, R.L., 29 collectively negotiated decisions, 12
Campbell, J.L., 28, 30, 41 ‘common good’, 248, 249–51
capacities, of network actors, 37 and governance networks,
Carlsson, L., 51 250–1
Castells, M., 9 communication, 114, 115, 122,
Cawson, A., 139 266–8, 286
central planning, 5 community democracy, 241–3
Chalmers, J., 189 competitive democracy, 236–9, 275
Chambers, S., 266 complexity, 5
change compromise formation, 100
managing in networks, 83–7 conflict, 27, 28, 33, 133, 222, 240,
in networks, 81–2 243, 305
Chan, J., 87 at institutional level, 149
Chayes, A., 159 at interaction level, 149
Chayes, A.H., 159 excessive, 150
Checkel, J.T., 155 implications of level within
Chiles, T.H., 206 networks, 151
China, ‘harmonious society’, 136 meaning of, 135
civil society, 235 negative and positive functions,
Clegg, S.R., 30 137–8
closedness, of governance networks, ‘conflict avoidance’, 53
99–100, 301 conflict reduction, 146, 170
closed networks, governing, 124–5 conflict resolution, 147
closed policy communication systems, Connolly, W.E., 243, 244, 250, 253
governing, 129–31 consensus, 254, 305
closure at institutional level, 149
cognitive, 118, 119 concept of, 133
explanations for, 120–3 drawbacks, 136
and ‘frames of reference’, 121–2 excessive, 150
and governance, 111–32 and group decision making, 136
and government, 112 implications of level within
and interdependence, 121 networks, 151
of networks, 118 insufficient consensus in policy
relations between explanations for, networks, 143–7
123–4 meaning of, 135
social, 118, 119, 120 and minorities, 136–7
types of, 118–21, 305 policy networks, and the
within networks, 118 overproduction of consensus,
Cobb, R.W., 135, 138, 142 141–3
Index 345
consensus – continued decentred theory, 77, 80–1, 88,

positive and negative functions, 90, 301
135–7 change in networks, 81–2
role and policy networks, 140–1 decision making, governance
surplus in policy networks, 138–40 networks, 12
consensus building, 134, 146 decisions
conservatism, 113, 118 collectively negotiated, 12
constitution, 101 documenting, 285, 287–8
content ‘degovernmentalization’, state, 3
managing, 205–4 deinstitutionalization, 209, 301
in process design, 202 and institutionalization, 62–4
context steering, 130 pressures for, 66
‘contextual interaction theory’, 221 Deitelhoff, N., 164
contextual mechanism of governance Deleuze, G., 19
network formation, 47–50 deliberate accountability, 281
mutual dependencies, 48–9 deliberative democracy, 264–5, 268,
perceptions, 49 280–1, 286, 293
resource and strategic dimension of democracy, 164
interdependencies, 47–8 aggregative theories of, 235
symmetrical and asymmetrical associative model, 237–8
interdependencies, 49–50 contribution of governance
contingency approach to networks, 271–3
management, 134 definition of, 234
Cook, 289 impact of governance networks,
cooperation, 52, 201, 217–18 233–4, 308–10
problems, 44 learning, 251–2
and trust, 206 liberal theories of, 234–5
coordination, 74, 117, 200 models of, 269–71
coordination games, 222, 224 ‘democracy of the affected’, 268
corporate social responsibility, democratic accountability, network
194, 315 governance, 274–96, 309–10
corporatist arrangements, 153 ‘democratic anchorage’, 15
Coser, L.A., 135, 137, 149 democratic elite theory, 280, 292
cosmopolitan democracy, 265 democratic ethos, 252, 253
Covaleski, M.A., 69 democratic legitimacy, 4
critical theory, 265–6 democratic theory, 300
Crozier, M., 137, 144 communicative aspect and
Cruikshank, B., 314 networks, 266–8
Cruikshank, J., 135 democratization, 270
Cunningham, F., 247, 248 Deng, F., 161, 162
Cutler, C.A. et al., 153, 156 Denmark
Czempiel, E.O., 155 Local Coordination Committees on
Preventive Labour Market
Damgaard, B., 191, 254 Measures, 192
Davies, A., 87 municipal reform, 191
Davis, G., 189 Denters, B.O. et al., 62
Deakin, S., 206 Dery, D., 133
Dean, M., 6, 17, 38, 106, 107, 178, DiMaggio, J., 15, 17, 29, 36, 104
179, 194, 314 Dimaggio, P., 207
346 Index
Dinsmith, M.W., 69 Esping, H., 191

