PUBLIC MANAGEMENT AND POLICY NETWORKS

Foundations of a network approach to governance
Erik-Hans Klijn and
Joop F. M. Koppenjan

E. H. Klijn
Department of Public
Administration

Abstract
In this article we address the elaboration of the central concepts of a theory of networks and of network management. We suggest that the network approach builds on several theoretical traditions. After this we clarify the theoretical concepts and axioms of the policy network approach and argue that this framework has important explanatory power both on the level of strategic interaction processes as well as on the level of institutional relations. We argue that government’s special resources and its unique legitimacy as representative of the common interest make it the outstanding candidate for fulfilling the
role of network manager, a role which means arranging and facilitating interaction processes within networks in such a way that problems of under or non representation are properly addressed and interests are articulated and dealt with in an open, transparent and balanced manner.

INTRODUCTION: NETWORKS AND THE GOVERNANCE DEBATE
The apparently broad consensus that has developed around the idea that government
is actually not the cockpit from which society is governed and that policy making
processes rather are generally an interplay among various actors has led to a full-scale
search for new governing methods and a discussion on governance and public
management (Kooiman 1993; Rhodes 1996). This has fused with discussions in public
administration on managerial reform and the adoption of business management
techniques under the heading of new public management (NPM), a framework for
implementing these reforms (Pollitt 1990; Kickert 1997).
Governance, public management and network management
Governance can roughly be described as the ‘directed inuence of societal processes’.
Many kinds of mechanisms, some of them quite complex, are involved and these do
not only originate from public actors (Kooiman 1993; Kickert et al. 1997). Thus it is
no surprise that the word ‘governance’ has become a catchword in the last few years
and that it has been used in many different contexts. Rhodes (1996) nds at least six
distinct uses of the concept. These seem to fall largely into two groups of de nitions.
In one case, governance pertains to notions of reducing the state and distinguishing
between government and governance. Government should be reduced or more done
with less (Osborne and Gaebler 1992), mainly by employing new public management
techniques. In the other case, the term governance is reserved for theories and cases
that take into account the interdependencies of public, private and semi-private actors.
In this denition governance refers to self-organizing networks. Ideas on network
management, which are the focus of this article, t in this category. The two
conceptualizations of governance have totally different perspectives on public manage-
ment and the role of government in society, and they draw their theoretical inspiration
from very different sources.
While ‘new public management’ represents an attempt to translate managerial ideas
from the private sector to public organizations, such as contracting out, client
orientation and the introduction of market mechanisms (Pollit 1990; Kickert 1997),
‘network management’ focuses more on mediating and co-ordinating interorganiza-
tional policy making. The theoretical basis for this alternative view is found in the
network approach to policy. This has acquired a prominent position in policy science
and public administration, demonstrated by the number of publications on policy
networks and network management in Europe (Wilks and Wright 1987; Rhodes 1988;
Marin and Mayentz 1991; Marsh and Rhodes 1992; Glasbergen 1995; Kickert et al.
1997) and in the United States (Milward and Wamsley 1985; Provan and Milward
1995; O’Toole 1997). Along with its growth in popularity, however, the policy network approach and ideas on network management also have met with frequent
criticism.
136
Public Management: an international journal of research and theory

Criticism of the network approach
The most important critical remarks can be categorized under ve headings:
(1) Lack of theoretical foundations and clear concepts. A major point of criticism is that
the network approach is not based on a solid theoretical body of knowledge as
a result of which a coherent theoretical framework is lacking and its concepts are
unclear (Borzel 1998).
(2) Lack of explanatory power. According to some critics the network approach is
primarily ‘metaphoric’. It is highly descriptive and does not provide for the
explanations of outcomes of policy processes (see Dowding 1995; Salancik 1995;
Blom-Hansen 1998; Borzel 1998).
(3) Neglect of the role of power. Another criticism is that the network approach, and
especially that part which focuses on network management, places too much
emphasis on the role of co-operation and consensus and ignores conict, power
and power differences (Brans 1997).
(4) Lack of clear evaluation criteria. Because the network approach rejects the use of
ex ante formulated goals as evaluation criteria, it is said that the approach does
not offer a clear framework for evaluation. The evaluation criteria it suggests are
considered to be vague and lacking a substantive norm. Thus the approach
insufciently acknowledges the inuence of the goals of governments (Propper
1996; Brans 1997).
(5) Normative objections against networks and the role of public actors within them. Critics
argue that the network approach considers government organizations to be the
same as any other organization, and neglects their role as guardian of the public
interest. They further contend that government strategies based on network
theory can seriously jeopardize policy innovation, the pursuit of the common
good and the primacy of politics (Ripley and Franklin 1987; Marin and Mayentz
1991; Rhodes 1996; De Bruijn and Ringeling 1997).

The focus of this article
We believe that some of this criticism is right and justied, and should be taken very
seriously. But in many cases, the criticism does not do justice to the network
approach. The aim of this article is to critically evaluate both the network approach and the
criticism it has encountered in order to clarify existing misunderstandings and to improve network
theory as a framework for the explanation, evaluation and improvement of public policy and
public management.

Klijn & Koppenjan: Public management and policy networks
137

We do this by systematically addressing the above mentioned ve categories of
criticism. We rst review the theoretical background and concepts underlying the
network approach (section 2). We show that the network approach is solidly rooted in
the theoretical history of policy science and organization theory and that it has a rather
well developed set of concepts. In section 3 we try to systematize the explanations of
interactions and outcomes of policy processes that the network approach offers. We try
to strengthen the framework on the institutional level by introducing the concept of
rules and trust. In section 4 we clarify the role of power and conict in networks, and
in section 5 we address the problem of evaluation and evaluation criteria in networks.
Section 6 elaborates upon the misunderstandings that surround the role of public
organizations in networks and here we provide management strategies that can help
governments to adequately operate under network conditions. In the last section we
discuss some topics which need further theoretical, empirical and normative con-
sideration.

THE THEORETICAL FOUNDATION OF THE NETWORK APPROACH
A major point of criticism of the network approach is that it lacks both a theoretical
foundation and clear concepts. But is this true? In this section we reject the suggestion
that the network approach has no theoretical basis. On the contrary, there is a rich
theoretical tradition on which the approach is founded. And in response to confusion
over what the core concepts of the network approach are, we suggest a framework that
can satisfy the demand for conceptual clarication.