‘disconnecting’ games, 226 ethnography, 82
discursive democracy, 271–2, 302 Etzioni-Halevy, E., 237, 239,
discursive structures, 312 280, 292
and storylines, 312 EU, 4
documentation procedures, 288–9 Employment Guidelines, 2
Donzelot, J., 187 Europe, 139–40
Dowding, K., 138, 184 European Employment Strategy (EES),
Dryzek, J.S., 164, 268, 270, 271 195
Dunleavy, P., 29 Evan, W.M., 3
Duverger, M., 27
Dyrberg, T.B., 30 face-to-face communications, 117
failure
Easton, D., 157 in governance, 171–2
Eckersely, R., 268 governance networks, 15, 102–3,
economic governance, 108 108–9, 301, 304–6
Edelenbos, J., 201, 202, 205, 206, 209, Fairclough, N., 289
211, 212, 213 feminism, 270–1
Eeten, M. van, 146, 147 Ferejohn, A., 45, 290
effective governance, 97, 98 Finer, H., 27
effectiveness, 157–9 Fisher, R., 147
demand for, 156–7 Flam, H., 147, 207
and legitimacy, 163–5, 305–6 Fleming, J., 85
effective network governance, 98 flexible policy solutions, 97–8
‘efficiency of history’, 37, 105 Florini, A., 161, 162
Ehrmann, J.R., 205 Forester, J., 147, 211
Eijk, C. van der, 142 formal organizations, 117
Eising, R., 153 Forrester, 204
Elder, C.D., 135, 138, 142 Foucault, M., 19, 30, 38, 39, 40,
electoral mandates, 292 41, 106, 178, 179, 185,
electoral representation, 277–8, 279 192, 301
elites, 237, 239 fragmentation, 5, 44, 79, 102
Elster, J., 44, 45, 50, 53, 56, 135 ‘frames of reference’, 121–2, 123
Employment Guidelines, European governing closed frames of
Union, 2 reference, 128–9
employment policy, 195–6 Frances, J., 79
‘Empowered Participatory franchise, 270
Governance’ (EPG), 239 Franklin, G., 134, 138
empowerment, 188 freedom
endogenous evolution, 44 concepts of, 264
‘energy’, 65 political, 309
Entmann, R., 289 Freeman, J.L., 138
environmental justice movement, 267 free riding, 34, 52–3, 303
‘epistemic communities’, 72 Frewer, L.J., 188
equality, 254–5 Friedberg, E., 137, 144
and governance networks, 257–8 Friedland, R., 36
political equality, 263 Friend, J.K. et al., 147, 200, 201
within networks, 257 functional democracy, 292
Esmark, A., 196, 117, 280 functional differentiation, 115
Index 347
functionalist inclusion principle, 280 governance network failure, 5,

functionally delimited groups, 95–110, 304–6
autonomy, 275 and governability theory, 102–3,
functional representation, 275 108–9, 301
Fung, A., 239, 240, 241 and governmentality theory, 106–8
and integration theory, 104–6
Gage, R.W., 200, 201, 202, 203, and interdependency theory,
211, 212 98–102
Galtung, F., 161 governance network formation,
‘game management’, 125, 127, 128 43–60, 183–98
game mechanism of governance actor calculation mechanism, 50–1
network formation, 51–6 contextual mechanism, 47–50
assurance problem, 53–4 game mechanism, 51–6
free riding problem, 52–3 mechanisms, 46–7
generosity problem, 54–6 governance network research, 3–7,
problem of continuous 14–20
cooperation, 51 first generation, 14
games, 89 second generation, 14, 16, 43
‘disconnecting’, 226 governance networks, 2–3, 262–3,
linking and segmenting games, 297–9
225–6 analysis of change in, 82
game structuring, 173 autonomy, 190–1
game theory, 221–2, 221–3, 300 benefits of, 216–17
Gay, P. du, 279 closed nature of, 111
Geddes, B., 61 closedness of, 99–100
Geertz, C., 82 and closure, 120–1
generalized trust, 51 and common good, 250–1
generosity problem, 54–6 conditions for success and failure, 7
Gibbons, M. et al., 141 as continuous cooperation, 52
Giddens, A., 208 contribution to democracy, 271–3
Glasbergen, P., 133, 147 decentred theory, 80–1
goal-achieving strategies, 204–5 decision making, 12
Gordon, C., 106 defining, 8–11
Goss, S., 85, 169, 233 definitions of, 9, 102, 104
governability, definition of, 103 democratic implications, 8, 308–10
governability theory, 17, 18, 301 democratic legitimacy, 15–16, 247
and governance network failure, democratic potential, 284
102–3, 108–9, 301 dependence of actors, 9
and metagovernance, 172–5 drawbacks of, 216
governance dynamics of, 7, 303–4
and closure, 111–32 effectiveness, 154, 156, 158
definitions of, 77, 88 and equality, 257–8
failure, 171–2 formation, 183–98
in issue networks, 146–7 governing outputs and outcomes,
and network management, 215–30
199–200 impact on democracy, 233–4
networks as, 154–6 impact of political cultures and
notion of, 8–9 traditions, 311
openness, 112 increased legitimacy, 69
348 Index
governance networks – continued governing

institutional ambiguity, 26 outputs and outcomes of
institutionalized framework, 10 governance networks, 215–30
interdependence, 121, 191 performance of networks, 194–6
interdependency, 9–10 through formation of autonomy
internal cohesion and external and interdependencies, 190–4
boundaries, 311 governing processes, 3
and learning democracy, 254 government
legitimacy, 153 art of, 38–9
and liberal democracy, 234–7 and closure, 112
linkage of internal and external Foucault on, 185–6, 192
factors, 69 governmentality
management of the consensus- and advanced liberal government,
conflict dimension, 149 185–7
metagovernance, 306–8 definition of, 106
mobilization, 183–98 Foucault on, 186
negotiations, 10 governmentality theory, 17, 19
operationally autonomous and governance network failure,
actors, 9 106–8, 109–10
optimization of functioning, 13 and metagovernance, 178–80
outcomes, 108–9, 225–6 government failure, 95
and participation, 247–61 Guattari, F., 19
path dependent transformation, guiding strategies, 204–5
32–3 Gutmann, A., 281, 286, 294
patterns of relationships in, 61
as a pluricentric governance system, Haahr, J.H., 195
11–12 Habermas, J., 265, 266, 284, 286
policy making in, 5 Hage, J., 212
policy solutions, 97–8, 99 Hajer, M., 3, 26, 40, 49, 101, 141, 312
positive approach, 78–9 Hall, P.A., 28, 33, 41, 140
and postliberal democracy, 236–45 Hall, T.E., 216
production of public purpose, 11 Hammersley, M., 82
self-regulation, 10 Hanf, K., 48, 113, 147, 199, 201,
social exclusion, 119 203, 213
and social systems, 117–18 Hansen et al., 254
as a solution, 153 Hardin, G., 160
and state, 173–4 Harding, A., 192
as a synthesis of state and Hardin, R., 160
market, 11 Hart, P., 136
as a threat to democracy, 243 Häusler, J. et al., 56
and traditional governance, Hay, C., 43, 44, 45
298–9 Healey, P., 211
governance processes Heclo, H., 3, 4, 82, 143, 145
hierarchical coordination, 155 Hedström, P., 46
non-hierarchical coordination Heffen, O. van, 169
(steering), 155 Heffen, O.V. et al., 3
governance structures, and Heisler, M.O., 139
processes, 156 Held, D., 265, 277
‘governance turn’, 153 Héritier, A., 155, 157, 159
Index 349
Hermansson, J., 52, 53 institutional dissociation, 210

Hernes, G., 49 institutionalization
Hertting, N., 47, 53, 56, 57 concepts of, 63
hierarchical regulation, 172 and deinstitutionalization, 62–4:
hierarchies, 154, 155, 157, 299, 311 characteristics of members,
Hirst, P., 3, 237, 238, 239, 280 71–3; functional factors,
historical institutionalism, 28, 31–3, 69–70; operating environment,
41–3, 64, 300–301 73–4; political factors,
Hjern, B., 5, 43, 73, 139, 221 66–9; social pressures, 70–71;
Hodgson, G.M., 32 tasks, 74
Hodson, D., 195 rational choice conceptions, 63–4
Holden, B., 164, 234 institutionalized discourses of
‘holistic governance’, 32 governance, tactical polyvalence
Holzinger, K., 165 of, 40–1
homelessness policy, 59n6 institutionalized framework, 10, 25–7
Hoppe, R., 141, 144 institutional tensions, 209–10
horizontal coordination, 33 institutional theory, 27, 300
horizontal networks, 3, 34 institutions, 31
Hösli, M., 154 inertia in, 65–6
Howarth, D., 312 need for, 50
Hull, C., 5, 43, 221 and power, 30
human interaction, 117 integration theory, 17, 18–19
human rights, 158, 161 and governance network failure,
Huntington, S.P., 63 104–6, 109
Hurd, I., 159 and metagoverance, 175–7
interaction, 117
identities, changing, 252–4 structures of, 312
identity, of network actors, 37 interaction processes, 150
implementation resistance, 13 interactive policy development,
incentive and opportunity structure, 47 209–10, 211
inciting governance, 130 interdependencies, governing through
inclusion, 275 formation of, 190–4
and accountability, 278–82 interdependency, 9–10, 47–8, 121
in networks, 255–7 interdependency theory, 17, 18,
‘inclusion of the affected’, 255–6 98–102
inertia, in institutions, 65–6 and governance network failure,
informality, 51, 56, 58 98–102, 108–9
informal networks, 58 and metagovernance, 170–2,
information, as a resource, 48 180–1
informed consent, 285 interest formation, 188
INGOs (international non- interest groups, 139, 145
governmental organizations), 162 interference (structural coupling
input legitimacy, 306 between systems), 116
instability, 303 ‘intermediary groups’, 6
institutional ambiguity, governance interorganizational negotiation, 4
networks, 26 interorganizational relations, 3, 140
institutional design, 206–7, 210–11 inter-organization theory, 47, 139
definition of, 207–8 interpenetration, 116
types of strategy for, 208–9 intersystem relations, 116
350 Index
‘iron triangles’, 4, 138, 139, 145 Kok, W.J.P., 142