The theoretical roots of the policy network approach
The use of the network concept in policy science dates back to the early 1970s. In
implementation studies, especially in what has become known as the ‘bottom-up
approach’ (see Hjern and Porter 1981), as well as in intergovernmental relations
literature (see Friend et al. 1974; and the very inuential work of Scharpf: Scharpf et
al. 1978), the concept has been used to map relation patterns between organizations
and to assess the inuence of these patterns for policy processes. In these two early
uses of the network approach to policy one can nd the inuence of theoretical notions
from interorganizational theory and insights from the interactive perspective on public
policy (Hufen and Ringeling 1990; Klijn 1997).
The interactive policy approach in policy science is visible in the work of authors
such as Allison (1971), Cohen et al. (1972) and Lindblom (Lindblom 1965; Lindblom
and Cohen 1979). In their work policy appears as the result of an interaction between
a multitude of actors. Con icting interests characterize policy processes and problem
denitions are dynamic and unpredictable.
138
Public Management: an international journal of research and theory
The policy network approach builds on this process model since it also focuses
attention on the interaction processes between interdependent actors and the com-
plexity of objectives and strategies as a consequence of that interaction. An important
difference with the process model is that in the network approach, more attention is
given to the institutional context in which complex interactions take place. In an
attempt to elaborate the institutional context of complex interaction processes,
network theoreticians are inspired by interorganizational theory (Levine and White
1961; Negandhi 1975; Aldrich 1979).
The central starting point of the interorganizational approach is that the environ-
ment of organizations consists of other organizations. In order to survive, an
organization requires resources from other organizations. These organizations engage
in exchange relations with each other and a network of mutually dependent actors
emerges. There is substantial attention in interorganizational theory for the links
between organizations and the strategies used by organizations to inuence the
exchange processes (see, for example, Levine and White 1961; Cook 1977; Aldrich
and Whetten 1981; Benson 1982).
As it has evolved, the policy network approach has developed its own, distinctive
theoretical framework. The network approach assumes that policy is made in complex
interaction processes between a large number of actors which takes place within
networks of interdependent actors. These actors are mutually dependent so policy can
only be realized on the basis of co-operation. This co-operation, however, is by no
means simple or spontaneous, and it requires types of game management and network
constitution. The central starting points of the network approach are elaborated
below.

Clarication of central concepts: Policy networks as a context of interactions
The network approach assumes that actors are mutually dependent. Actors cannot
achieve their objectives without resources that are possessed by other actors (Scharpf
1978; Benson 1982; Rhodes 1988). Interaction patterns between actors emerge
around policy problems and resource clusters, and these patterns acquire a degree of
sustainability because of the limited substitutability of resources. Rules develop which
regulate the behaviour of actors and resource distribution in the network, and this also
inuences interactions within networks. Resource distribution and rules are gradually
shaped in interactions, but they are also solidied and altered in these interactions
(Giddens 1984). Thus the created policy networks form a context within which actors
act strategically and in which strategic action is confronted by the strategic action of
others.
Within networks,
1
series of interactions occur around policy and other issues.These
series of interactions can be called games (Crozier and Friedberg 1980; Rhodes 1981;
 
Klijn & Koppenjan: Public management and policy networks
139

Scharpf 1997). Their positions in the network and the strategic action in the game
determine the positions of the players. During the game, actors operate within the
established resource distribution and set of rules, which are to a large extent framed
by the network. In addition, they have to operate strategically in order to handle the
given dependencies in the game so that they can achieve their own objectives. During
this action, they interpret the existing rules that are, after all, ambiguous (March and
Olsen 1989; Klijn 1996b).
Policy processes can thus be seen as a collection of games between actors. In these
games, each of the various actors has its own perceptions of the nature of the problem,
the desired solutions and of the other actors in the network. On the basis of these
perceptions, actors select strategies. The outcomes of the game are a consequence of
the interactions of strategies of different players in the game. These strategies are
however inuenced by the perceptions of the actors, the power and resource divisions
in the network and the rules of the network. We will elaborate on this in the next
section.

Concerted action and network management
A central question within the network approach is how concerted action is established
around a concrete issue. Actors need to co-operate in order to achieve satisfying
outcomes. This is not always easy, despite durable dependencies, since major conict
may arise at the process level about, for instance, the distribution of costs and bene ts
of a solution. Policy is made and policy processes occur in the tension between
dependency and the diversity of goals and interests. And while this tension can be
more or less regulated by the rules and resource distribution in the network, the
tension will exist and needs to be solved in any policy game.
Since co-operation and collaboration of goals and interests does not happen of its
own accord, steering of complex games in networks is necessary. These steering
strategies, i.e. network management, are primarily focused on the improvement of co-
operation between involved actors (O’Toole 1988). The sometimes implicit assump-
tion is that satisfying outcomes for actors are not possible without network
management. Network management is thus an independent variable in the develop-
ment of policy processes.
In the literature on network management, a distinction is made between two types
of network management strategies: process management and network constitution (for
extensive discussion, see Klijn et al. 1995; Kickert et al. 1997). Process management
intends to improve the interaction between actors in policy games. In essence this
concerns steering strategies that seek to unite the various perceptions of actors and
solve the organizational problem that various organizations, in having autonomously
developed their own strategies, are not automatically in concert with one another. In
doing so, actors cannot unilaterally determine each other’s strategy. What is important
140
Public Management: an international journal of research and theory

is that strategies of process management assume the structure and composition of the
network as given. So, rules (formal or informal), resource divisions and existing actors
are treated as a given starting point for the management strategies. Important process
management strategies are:
c
the selection and activation of actors (Friend et al. 1974; Hanf and Scharpf
1978); to take successful policy initiatives actors with the necessary resources
must be selected and motivated to participate;
c
the improvement of mutual perception about an issue or solution; given the fact
that actors have different perceptions of the problem, solution and existing
situation, network management must be aimed at creating a minimum con-
vergence of perceptions and at creating packages of goals which are acceptable
for a workable coalition of actors;
c
the creation of temporary organizational arrangements between organizations;
because co-ordination between different actors is not secured, organizational
arrangements have to be created to sustain interactions and co-ordinate strate-
gies;
c
the improvement and supervision of interactions by means of process and
conict management (Susskind and Cruikshank 1987; Mandell 1990).
Network constitution is focused on realizing changes in the network. Based on the
assumption that the institutional characteristics of the network also inuence strategies
and co-operation opportunities of actors, attempts can be made to change one or
more of these characteristics. In general, these strategies are time consuming since
they seek institutional change. As a result, they are usually unsuitable for inuencing
policy games that are already underway. Network constitution strategies can be
focused on:
c
changing the position of actors or the introduction of new actors; introducing
new actors into the network can bring new perceptions but can also change
given positions of power and regularities in interaction;
c
changing the rules (for instance those that regulate access to a process) (Ostrom
1986); changing both formal and informal rules can lead to different patterns of
interactions of frames. Changing conict-regulating mechanisms, for instance,
can lead to other strategies and interaction patterns because actors want to avoid
binding conict resolution;
c
reframing (fundamentally alter ideas about the functioning and the substantive
problems of the network) (Rein and Schon 1992); sometimes it seems necessary
to initiate radical changes in perceptions of sectoral problems or ways of doing
things. Mostly central government tries to achieve this by radical system
changes, but sometimes actors within networks themselves try to achieve this.
The literature on the network approach explicitly mentions that network management
is far from easy. It requires knowledge of the network and numerous skills including
Klijn & Koppenjan: Public management and policy networks

negotiation skills since network management strategies are conducted in a situation of
mutual dependency. Thus a network manager is not a central actor or director, but
rather a mediator and stimulator (Forester 1989). This role is not necessarily intended
for only one actor. Even though public actors often assume the role of network
manager, other actors can do so as well. Which actor has the authority and opportunity
to full the role of network manager is most certainly in uenced by the strategic
position of actors and the rules, including rules of behaviour, in use in the network
(Ostrom 1986; Burns and Flam 1987). The most important theoretical assumptions of
the network approach are summarized in Table 1.

THE NETWORK APPROACH AS AN EXPLANATORY MODEL
The network concept is said to be merely a metaphor, and that because of this the
network approach lacks real explanatory power. The question is: to what extent can

Table 1: Theoretical assumptions of the policy network approach

Policy Network Policy processes Outcomes
 Network management
Actors are mutually dependent for reaching objectives.
 Within networks, interactions between actors over policy and issues take place focused on solving the tension between dependencies on the one hand and diverging and con icting interests on the other. Policy is the result of complex interactions between actors who
participate in concrete games in a network.
 Given the variety of goals and interests and – as a result – the actual and potential con ict over the distribution of costs and benefits, co-operation is not automatic and does not develop without problems.
Dependencies create sustainable relations between actors.
 In doing so actors depart from perceptions they hold about the policy
area, the actors and the decisions at stake.
  Concerted action can be improved through incentives for co-operation,
through process and con ict management, and through the reduction of risks linked to co-operation.
Dependencies create some veto power for various actors. Actors select specific strategies on the basis of perceptions.
  
The sustainability of interactions creates and solidies a distribution of resources between actors. Policy processes are complex and not entirely predictable because of the
variety of actors, perceptions and strategies.  
In the course of interactions, rules are formed and solidied which regulate actor behaviour.   
Resource distribution and rule formation lead to a certain closeness of networks for outside actors.   

142
Public Management: an international journal of research and theory

structural approaches to networks explain policy outcomes and processes? How do
these network features relate to actor behaviour and their interactions (see Downing
1995; Blom-Hansen 1997; Borzel 1998)? In this section we examine the nature of the
explanations the network approach comes up with and how they deal with the multi-
level character of policy processes in networks.
We suggest that the issue of co-operation is central to the policy network approach.
Given the dependency of actors on each other’s resources, policies can only be
developed when actors make their resources available. Therefore, in the network
approach, explanations for the success or failure of policy processes are based on the
extent to which co-operation has been achieved. Co-operation cannot be achieved
when interactions between actors stagnate, are blocked or have led to undesired or
unforeseen consequences, or because interactions are inuenced by institutional
characteristics. Explanations that fall into this latter category are based on the premise
that rules limit the behaviour of actors even when the actors themselves may have
violated these rules consciously or unconsciously.
Thus, explanations for the development ofinteraction processes in networks are found
both in institutional characteristics – the resources and the rules – and in the character-
istics of the interaction situation – the players, their stakes and their strategies (also see,
for instance, Scharpf 1997). As both Dowding (1995) and Blom-Hansen (1997) state, to
arrive at explanations network orientations which focus on the structural dimensions of
networks should be combined with an actor and interaction approach.
Process variables as factors for success and failure
The network approach assumes that policy outcomes are the result of interaction of
strategies of various actors. The involvement of these actors is a consequence of the
fact that they possess resources that require their involvement in the handling and
solution of a particular problem. They can block interaction processes by withdrawing
their resources: they have veto power. Replacement of these resources is not always
possible and when it is, it might be costly and time consuming. The same can be said
for attempts to coerce co-operation, for instance by ordering a municipality to change
its zoning plan if it does not do so of its own accord.
An important explanation for failing to realize concerted policy outcomes is the fact
thatactorsareinsufciently aware of theirexternal dependencies. In this casethey assume
that they can solve the problem alone or that they can impose their solution on other
actors. But even when actors are awareof their external dependencies, it is often quite an
undertaking to bring the various goals and interests together. Differences and disagree-
ments in perceptions between actors may cause conicts and block the interaction. Only
when actorsareable to bring their perceptions together and formulate commongoals and
interestswill policy games lead tosatisfactory outcomes. Learning processesarethus very
important in policy games and process management seeks to stimulate these. Preferences
Klijn & Koppenjan: Public management and policy networks
143

of actors are not xed (Weick 1979; March 1988). Discovering new goals that are
interesting to actors can prevent stagnation.
On the other hand, actors may lose interest in policy games so that stagnation
occurs. These can be the consequence of the low priority that a policy problem has in
the perception of one or more actors. Stagnation and blockades may also be a
consequence of an undesirable balance between interaction costs and expected
outcomes of policy games or of risks related to policy games as a consequence of
unexpected strategies of others. Thus there is a risk that as soon as they have proted
in the interaction process, actors exit or threaten to exit, which leaves other actors
empty-handed. This problem typically exists when parties commit to transaction-
speci c investments that cannot, or not easily, be used for other transactions
(Williamson 1979; Barney and Hesterly 1996). And then there is always the danger
that outsiders may prot from the mutual efforts of a particular group without having
made any contribution to the endeavour (free-rider problem). In this case, network
management should focus on the organization of interactions and protection of
interests of the actors involved (De Bruijn et al. 1998).
Concerted action thus requires that actors are able to assess their mutual
dependencies and possibilities of co-operation and that the risks and costs involved are
limited. Lack of awareness of mutual dependencies, con icts of interests, interaction
costs and risks are important explanations for the failure of concerted policy.
Conversely, the emergence of concerted action is explained through the acknowl-
edgement of mutual dependencies, converging perceptions, the existence of incentives
which improve co-operation and the limitation of interaction risks through the
application of types of game management.

The structure of the network as explanation
As argued, explanations for the success or failure of policy processes are divided into
process variables and factors connected to the structure of the network. Concerted
solutions have to be established in games. When network level factors for the
explanation of success and/or failure of policy and policy processes are included, game
and network are linked; after all, it concerns the question of how characteristics at the
network level inuence the development of the game.
Intensive interaction between actors creates a specic resource distribution that
inuences the functioning of the network. Actors recognize/acknowledge that certain
resources are relevant or even necessary to the realization of policy outcomes. These
resources provide actors with veto power. The resources enable them to veto
interaction processes and they thus acquire a privileged position in the network and in
the games within that network. The greater the veto power of an actor, the more
indispensable the actor is to the policy games. The success of policy games is thus
partially determined by the degree to which indispensable resources, and the actors
144
Public Management: an international journal of research and theory