issue networks, 49, 112, 143, 144–6 Kooiman, J., 3, 5, 13, 14, 32, 33, 102,
governance in, 146–7 103, 113, 153, 155, 156, 157, 169,
172, 173, 174, 175, 184, 233
Jachtenfuchs, M., 153 Koppen, 116
Janis, I.L., 136 Koppenjan, J.F.M., 5, 10, 13, 15, 51,
Jenkins-Smith, H.C., 82, 146 96, 99, 100, 125, 126, 127, 128,
Jessop, B., 1, 5, 9, 11, 12, 14, 15, 87, 139, 141, 145, 146, 148, 170, 171,
96, 97, 98, 169, 171, 190, 233 184, 191, 199, 200, 201, 202, 204,
Johansson, R., 191 208, 213
John, P., 112 Koppenjan, J.F.M. et al., 145
Johnson, J.D., 45, 46 Krasner, S.D., 32, 38
Johnson, N., 27
joint knowledge production, 205–6 Laclau, E., 30, 39, 40, 250, 251
Jong, M. de, 211 laissez faire liberalism, 106
Jordan, G., 119, 134, 140 Lane, C., 206
Josling, T.E., 209 Laumann, E.O., 139, 143
journalists, 289 Lawrence, T.B. et al., 209
learning, 105, 299
Kahn, R.L., 65 learning democracy, 251–2
Kaiser, K., 162 and governance networks, 254
Katz, D., 65 legal and financial methods, of
Kaufmann, A., 247 network formation, 191–2
Kaufmann, F.X. et al., 199 Le Galès, P., 157
Keck, M., 161 legitimacy, 159–63
Kenis, P., 143, 157 demand for, 156–7
Keohane, R., 162 and effectiveness, 163–5
Kersbergen, K.V., 9, 11 of network governance, 153
Kickert, W.J.M., 3, 13, 15, 51, 62, 125, Lehmbruch, G., 139
126, 127, 170, 171, 184, 202, 203 Leroy, 134
Kickert, W.J.M. et al., 121, 133, 146, Levi, M., 51
169, 170, 171, 191, 199, 200, 233 Levine, S., 139
Kingdon, J.W., 143, 212 Levy, A., 129
King, L.A., 209 liberal democracy, and governance
Kiser, L.L., 121, 219 networks, 234–7
Kiss, G., 114 Lindblom, C.E., 154, 205
Klijn, E.H., 4, 5, 10, 13, 15, 47, 48, 51, linkages, 116
96, 99, 100, 140, 141, 145, 146, linking and segmenting games,
147, 148, 170, 171, 191, 199, 200, 225–6
201, 202, 203, 204, 206, 207, 208, Lipsky, M., 5
212, 213 listening, to others, 254
Klok, P.J., 221 local governance, 112
Kneer, G., 116 local public sector, 31–2
Knoke, D., 139, 143 Lowi, T.J., 145, 161, 216, 262
‘knowledge disputes’, 205 Luhmann, N., 111, 114, 115, 116,
knowledge production, 141, 144 124, 129, 160
Knox, H. et al., 312, 313 Lukes, S., 314
Koch, H., 247, 249 Lynn, L.E., 204
Kohler-Koch, B., 153, 154 Lynn, L.E. Jr, 222
Index 351
McGuire, M., 199, 203, 212 hands-off, through institutional