who own them, are involved (Scharpf 1997). Changes in the resource distribution in
the networks are, therefore, reected in the policy games. Thus, ten years ago it would
have been unthinkable that commercial project developers would have any role at all
in urban renewal policy in The Netherlands. But now that resources in the housing
network have been redistributed and national government subsidies for public housing
have declined, these actors are involved.
In addition, the degree to which actors participating in an interaction process are
also connected in more encompassing interaction patterns and rules is important
(Ostrom 1986; Burns and Flam 1987; Ostrom et al. 1994; Klijn 1996b). Interactions
between actors from various networks may be difcult because they do not interact
otherwise or have few rules to regulate their interactions. In other words, automatic
co-ordination mechanisms and a degree of trust are lacking, and this results in higher
interaction costs (see Hindmoor 1998, on the embeddedness of interactions).
Rules play an important role in the development of policy processes. Rules enable
actors to depart from minimal institutional agreements in their interaction. This
reduces transaction costs and simpli es collaboration (Scharpf 1997; Hindmoor 1998).
Initially, without knowledge of the network, it is difcult to arrive at general
statements about the inuence of rules on policy networks. Rules are social
constructions of actors in a network, and they differ from network to network.
Research has shown, however, that rules of conict management and mediation, as
well as rules to protect autonomy and position, are important for determining the
possibility of co-operation (Klijn 1996a; Scharpf 1997). The stronger the territorial
demarcations in a network and the weaker the rules for conict management and
mediation, the more difcult decision making will be. The lack of trust and useful
sanctions makes it dif cult to prevent exploitative behaviour on the part of actors.
These examples illustrate the structuring nature of rules in networks. They can
improve or limit certain styles of interaction. Thus the lack of conict regulating
mechanisms and trust will more quickly lead to non-co-operative outcomes of ‘mixed-
motive’ policy games which usually result in less for the actor than more co-operative
strategies.There are no mechanisms to prevent or decrease incidences of opportunistic
behaviour (Scharpf 1997). The central characteristic by which these outcomes are
produced is trust. Scharpf concludes: ‘In other words, being able to trust, and being
trusted, is an advantage – but exploiting trust may be even more advantageous’
(Scharpf 1997: 89). Rules are one of the most important pillars of trust, but herein
lies a problem. Actors can violate rules, whether formal or informal, because the
result might be attractive to the actor. In this sense, rules do regulate but not
determine and they can be changed. Each analysis of decision making in networks must
take this into account. Particular attention should be focused on the process of
reformulation and reinterpretation of rules as a consequence of circumstances external
to the network and the strategic choices of actors. Table 2 summarizes the most
important explanations for the success and failure of policy processes in networks.
Klijn & Koppenjan: Public management and policy networks
145

THE ROLE OF POWER AND CONFLICT IN NETWORKS
Within the network approach literature, works that emphasize network management
are critiqued for elevating co-operation to a norm and for insuf ciently considering the
role of conict and differences in power (Brans 1997; De Bruijn and Ringeling 1997).
Indeed, co-operation is an important element of network theory, both with respect to
the explanations of success and failure and the development of prescriptions. But
power and conict are not excluded from consideration.

Power, con ict and durable relations
Without co-operation, actors who nd themselves in situations of mutual depend-
encies cannot realize their objectives. This does not mean that co-operation is
established without conict. Nor does it mean that actors will manage to co-operate.
Durable dependency relations do not necessarily mean that no conict will emerge
over the distribution of costs and bene ts in concrete policy processes. As an example,
one need only consider labour relations where employers and labour unions maintain
a durable relationship characterized by both co-operation and conict. It is exactly this

Table 2: Some of the more important explanations for success and failure from the perspective of the network management approach

Explanations for success and failure
At interaction level At network level

The degree to which actors are aware of their mutual dependencies.  The degree to which actors possess veto power because their resources
are indispensable.
The degree to which actors succeed in redi ning diverging and
con icting interests into a common interest. The degree to which actors with veto power are actually involved in the
process.
The degree to which interaction costs are balanced favourably or
unfavourably with perceived outcomes of the interaction. The degree to which actors in a game belong to the same network, so
that they also interact with each other elsewhere and have developed
mutual rules.
The degree to which risks of the interaction within the game as a
consequence of strategies of other actors are limited. The degree to which dened problems and solutions, and the way these
are handled, t within the rules developed in the network.
The degree to which game management is foreseen (mutual perception
development; arrangements; game and con ict management). 

146
Public Management: an international journal of research and theory

tension between co-operation and conict which needs to be resolved (Scharpf 1997).
The lack of a dominant actor does not imply that resources are equally distributed
among actors (Knight 1992). Also, rules may operate to the advantage of some, and
to the disadvantage of other actors. This is implied by the fact that rules have been
formed during earlier interactions. The inequalities resulting from earlier interactions
are incorporated into the existing rules. A change of rules is thus also (but not
exclusively) a battle for power between actors (Burns and Flam 1987). In this sense
the network approach is quite attentive to ‘invisible’ forms of power, traditionally
known as ‘the mobilisation of bias’ (Bachrach and Baratz 1962) such as rules which
shape the problem denitions and entrees of the actors in games and networks.
In short: the differences in the distribution of resources matter. Actors will use
them to inuence the process and the substance of the interaction. A project developer
or a municipality will generally be able to wield more inuence over building plans
than citizens’ organizations. Citizens lack the ‘know how’ and organizational capacity
to be present throughout the process and provide input.

Veto power, network management and less privileged interests
Nevertheless, less powerful actors may influence decision making. They can use their
veto power and their ability to use resources for blocking decision making and thus
create stagnation or blockade. Since stagnation and blockade result in extra costs – at
the very least – more powerful actors need to consider their less powerful colleagues.
In order to encourage actors not to use their veto power, some degree of convergence
of perceptions must be achieved. This is also a demonstration of the importance and
the need for process management. Furthermore, and certainly as important, this leads
to the consideration of information and interests of other actors to enhance the quality
and support of policy initiatives. The starting point of process management is to
enhance the learning capability of policy processes by including information and
interests of various actors so that more complete policy initiatives can be developed.
From a network approach, the involvement of actors is not only recommended for
normative reasons, but also for reasons of effectiveness and eficiency. Expertise and
knowledge for handling policy as well as other sorts of problems is not available in one
place only and thus a confrontation of policy initiatives with information and interests
of other actors is necessary. Power differences influence the way in which this process
evolves. As long as actors hold veto power, they have influence.
A more serious problem occurs when actors have no veto power and/or are
excluded from interaction by other parties.This can happen when interaction patterns
between actors result in a certain degree of network insularity (Laumann and Knoke
1987; Rhodes 1988). Outsiders can only access the network if they familiarize
themselves with the rules of behaviour and the language of the network (Klijn 1996a).
Klijn & Koppenjan: Public management and policy networks
147

In the development of theory in network management, substantial attention is given to
this aspect of networks and to the negative policy effects this may have for the
environment. When formulating prescriptions, opportunities for dealing with the
limitations of this closed nature are sought in network constitution and in the use of
process norms. These will be clari ed in the next section.

SUCCESS AND FAILURE OF POLICY: THE SEARCH FOR EVALUATION
CRITERIA
The way in which success and failure are assessed in policy analysis is affected by use
of the network approach.The criteria that have been proposed for evaluating outcomes
and policy processes in networks have been criticized for being too vague and not
sufciently taking into account the goals of public actors (Marin and Mayentz 1991;
Propper 1996). If, however, we start from the premise that a variety of actors is
involved and that they all have different objectives, then it should not be expected that
the process and outcome can be evaluated in terms of the objectives of one actor, even
if this is a public actor.This type of top-down approach is not in tune with the network
approach. Instead, criteria are needed which consider the multi-actor, dynamic
character of interaction in networks.

A unilaterally determined substantive criterion is not tenable
In classic top-down approaches, success and failure of policy processes are measured in
terms of a public actor’s effectiveness in achieving goals. The justi cation for this norm
is that this actor represents the public interest and is the central manager in policy
processes. We argue that this yardstick is not appropriate in the network approach for
a number of reasons.
First, there is the problem of the ‘classic goal achievement method’, namely the
accurate determination of the formulated objective. In networks actors are relatively
autonomous and there is no central, co-ordinating actor. Each of the actors has their
own objectives, so it is unclear whose objective should serve as the yardstick. The
pragmatic choice of the public actor’s objective is not helpful. Frequently, several
public actors are involved in decision-making processes so that it is even difcult to
determine what ‘the public interest’ is. The solution of de ning the public or common
interest in terms of elected political representation is based on an unrealistic and naive
assumption about the accumulation of preferences of citizens that has long since been
falsi ed by theory and research. Calling upon the common interest is increasingly
difcult in a society that continues to fragment (see next section).
The problem of nding the right evaluation criterion is not solved by using a
collectively achieved formulation of a societal problem or objectives as a yardstick
148
Public Management: an international journal of research and theory
________________________________________
either (compare Glasbergen 1995). It is unlikely that actors possess a common
perception of the problem or objective at the beginning of a process given the large
number of parties involved and their diverging interests.
There is an additional problem that makes the use of ex ante formulated objectives
in the network approach untenable. In interaction processes, actors adapt their
perceptions and objectives based on the responses of other parties and events in the
environment. As a result, they arrive at a conclusion through a goal-seeking process.
Ex ante problem formulation or objective setting as a yardstick, whether or not it has
been arrived at collectively, does not take this into account. After all, the problem
formulation and the objective(s) will change in the course of the process. If the
evaluator only focuses on initial problem de nitions or objectives, an important
element of interaction processes may be ignored – namely that the perceptions of
problems and solutions are subject to constant change – and learning processes will be
evaluated negatively.
A nal problem with evaluating success and failure by means of a prior and collectively
agreed upon yardstick is the following: if parties do not participate in the interaction
process, the chances are high that their interests and preferences will not be represented.
The question of the degree to which the nal solution furthers or jeopardizes their
interestsis not considered in the evaluation. In the network approach, the process used to
arrive at a possible common problem formulation is critical.
Assessment of substance: ex post satisfying and win-win situations
An adequate yardstick should take into consideration the fact that various actors with
diverging interests interact, that objectives are dif cult to measure, that objectives shift
and that the interests of those involved may be overlooked. In the network
management literature, the solution to this is found in the ex post satisfying criterion
(Teisman  1992/1995  1992/1995 ). This means that the starting point of the assessment of policy
process outcomes is based on the subjective judgements of individual actors. The
practice of interviewing various actors about their satisfaction after the conclusion of
the process can solve several problems.
At the end of the process the actors have to determine how the outcome has
beneted them, what the outcome has cost them and how the outcome ts in the
changing environment. In this assessment, both substantive and process elements are
weighted, something that a researcher cannot possibly do. Since this is an evaluation
viewed in hindsight, justice is done to the development of objectives and problem
formulation during the process. Learning behaviour receives the appreciation it
deserves. One potential danger is that actors will, in retrospect, suggest rationaliza-
tions that mask possible de ciencies. Comparing the subjective judgement of actors to
actual achieved outcomes and to the interests of the actors involved might alleviate this
problem. Also, an assessment of learning results should take into account the
Klijn & Koppenjan: Public management and policy networks
149
________________________________________
development of the substantive content of policy proposals. Policy proposals are better
when they have been able to incorporate the various goals and desires of actors and
have included or explicitly rejected criticism of earlier policy proposals. The ex post
satisfying criterion solves problems of measurability, assessment and dynamics.
What remains are problems of intersubjectivity and exclusion. The problem of
intersubjectivity refers to the fact that actors’ statements may diverge strongly and will
not directly lead to a general assessment of success or failure of the policy process. In
other words, there is a need to assess the individual judgement of actors at a higher
level. For this, the win-win situation criterion is used in the network approach. When
actors have succeeded in reaching an outcome that represents an improvement from
the earlier situation for all, or when an undesirable situation is avoided through co-
operation, we speak of a win-win situation (compare the Pareto criterion, see for
instance Dery 1984). The nature of the improvement may differ for the various
parties.Also a party may have actually lost, but this is compensated by other elements.
A win-win situation can be assessed by aggregating the individual ex post judgement
of actors at a more collective level. It is also conceivable that the actors involved are
given the opportunity to arrive at an assessment of the process and its outcome
together. Here too, statements will have to be validated in relation to objectives
formulated by actors and realized outcomes.
The necessity of process norms
Not all forms of co-operation are of equal interest to all parties, nor are they always
desirable from a wider perspective. It is conceivable that actors who worked together
on a problem nd themselves in a type of group-think situation, with the interests of
the outside world insufciently considered or not considered at all. Win-win situations
may have been achieved precisely because certain actors were excluded from the
decision making or because costs were placed elsewhere.
It is important that interaction processes are accessible to third parties, that careful
assessments are possible and that contact is maintained with the outside world. This
means that in addition to the win-win situation criterion, process criteria such as
openness, carefulness, reliability and legitimacy are included when evaluating inter-
action processes in networks. Also, the external effects of these processes should be
included (Kickert et al. 1997).
Thus in the network approach, the ex post judgement of actors about the process
and the outcome, in combination with process criteria and concern for external
effects, are used in order to determine the success or failure of policy processes. These
are considered to be better indicators for success and failure than the ex ante
formulated objectives of one actor.
150
Public Management: an international journal of research and theory
________________________________________
PUBLIC ACTORS AND POLICY NETWORKS
In the network approach, public actors do not play the dominant role they often are
ascribed in other public administration perspectives. This has evoked criticism. The
network approach is accused of considering government merely as ‘an actor among
actors’, which can lead to problems of democratic legitimacy or accountability (Hirst
1994; Rhodes 1997). Sometimes critics describe networks as closed subsystems
dominated by established interest groups, which impede innovation and maximize
their private interests at the cost of others. As a result the common interest is
neglected and the primacy of politics is eroded (Ripley and Franklin 1987; Marsh and
Rhodes 1992; De Bruijn and Ringeling 1997). From this point of view it is
inconceivable to present a network approach as a normative theory. Yet this is what we
do when we derive evaluation criteria and prescriptions from network theory.
In this section we warn against mixing up the real world and the theoretical
framework that is used to analyse, evaluate and improve it. If the real world does not
t our norms, this does not mean we can refrain from it, or that our attempts to
improve it are not subject to the same normative shortcomings. The fact that
government is confronted with the reality of its dependency upon other actors does
not imply that its position is not special. Nor does developing network strategies and
strategies for network management imply that the shortcomings of existing realities
are accepted and elevated to guiding principles. So, there are clearly misunderstand-
ings about the position of government in network theory. In this section we will clarify
the ideas within network theory on the role of governments as public managers in
networks and the risks and potential these have for representation of the public interest
and the primacy of politics.