McMackin, J., 206 design, 173
Macpherson, C.B., 234, 247, 252, 277 of identities, 175–6
macro-micro-macro schedule, 57 and integration theory, 175–7
Maher, I., 195 and interdependency theory, 170–2
‘management by chaos’, 127 legislative level, 220
management ethics, 194 levels of, 218–21
management reforms, police service, of network actors’ capacities, 176
85–7 as network management, 199–214
mandates, 293–4 by public authorities, 226–8
Mandell, M.P., 133, 146, 147, 199, and state, 174
200, 201, 202, 203, 211, 212, 213 theoretical approaches, 169–82
Manin, B., 268, 277 via policy formulation, 223–4
March, J.G., 3, 4, 10, 13, 15, 17, 29, methodological challenge, 310
35, 36, 37, 38, 41, 49, 104, 105, methodological individualism, 46
112, 113, 118, 119, 134, 175, 176, micro-level theory of network
177, 188, 190, 207, 234, 242, 274, formation, 35
285, 289, 290 Miller, G., 59
Marin, B., 3, 9, 14, 134, 143, 153, 156, Miller, P., 6
157, 169 Mill, J.S., 252
Marinetto, M., 78, 184 Milward, H.B., 15, 69, 139, 143, 169,
marketization, 78, 79, 95 199, 206
marketization strategy, 1 minority protection, 235
markets, 154, 155, 156, 157, Mintzberg, H., 4
299, 311 mobilization of
private, 299 agency, 187–9
quasi-markets, 299 bias, 136, 142
Marsh, D., 4, 81, 82, 134, 143 mode of coordination, 4
Mastenbroek, W.F., 137, 149 monitoring, 288
Mayntz, R., 3, 4, 5, 9, 11, 12, moral constituency, 281
13, 14, 15, 102, 104, 113, Mouffe, C., 243, 244, 250, 266, 314
134, 143, 153, 155, 156, multi-actor and multi-purpose games,
157, 169, 172, 173, 174, 143–4
184, 191, 234 Multilateral Agreement on Investment
Mazmanian, D., 221 (MAI), 158
Meier, K.J., 169, 213, 216, 217, 218 multilevel governance networks, 6
members, characteristics of, 71–3 multilevel governmental systems, 220
Merry, U., 129 multi-theoretical approach, network
metagovernance, 7, 15, 44, 96, 101, governance, 299–303
106, 199, 221–3
advanced liberal metagovernance, Nassehi, A., 116
178 National Action Plans on
as constitutional decision Employment (NAPs), 2, 195–6
making, 219 national bridge-building project, 2
differences between theories, 180–1 negotiated governance, 2
and governability theory, 172–5 negotiation, 100
of governance networks, 306–8 governance networks, 10
and governmentality theory, neocorporatism, and policy networks,
178–80 139–40
352 Index
neo-institutionalist theory, 188 implication of level of consensus