The special position of government
Network theory by no means presumes that governments are like other actors.
Governments have unique resources at their disposal and work to achieve unique goals.
They occupy a special position, which in most cases cannot be lled by others.
Resources that determine this special position include: sizeable budgets and personnel,
special powers, access to mass media, a monopoly on the use of force and democratic
legitimization. Access to these resources provides governments with considerable
power. However, they also encounter certain limitations as a function of their
uniqueness (Kickert et al. 1997):
• The tasks of government de ne to a great extent its interdependence and often
condemn it to interactions with particular social and administrative partners that
it cannot freely choose.
• In performing its duties, government is frequently not allowed to ‘goal bargain’.
In this respect, it often does not have the option of carrying out tasks through
negotiation.
• Governments are bound to the norms and rules they wish to impose on others:
principles of good government, consideration for minorities and adversaries,
guidelines of democratic regulations, et cetera.Where other actors operate with
strategic ingenuity, governments are expected to show exemplary behaviour.
• Because of its public nature and democratic monitoring, more demands are
made on government’s strategic interactions.These actions are scrutinized by the
watchful eye of the media.
• Government is not only expected to operate effectively and efciently, its actions
must also be legitimate: they must be ‘backed’ by politicians and political
parties, but there must also be social acceptance of public policy.

So, while their unique position means governments have access to special resources, it
also limits their possibilities to use them in order to attain their goals in network
situations.
But the objective statement of these conditions may not be getting at the real core
of the debate about the role of governments in networks. The debate is mainly about
the normative implications of becoming engaged in networks: public actors represent
the common interest and that is why they should not enter into interactions and
partnerships with other parties in society. One counter argument that makes such a
normative position problematic is that since governments are indeed dependent, this
way of thinking simply does not help when they have to deal with the genuine
complexities of their tasks and environment. But we may also go more deeply into the
question regarding why the representation of the common interest is at odds with
engaging in network processes. What does representation of the common interest mean?
Does it imply that the objectives of governmental organizations are superior to those of
other actors, because they are mandated by elections (i.e. the will of the people)?
Political science and public administration research of the last fty years has shown
that there are many problems with the aggregation of the preferences of voters in
governmental policies and that representative bodies function far from optimally. In
concrete game situations it appears that public organizations de ne the common
interest in a variety of ways and frequently use it to legitimize their own organizational
objectives. This situation is bound to inuence our appreciation of the concept of
‘common interest’.
If common interest has any meaning at all, it is because it refers to the importance
of criteria such as proportional representation, openness, equity, fairness, reliability, et
cetera. Note that these are all criteria that concern the quality of the interaction
process rather than the content of policies or decisions. And that it is precisely network
management that has been developed as an instrument to ensure that interaction
processes have these qualities. Network management is the means by which the quality
152
Public Management: an international journal of research and theory
________________________________________
and openness of processes can be achieved. If the representation of the common
interest is dened in this way, there is no reason to persist in the claim that entering
into network-oriented processes is contrary to the representation of the common
interest. Quite the contrary: because it is the task of governmental organizations to
uphold and further the common interest, they should, rather than refraining from
network games, actively seek to organize and manage them.
However, there remains the criticism that the position of representative bodies will
be eroded by engaging in negotiation processes with private and semi-private partners
(Rhodes 1997). Because of the non-transparent and uncontrolled processes in
networks, the primacy of politics will be threatened. But then again, the position can
be taken that by responsible and prudent engagement in networks, government can
help prevent processes from succumbing to these shortcomings. The network
approach, with its ideas about process management and network constitution, seeks to
contribute to the improvement of the open and democratic nature of interaction
processes in policy networks.
Public management in networks: Roles for governmental organizations
When confronted with a network-like situation, governments may choose among the
following options.
First, they may choose not to join in network games. This means that they will try
instead to unilaterally impose their ideas and goals on other social actors. This will
require a huge investment in decision-making and implementation activities since there
are existing dependencies that will need to be dealt with and the power of the
opposition will need to be broken. The risks are high: is there sufcient and stable
political support for such a strategy? How sure can we be that goal attainment means
effectiveness and efciency, given that policy development is based on imperfect
information and that the strategic behaviour of target groups must be taken into
account? And what does this mean for relations with parties on whom governments
remain dependent both in the future and in parallel situations?
Second, governments may decide to carry out their tasks in co-operation with other
public, semi-public and private actors. Often, entering into dialogue with non-
governmental organizations is considered quite legitimate and a standard operating
procedure. We explicitly mention co-operation with other public actors because it is
entirely possible that various governmental organizations, in performing their tasks,
discover that they are dependent upon each other. But not every form of co-operation
is acceptable or manageable. For instance, hierarchical supervisory relations between
public actors may limit the possibilities of horizontal co-operation.
Third, government can take up the role of process manager and try to facilitate
interaction processes aimed at the resolution of certain problems or the realization of
projects. The fact that government is supposed to protect the common interest,
Klijn & Koppenjan: Public management and policy networks
153
________________________________________
safeguard democratic values and be publicly accountable for its actions frequently
makes it acceptable to others as a process manager. However, it is not always possible
or sensible for governmental organizations to accept such a role. For instance, the
Dutch Ministry of Housing, Physical Planning and Environment was project leader of
the planning process regarding expansion of the national airport in the second half of
the 1980s. It was very difcult for the ministry to protect environmental interests
while it was supposed to be performing the role of a non-partisan process manager.
Fourth, governments may choose to take up the role of network builder. Given the
special resources of governments and their role as representative of the common
interest, governments seem to be eminently suited for this role. But at the same time,
strategies aimed at changing network features have to themselves be handled in game
situations and need to be negotiated with other involved parties in order to result in
stable network changes. These changes cannot be achieved instantaneously. This means
that network constitution is not instrumental to the realization of substantive
government goals in concrete game situations (compare Hood and Jackson 1991). A
serious danger in operating within games for governments is that these four roles get
confused. This may occur for strategic reasons, or if government is inexperienced with
a new role and, in the middle of a difcult situation, reverts to old routines. Ten
Heuvelhof (1993) mentions the risk of alternating between the roles of imposing and
negotiation. We have already referred to the danger of pursuing a substantive objective
while acting as network manager. Also, network constitution and process management
are not necessarily compatible. Clearly confusion of roles can lead to misunderstand-
ings and conict among actors and can prove to be costly in terms of effectiveness and
ef ciency, but especially with regard to the reliability and legitimacy of govern-
ment.

CONCLUSION
In this article, we have argued that the policy network approach has developed into a
relatively elaborate, empirically grounded and recognizable theoretical framework.
With the help of this framework, policy processes can be analysed, explained and
evaluated. And it offers clues on which prescriptions regarding strategies, game
management and network structuring can be based.
Despite this scholarly and substantial work, the network approach can hardly be
considered to be widely accepted as a theory on which practitioners in the public
sector base their actions. It is therefore not surprising that the descriptive and
explanatory aspects of the theory until now have received more attention than its
prescriptions. However, we believe that network theory will prove to be an important
source of inspiration for the development of public management. The nature of tasks
that governments in contemporary complex societies are confronted with will not
allow for command and control reactions. Because of the ambiguity and complexity of
154
Public Management: an international journal of research and theory
________________________________________
these tasks, governments will have to learn to enter into partnerships with other
parties. Network management strategies will have to become part of their standard
operating procedures.
In The Netherlands, for instance, this view is widely recognized and experiments
have even been started to develop public policies in co-operation with politicians, civil
servants, private companies, pressure groups and citizens. But at the same time there
is a hesitation and resistance to abandoning existing routines and to giving up the
power to determine the content of policies unilaterally. As a result these experiments
often remain marginal and half-hearted.
So, where do we go from here? We think it is important that the content, conditions
and implications of the network approach as a prescriptive theory for public
management be further developed, tested and evaluated. In the rst place this will
make it possible to further elaborate network management strategies on the level of
concrete techniques and supporting instruments. Until now this part of the network
approach has been the subject of mostly conceptual and theoretical work. It is now
time to apply these ideas in practice and to test them empirically. This new phase will
open up a whole new range of practical challenges and research questions.
A second theme that deserves attention is that of the institutional conditions for the
introduction of network management strategies. What are the institutional barriers and
how can they be removed? For instance, what are the positive and negative incentives
for interorganizational co-operation that exist within governmental organizations? And
what interorganizational arrangements for mediation and arbitrage are available within
policy sectors at the different governmental levels? This institutional theme brings us
also to the potential of international comparative studies. Is network theory a typical
product of countries with coalition governments, a strong consensual political culture
and a decentralized state system? Can it be applied in unitary states with a majority
system? If so, are network management strategies universal or subject to path
dependent developments?
Third, there is the question of the implications of applying network theory.
Applying ideas of network management means a rede nition of the traditional roles of
politicians, civil servants, interest group involvement and citizen participation. We
mentioned earlier the need to rede ne the concept of the common interest. The same
is true for ideas about the primacy of politics and accountability. Does network
management mean the end of politics or can politicians develop new leading roles?
Does co-operation between governments and other parties result in the blurring of
responsibilities, or can new arrangements and procedures be found which allow for
both co-operation and accountability?
NOTE
1
We dene policy networks as a (more or less) stable pattern of social relations between
interdependent actors, which take shape around policy problems and/or policy programmes.
Klijn & Koppenjan: Public management and policy networks
155
________________________________________
REFERENCES
Aldrich, H. A. (1979) Organisations and Environments, Englewood Cliffs, NJ: Prentice-Hall.
Aldrich, H. And Whetten, D. A. (1981) ‘Organisation-Sets, Action-Sets and Networks: Making The
Most out of Simplicity’, in P. C. Nystrom and W. H. Starbuck (eds) Handbook of Organizational
Design. Oxford: Oxford University Press.
Allison, G. T. (1971) The Essence of Decision, Little Boston, MA: Brown & Company.
Bachrach P. And Baratz, M. S. (1962) ‘Two Faces of Power’. American Political Science Review, 56:4
pp947–52.
Barney, J. B. And Hesterly, W. (1996) ‘Organisational Economics: Understanding the Relationship
Between Organisations and Economic Analysis’ in S. R. Clegg, C. Hardy and W. R. Nord (eds)
Handbook of Organisation Studies. London: Sage.
Benson, J. K. (1982) ‘A Framework for Policy Analysis’ in D. L. Rogers and D. A. Whetten (eds)
Interorganisational Co-ordination: Theory, Research, and Implementation. Ames: Iowa State University
Press.
Blom-Hansen, J. (1997) ‘A New Institutional Perspective on Policy Networks’. Public Administration,
75:winter pp669–93.
—— (1998) ‘A “New Institutional” Perspective on Policy Networks’. Public Administration, 75:winter
pp115–693.
Borzel, T. A. (1998) ‘Organising Babylon – on the Different Conceptions of Policy Networks’. Public
Administration, 76:summer pp253–73.
Brans, M. (1997) ‘Challenges to the Practice and Theory of Public Administration in Europe’. Journal
of Theoretical Politics, 9:3 pp389–415.
Burns, T. R. And Flam, H. (1987) The Shaping of Social Organisation: Social Rule System Theory with
Application. London: Sage.
Cohen, M. D., March, J. G. And Olsen, J. P. (1972) ‘A Garbage Can Model of Organisational
Choice’. Administrative Science Quarterly, 17:1 pp1–25.
Cook, K. S. (1977) ‘Exchange and Power in Networks of Interorganisational Relations’. The Sociological
Quarterly, 18:1 pp62–82.
Crozier, M. And Friedberg, E. (1980) Actors and Systems; The Politics of Collective Action, Chicago,
IL/London: University of Chicago Press.
De Bruijn, J. A. And Ringeling, A. B. (1997) ‘Normative Notes Perspectives on Networks’ in W. J. M.
Kickert, E. H. Klijn and J. F. M. Koppenjan (eds) Managing Complex Networks. London: Sage.
De Bruijn, J. A., Ten Heuvelhof, E. F. And In’t Veld, R. J. (1998) Procesmanagement. Over Procesontwerp
en Besluitvorming. Schoonhoven: Academic Services.
Dery, D. (1984) Problem Denition in Policy Analysis, Kansas, KS: University Press of Kansas.
Dowding, K. (1995) ‘Model or Metaphor? A Critical Review of the Policy Network Approach’.
Political Studies, XLIII pp136–58.
Forester, J. (1989) Planning in the Phase of Power, Berkeley, CA: University of California Press.
Friend J. K., Power, J. M. And Yewlett, C. J. L. (1974) Public Planning: The Inter-Corporate Dimension,
London: Travistock Publications.
Giddens, A. (1984) The Constitution of Society: Outline of the Theory of Structuration, London/Berkeley,
CA/Los Angeles, CA: Macmillan.
Glasbergen, P. ed. (1995) ‘Managing Environmental Disputes’, in Network Management as an Alternative.
Dordrecht/Boston, MA/London: Kluwer Academic Publishers.
Hanf, K. And Scharpf, F. W. eds (1978) Interorganisational Policy Making, London: Sage.