Netherlands, 136, 226 and conflict within, 151
interactive policy development, 209 managing change in, 83–7
network literature, 63 notion of, 9
network, definition of, 74 as representative institutions, 275
network composition, 208 and rules, 64
network dynamics, 77, 303–4 selection of members, 294
network formation success in, 83
bottom-up explanation, 36 and systems theory, 117
through legal and financial tools for, 84
methods, 191–2 ‘network structuring’, 125, 127, 128
through norms, 192–4 network transformation, 304
network governance, 89 new institutionalism, 27–8
democratic accountability, 274–96 New Public Management (NPM), 9,
multi-theoretical approach, 31, 43, 70
299–303 Neyer, J., 157
political character, 313–14 NGOs (non-governmental
network interactions, 209 organizations), 158, 162
setting up and facilitating, 201–6 Nielsen, K., 12
network literature, Netherlands, 63 non-hierarchical coordination
network management, 15, 84–5, 169, (steering), 155
306–7 non-hierarchical forms of
analytical skills, 212 governance, 3
choice and implementation of Nooteboom, B., 100, 206
strategies, 212 normative institutionalism see social
design for network interactions, constructivist institutionalism
212 norms, network formation through,
and governance, 199–200 192–4
metagovernance as, 199–214 Noworthy, H.P.S. et al., 141
skills and competencies, 211–12 Nye, J.S. Jr., 162, 272
strategies, 213–14
substantive knowledge, 212 Oliver, C., 63, 65, 69
network outcomes, 208 Olsen, J.P., 3, 10, 13, 15, 17, 29, 35,
network participation, 170 36, 37, 38, 41, 104, 105, 175, 176,
networks, 154, 155 177, 188, 190, 207, 234, 242, 274,
access to/inclusion in, 255–7 285, 289, 290
activation, 204 Olson, M., 52, 142, 145, 146, 201
application of democratic openness, in process design, 201
principles, 263–4 operating environment, 73–4
change in, 81–2 operationally autonomous actors, 9
and the communicative aspect of OPOV (one person, one vote)
democratic theory, 266–8 principle, 277
concept of, 138 opportunistic action, 34, 100
equality of members, 72–3 ‘organizational field’, definition of, 104
equality within, 257 organizational resources, 203
failure in, 83–4 organizations, as open systems, 4
as governance, 154–6 organization theory, 4
governing performance of, 194–6 Ostrom, E., 29, 45, 52, 64, 72, 121,
heterogeneous, 72 147, 207, 208, 219, 220
Index 353
O’Toole, L.J. Jr, 45, 48, 49, 57, 133, policy implementation, 200, 221
139, 146, 147, 169, 203, 213, 215, policy making, in governance
216, 217, 218, 222, 223, 224 networks, 5
outcome democracy, 239–41 policy networks, 4, 64, 78, 144
output legitimacy, 306 factors for success and failure, 65
outputs and outcomes of governance insufficient consensus, 143–7
networks, 215–30 and neocorporatism, 139–40
and the overproduction of
Parkinson, J., 268 consensus, 141–3
Parson, T., 157 and the role of consensus, 140–1
participation, 188 surplus of consensus, 138–40
and governance networks, 247–61 policy paradigms, 140
‘participatory’ tradition, 247–8, 302 policy problems, 12–13
Pateman, C., 247, 248, 249, framework for, 101
251–2, 255 policy solutions, 99
path dependent development, 28 political accountability, 279
path dependent transformation, 32–3 political cultures and traditions,
Pedersen, O.K., 12, 30 impact on governance
Pedersen, O.K. et al., 259 networks, 311
Perri et al., 85 political equality, 263, 270, 308
Peters, B.G., 3, 15, 27, 28, 44, 62, 112, political freedom, 309
169, 233, 234, 279, 290 political theory, 4
Pettit, P., 264 politics, 162, 163
Pharr, S.J., 71 polity, 163
Pierre, J., 3, 14, 15, 44, 112, 169, 234 polity reform, 165
Pierson, P., 28, 32, 64 Polsby, N., 63
Plato, 157 Popper, K., 45
pluralism, 250 popular control, 263, 269, 270, 308
plurality, in autonomous network Porter, D.O., 73, 139
actors, 105 positivist approaches, 44, 77, 78–9,
pluricentric governance systems, 80, 81
11–12 postliberal democracy, 275
pluricentric mode of coordination, 4 and governance networks, 236–45
police service, management reforms, similarities and differences between
85–7 theories, 245–6
policy, definition of, 3 poststructuralism, 29–30
policy analysis, 4–5 poststructuralist institutionalism,
policy communication systems, 38–41
117–18 differences from social
governing closed policy constructivist (or normative)
communication systems, institutionalism, 41
129–31 Powell, W.W., 15, 17, 29, 36, 79, 104,
policy communities, 49, 112–13, 118, 154, 207
139–40 power, 39, 133, 197
closure in, 119 governmentalization, 179
policy core, 140 and institutions, 30
policy discourse, 141, 312 soft, 272–3
policy formulation, 225 Power, M., 276
policy games, 148 power structures, 311, 314
354 Index
Pratchett, L., 131 Rein, M., 67, 101, 119, 121, 140, 146
preferences, 227 relationship patterns, 61
Pressman, J.L., 55, 144 representation, 275
‘prisoners-dilemma’, 34, 53, 223, 227 representation through deliberation,
privatization, 298 275, 280–1, 286
proactive governance, 98 representative institutions, networks
process design as, 275
content in, 202 resource dependencies, 47, 68, 139
progress in, 202 resource pooling, 100
process management, 170, 200, 201–6 resources, 313
adaptive process management, 202 responsiveness, 276
strategies, 203 and accountability, 290–92
‘proto-institutions’, 209 adequate, 293–5
Provan, K.G., 15, 69, 206 Rhodes, R.A.W., 3, 4, 9, 11, 14, 15, 27,
Przeworski, A. et al., 279 31, 43, 44, 45, 49, 67, 68, 77, 78,
public administration, 171 79, 80, 81, 82, 83, 85, 87, 90, 98,
public authorities, 27, 218, 307 111, 112, 113, 118, 119, 134, 138,
metagovernance by, 226–8, 307 140, 143, 144, 169, 170, 171, 184,
modification of network’s 199, 233, 234
structure, 228 Richards, D., 43, 44, 78, 82
and network outputs and outcomes, Richardson, J., 81, 82, 169
223–4 Ripley, R.B., 134, 138
powers of, 219 risk, 5
public choice assumption, 46 Risse, T., 153, 155, 158, 163, 165
‘public ethos’, 279 Risse, T. et al., 158
publicity, 259, 275–6 Rogers, D.L., 200, 204
and accountability, 284–6 Rose, N., 3, 6, 107, 178, 314
sufficient, 287–90 Rosenau, J.N., 155, 265
public management and performance Rousseau, D. et al., 206
model, 216–17 Rousseau, J.J., 164, 248
public–private partnerships, 9, 158, Røvik, 36
189, 196 Rowe, G., 188
public purpose, 11 rules, and networks, 64
public sphere, 285–6, 287, 289, 292
Putnam, R.D., 71 Sabatier, P.A., 5, 49, 82, 126, 140, 144,
146, 221
rational choice approach, 45–50 Sabel, C. et al., 262
rational choice institutionalism, 28–9, safety, in process design, 202
33–5, 300 Sako, M., 206
similarity to historical Salamon, L.M., 84
institutionalism, 41–2 sanctions, 290–1, 294–5
rational reflexivity, 286 Sandel, M.J., 242
Rawls, J., 240, 265 Saward, M., 184
reflexive government, 107 Schaal G.S., 164
reflexive rationality, 12 Schaap, L., 99, 111, 114, 115, 117,
‘regimes of practices’, 40 118, 120, 130
‘regulated self-regulation’, 314 Scharpf, F.W., 3, 4, 10, 12, 13, 15, 29,
regulation, 15 44, 46, 48, 49, 50, 52, 57, 102, 113,
Reinicke, W.H., 153, 156, 161, 162 133, 135, 146, 147, 153, 154,
Index 355
Scharpf, F.W. – continued social systems, 115