Hindmoor, A. (1998) ‘The Importance of Being Trusted: Transaction Costs and Policy Network
Theory’. Public Administration, 76:spring pp25–43.
156
Public Management: an international journal of research and theory
________________________________________
Hirst, P. (1994) Associative Democracy: New Forms of Economic and Social Governance, Cambridge: Polity
Press.
Hjern, B. And Porter, D. O. (1981) ‘Implementation Structures: A New Unit for Administrative
Analysis. Organisational Studies, 3 pp211–37.
Hood, C. C. And Jackson, M. (1991) Administrative Argument, Aldershot: Dartmouth.
Hufen, J. A. M. And Ringeling, A. B. (eds) (1990) Beleidsnetwerken; Overheids- Semi-Overheids- en
Particuliere Organisaties in Wisselwerking, ’s-Gravenhage: VUGA.
Kickert, W. J. M. ed. (1997) Public Management and Administrative Reform inWestern Europe, Cheltenham:
Edward Elgar Publishing.
Kickert, W. J. M., Klijn, E. H. And Koppenjan, J. F. M. eds (1997) Managing Complex Networks,
London: Sage.
Klijn, E. H. (1996a) Regels en Sturing in Netwerken. De Invloed van Netwerkregels op de Herstructuring van
Naoorlogse Wijken, Delft: Eburon.
—— (1996b) ‘Analysing and Managing Policy Processes in Complex Networks: A Theoretical
Examination of the Concept Policy Network and its Problems’. Administration and Society, 28:1
pp90–119
—— (1997) ‘Policy Networks: An Overview’ in W. J. M. Kickert, E. H. Klijn and J. F. M.
Koppenjan (eds) Managing Complex Networks. London: Sage.
Klijn, E. H., Koppenjan, J. F. M. And Termeer, C. J. A. M. (1995) ‘Managing Networks in the Public
Sector’. Public Administration, 73:3 pp437–54.
Knight, J. (1992) Institutions and Social Con ict, Cambridge: Cambridge University Press.
Kooiman, J. ed. (1993) Modern Governance: New Government-Society Interactions, London: Sage.
Laumann, E. O. And Knoke, D. (1987) The Organisational State: Social Choice in National Policy Domains,
Wisconsin, WI: University of Wisconsin Press.
Levine, S., White, P. E. (1961) ‘Exchange as a Conceptual Framework for the Study of
Interorganisational Relationships’. Administration Science Quarterly, 5 pp583–601.
Lindblom, C. E. (1965) The Intelligence of Democracy: Decision Making through Mutual Adjustment, London:
Free Press.
Lindblom, C. E. And Cohen, D. K. (1979) Usable Knowledge: Social Science and Social Problem Solving,
New Haven, CT/London: Yale University Press.
Mandell, M. P. (1990) ‘Network Management: Strategic Behaviour in the Public Sector’ in R. W.
Gage and M. P. Mandell (eds) Strategies for Managing Intergovernmental Policies and Networks, New
York: Praeger.
March, G. (1988) ‘The Technology of Foolishness’ in J. G. March (ed.) Decisions and Organisations.
Oxford: Basil Blackwell.
March, J. G. And Olsen, J. P. (1989) Rediscovering Institutions: The Organisational Basis of Politics, New
York: Free Press.
Marin, B. And Mayentz, R. eds (1991) Policy Networks: Empirical Evidence and Theoretical Considerations,
New York: Free Press.
Marsh, D. And Rhodes, R. A. W. (1992) Policy Networks in British Government, Oxford: Clarendon
Press.
Milward, H. B. And Wamsley, G. L. (1985) ‘Policy Subsystems, Networks and the Tools of Public
Management’ in K. Hanf and Th. A. J. Toonen Policy Implementation in Federal and Unitary
Systems, Dordrecht: Martinus Nijhoff Publishers.
Negandhi, A. R. ed. (1975) Interorganisation Theory, Kansas City: Kansas University Press.
Osborne, D. And Gaebler, T. (1992) Reinventing Government: How the Entrepreneurial Spirit is Transforming
the Public Sector, Reading, MA: Addison-Wesley.
Klijn & Koppenjan: Public management and policy networks
157
Ostrom, E. (1986) ‘A Method of Institutional Analysis’ in F.-X. Kaufman, G. Majone, V. Ostrom (eds)
Guidance and Control in the Public Sector: the Bielefeld Interdisciplinary Project. Berlin: De Gruyter.
Ostrom, E., Gardner, R. And Walker, J. (1994) Rules, Games and Common Pool Resources, Ann Arbor,
MI: University of Michigan Press.
O’Toole, L. J. (1988) ‘Strategies for Intergovernmental Management: Implementing Programs in
Interorganisational Networks’. Journal of Public Administration, 25:1 pp43–57.
—— (1997) ‘Treating Networks Seriously: Practical and Research-Based Agendas in Public
Administration’. Public Administration Review, 57:1 pp45–52.
Pollitt, C. (1990) Managerialism and the Public Services: The Anglo-American Experience, Oxford: Blackwell.
Propper, I. M. A. M. (1996) ‘Success and Failure in the Management of Policy Networks’.
Beleidswetenschap, 10:4 pp345–65 (in Dutch).
Provan, K. G. And Milward, H. Brinton (1995) ‘A Preliminary Theory of Interorganisational Network
Effectiveness: A Comparative Study of Four Community Mental Health Systems’. Administration
Science Quarterly, 40: March pp1–33.
Rein, M. And Schon, D. (1992) ‘Reframing Policy Discourse’ in F. Fischer and J. Forester (eds) The
Argumentative Turn in Policy Analysis and Planning, Durham, NC: Duke University Press.
Rhodes, R. A. W. (1981) Control and Power in Central-Local Government Relations, Farnborough: Gower.
—— (1988) Beyond Westminster andWhitehall: The Subsectoral Governments of Britain, London: Unwin
Hyman.
—— (1996) ‘The New Governance: Governing without Government’. Political Studies, 44:4
pp652–67.
—— (1997) Understanding Government, Buckingham: Open University Press.
Ripley, R. B. And Franklin, G. (1987) Congress, the Bureaucracy and Public Policy, Homewood, IL:
Dorsey. ( rst published 1976)
Salancik, G. R. (1995) ‘Wanted: A Good Network Theory of Organization’. Administration Science
Quarterly, 40:June pp345–9.
Scharpf, F. W. (1978) ‘Interorganizational Policy Studies: Issues, Concepts and Perspectives’, in K.
Hanf and F. W. Scharpf (eds) Interorganisational Policy Making. London: Sage.
—— (1997) Games Real Actors Play: Actor Centred Institutionalism in Policy Research, Boulder, CO:
Westview Press.
Scharpf, F. W., Reissert, B. And Schabel, F. (1978) ‘Policy Effectiveness and Conict Avoidance in
Intergovernmental Policy Formation’, in K. Hanf and F. W. Scharpf (eds) Interorganisational
Policy Making. London: Sage.
Susskind, L. And Cruikshank, J. (1987) Breaking the Impasse: Consensual Approaches Resolving Public
Disputes, New York: Basic Books.
Teisman, G. R. ( 1992/1995  1992/1995 ) Complexe Besluitvorming: Een Pluricentrisch Perspectief op Besluitvorming over
Ruimtelijke Investeringen, ‘s-Gravenhage: VUGA.
Ten Heuvelhof, E. F. (1993) Gedragsnormen voor Overheden in Horizontale Structuren, ‘s-Gravenhage:
VUGA.
Weick, K. E. (1979) The Social Psychology of Organising (2nd edn), New York: Random House.
Wilks, S. And Wright, M. (1987) Comparative Government Industry Relations, Oxford: Oxford University
Press.
Williamson, O. E. (1979) ‘Transaction Costs Economics: The Governance of Contractual Relations’.
Journal of Law and Economics, XXII:2 pp233–61.
158
Public Management: an international journal of research and theory
..................................................... 

..................................................... 

Foundations of a network approach to governance
Erik-Hans Klijn and
Joop F. M. Koppenjan

E. H. Klijn
Department of Public
Administration
Erasmus University
Rotterdam
P.O. Box 1738
3000 BA Rotterdam, The
Netherlands
Tel:  [phone]  [phone]
Fax: [phone]
E-mail: [e-mail]
 J. F. M. Koppenjan
Faculty of Technology, Policy
and Management
Delft University of
Technology
Jaffalaan 5
2628 BX Delft, The
Netherlands
Tel:  [phone]  [phone]
Fax: [phone]
E-mail: J.F.M.Koppenjan@
sepa.tudelft.nl
Vol. 2 Issue 2 2000 135–158
Public Management ISSN 1461–667X print/ISSN  1470–1065  1470–1065 online
© 2000 Taylor & Francis Ltd http://www.tandf.co.uk/journals