155, 159, 163, 169, 172, 173, 179, closure, 114
199, 200, 201, 203, 204, 207, 213, formal organizations, 117
222, 233, 234, 299 and governance networks,
Schattschneider, E.E., 136, 142, 228 117–18
Schellenberg, 137 human interaction, 117
Schlosberg, D., 267 policy communication systems, 117
Schmitter, P.C., 139 society, 117
Schneider, V., 143, 157 societal dynamics, 5
Schön, D.A., 67, 101, 119, 121, societal fragmentation, 4, 5
140, 146 societal governance, 2
Schumpeter, J.A., 248 societal problem solving, 15
scope, 270 society, 117
Scott, J., 313 ‘socio-cybernetic systems’
Scott, R., 175 approach, 111
Scott, W.R., 29, 67, 177 soft power, 272–3
secretariats, 288 Sørensen, E., 13, 15, 16, 17, 43, 61,
‘Security, Territory and Population’, 185 236, 254, 310
selection bias, 61 sovereign power, 8, 262
selection of members, networks, 294 Spencer-Brown, G., 114
self-governing capacities, 185, 187 stability, 217
self-interest, 113, 118 state, 38, 89, 174, 189
self-organization, 190 ‘degovernmentalization’, 3
self-referentiality, 115 and governance networks, 173–4
self-regulation, 10, 169, 180 ‘hollowing out’, 184, 196
self-steering programmes, 116, 118, role of, 174
123, 124, 130, 199 state failure, 157
Selznick, P., 63 Steinmo, S., 28
‘shadow of hierarchy’, 299 Stinson, B.L., 205
Shafritz, 285 Stoker, G., 48, 78, 131
Shepsie, K., 64 strategic consensus, 146, 147
Sherlock, K.L. et al., 192 strategic externalities, 47
Sikkink, K., 161 Strecker, D., 164
Skelcher, C., 233, 234 ‘structural dilemma’, 54
Smith, M., 82 structures, of interaction, 312, 313
Smith, M.J., 78, 82 subgovernments, 138–9
social capital, 73 subsystems, 138–9
social closedness, 99 Sullivan, H., 131
social closure, 118 Susskind, L., 135
social constructivist approach, 188 Swedberg, R., 46
social constructivist (or normative) Sweden, framework laws, 191
institutionalism, 28, 29–30,
35–8 tactical polyvalence, of
differences from poststructuralist institutionalized discourses of
institutionalism, 41 governance, 40–1
social exclusion, 119 tasks, 74
social liberal government, 187 Tatenhove, 134
Social Network Analysis, 313 Taylor, R.C.R., 28, 41
social pressures, 70–71 ‘teamwork’, 32
356 Index
technologies veto power, 121, 122, 123, 126, 127,

of agency, 177–8, 188–9, 195 128, 129, 137, 305
of performance, 179, 194–5 governing, 125–8
Teisman, G.R., 147, 204 Villadsen, S., 191
Termeer, C.J.A.M., 128, 140
Teubner, G.C.M., 116, 130 Waarden, F.V., 9, 12, 138, 145
Thatcher, M., 153, 156 Wagenaar, H., 3, 141
Thelen, K., 28 Wamsley, G.L., 139, 143, 199
Thomas, G.M., 162 ‘war in ideas’, 272
Thomas, R., 87 Weber, M., 157, 159, 160, 217
Thompson, D., 281, 286, 294 Weimer, D.L., 211
Toke, D., 45, 49 welfare policies, 43, 193
top-down government, 3 welfare state, 106
Torfing, J., 8, 13, 16, 17, 28, services, 78
30, 43, 61, 236, 310, Westminster model, 78, 90
312, 313 Whetten, D.A., 200, 204
traditions, 90 Whetten, H.D.A., 139
transaction costs, 64, 96, 102, 204 Whitehouse, L., 194
transnational actors, democratic White, P.E., 139
accountability, 162 Wildavsky, A., 48, 49, 55, 144
Transnational Business Wildavsky, D., 82
Dialog, 162 Wilkinson, F., 206
transparency, 193, 194 Wilks, S., 140
Triantafillou, P., 193, 196 Williamson, O.E., 50, 64, 154
trust, 51, 88, 100, 227, 299 Willke, H., 130
and cooperation, 206 Winter, S., 73
Tsebelis, G., 46, 50 Wittgenstein, 89
Twist, M.J.W. van, 99, 113, 114, 115, Wolf, K.D., 157, 162, 233
118, 120 World Bank, 8
Wright, E.O., 239, 240, 241
uncertainty, 5 Wright, M., 140
unconscious cognitive Wynne, B., 141
closedness, 99
unconscious social closedness, 99 Young, I.M., 240, 255, 268,
Ury, W., 147 281, 285
US, 138–9 Young, O., 265
Veld, R.J. et al., 114, 140, 144 Zintl, R., 52

Versteeg, W., 3, 26, 101 Zofante, M., 49
vertical games, 56–7 Zouridis, S., 310
Verweij, M., 209 Zucker, L.G., 65

(Eva Sorensen, Jacob Torfing) Theories of Democrat (BookFi) PDF

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

(Eva Sorensen, Jacob Torfing) Theories of Democrat (BookFi) PDF

Transféré par

Droits d'auteur :

Formats disponibles

Theories of Democratic

POLITICIANS AND NETWORK DEMOCRACY (in Danish)

Also by Jacob Torfing

DISCOURSE THEORY IN EUROPEAN POLITICS: Identity, Policy and

List of Tables, Figures and Boxes xi

Introduction: Governance Network Research: Towards a

Part I Governance Network Dynamics

1 Theoretical Approaches to Governance

Actor calculation mechanism: preference for

Institutionalization and deinstitutionalization 62

Part II Governance Network Failure

5 Theoretical Approaches to Governance Network Failure 95

Similarities and differences 108

Part III Metagovernance

9 Theoretical Approaches to Metagovernance 169

10 Governing the Formation and Mobilization

Possibilities for public authorities to shape

Part IV Democratic Network Governance

13 Theoretical Approaches to Democratic Network

14 Governance Networks and Participation 247

15 Networks and Democratic Ideals: Equality,

Applying the standard democratic principles

16 Democratic Accountability and Network

The Second Generation of Governance Network

Governance networks are here to stay 297

I.1 Overview of four different governance network theories 17

2.1 Perceptions, calculations and games in governance

4.1 Ten lessons on how to manage your network 84

Mark Bevir, Professor, Department of Political Science, University of

Erik-Hans Klijn, Associate Professor, Department of Public

Linze Schaap, Leader of Centre for Local Democracy, The Faculty of

Jacob Torfing, Professor of Politics, Director of Centre for Democratic

Peter Triantafillou, Assistant Professor, Department of Social Science,

The structures and processes of public policy-making and societal

The construction of a huge, EU-funded shopping centre in major city

A large national bridge-building project stirred up a lot of public

The Employment Guidelines formulated by the European Union

The empirical examples demonstrate the need to abandon the false

The rise of governance network research

The study of governance networks is a novel research field that is founded

government, but subject to negotiations between a wide range of public,

break-through came in the field of policy analysis. Decision-making the-

most suitable answer to the challenge posed by increasing societal

The aims of this book

Governance network research is empirically oriented, but the study of

success and failure; 3. the forms and functions of metagovernance; and

Defining governance networks

Today, everybody talk about ‘governance’ and ‘networks’. The problema-

about the institutional conditions for ‘economic governance’, the private

emerging policy problems or exploit new opportunities. Networks that

Merits and problems of network governance

Whereas the early governance network theorists saw governance networks

are not bound by any common purpose or obligation. By contrast,

networks are claimed to have a large potential for proactive governance

political authorities must use their powers to influence the composition,

First and second generation research

Although the field of research is relatively new, it is possible to talk

1. How can we explain the formation, functioning and development of

Some governance network theorists have drawn our attention to the

democratic conduct (Sørensen & Torfing 2005a). However, more research

Table I.1 Overview of four different governance network theories

Conflict Interdependency theory Governmentality theory

who want to move beyond the established theories and explore