You are on page 1of 525


qxd 6/8/2006 7:36 PM Page i

Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page ii
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page iii


Edited by

SAGE Publications
London ● Thousand Oaks ● New Delhi
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page iv

Introduction © B. Guy Peters and Jon Pierre 2006 Chapter 15 © Kevin V. Mulcahy 2006
Chapter 1 © Peter L. Hupe and Michael J. Hill 2006 Chapter 16 © B. Guy Peters 2006
Chapter 2 © Peter DeLeon and Christine R. Chapter 17 © Richard D. Bingham 2006
Martell 2006 Chapter 18 © Wyn Grant 2006
Chapter 3 © Bryan D. Jones, Graeme Boushey and Chapter 19 © Kenneth Button 2006
Samuel Workman 2006 Chapter 20 © Walter Carlsnaes 2006
Chapter 4 © Davis B. Bobrow 2006 Chapter 21 © Tim Newburn 2006
Chapter 5 © Peter Bogason 2006 Chapter 22 © Ian Thynne 2006
Chapter 6 © B. Guy Peters 2006 Chapter 23 © Evert Vedung 2006
Chapter 7 © Irene S. Rubin 2006 Chapter 24 © Aiden R. Vining and David
Chapter 8 © Søren C. Winter 2006 L. Weimer 2006
Chapter 9 © John Uhr 2006 Chapter 25 © Gary Bryner 2006
Chapter 10 © Helen Fawcett 2006 Chapter 26 © Geert Bouckaert and John
Chapter 11 © Harold L. Wilensky 2006 Halligan 2006
Chapter 12 © Michael Moran 2006 Chapter 27 © Herbert Gottweis 2006
Chapter 13 © Susan Marton 2006 Chapter 28 © Jon Pierre 2006
Chapter 14 © Christoph Knill 2006

First published 2006

Apart from any fair dealing for the purposes of research or

private study, or criticism or review, as permitted under the
Copyright, Designs and Patents Act, 1988, this publication may
be reproduced, stored or transmitted in any form, or by any means,
only with the prior permission in writing of the publishers, or in
the case of reprographic reproduction, in accordance with the
terms of licences issued by the Copyright Licensing Agency.
Enquiries concerning reproduction outside those terms should be
sent to the publishers.

SAGE Publications Ltd

1 Oliver’s Yard
55 City Road
London EC1Y 1SP

SAGE Publications Inc.

2455 Teller Road
Thousand Oaks, California 91320

SAGE Publications India Pvt Ltd

B-42, Panchsheel Enclave
Post Box 4109
New Delhi 110 017

British Library Cataloguing in Publication data

A catalogue record for this book is available from

the British Library
ISBN-10 0-7619-4061-8 ISBN-13 978-0-7619-4061-6

Library of Congress Control Number available

Typeset by C&M Digitals (P) Ltd., Chennai, India

Printed in Great Britain by Cromwell Press Ltd, Trowbridge, Wiltshire
Printer on paper from sustainable resources
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page v


List of Contributors ix

Introduction 1
B. Guy Peters and Jon Pierre

Section 1: Making Policy 11

1 The Three Action Levels of Governance: Re-framing the Policy Process

Beyond the Stages Model 13
Peter L. Hupe and Michael J. Hill

2 The Policy Sciences: Past, Present, and Future 31

Peter DeLeon and Christine R. Martell

3 Behavioral Rationality and the Policy Processes: Toward A New Model

of Organizational Information Processing 49
Bryan D. Jones, Graeme Boushey and Samuel Workman

4 Policy Design: Ubiquitous, Necessary and Difficult 75

Davis B. Bobrow

5 Networks and Bargaining in Policy Analysis 97

Peter Bogason

6 Concepts and Theories of Horizontal Policy Management 115

B. Guy Peters

7 Budgeting 139
Irene S. Rubin

8 Implementation 151
Søren C. Winter

Section 2: Substantive Policy Areas 167

9 Constitution and Rights 169

John Uhr
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page vi


10 Social Policy: Pensions 187

Helen Fawcett

11 Social Policy: Is there a Crisis of the Welfare State? 201

Harold L. Wilensky

12 Health Policy 219

Michael Moran

13 Education Policy 231

Susan Marton

14 Environmental Policy 249

Christoph Knill

15 Cultural Policy 265

Kevin V. Mulcahy

16 Tax Policy 281

B. Guy Peters

17 Industrial Policy in Developed Nations 293

Richard D. Bingham

18 Agriculture and Food 309

Wyn Grant

19 Transportation and Infrastructure 323

Kenneth Button

20 Foreign Policy 339

Walter Carlsnaes

21 Criminal Justice Policy 365

Tim Newburn

22 Privatisation by Divestment 381

Ian Thynne

Section 3: Evaluating Policy 395

23 Evaluation Research 397
Evert Vedung

24 Efficiency and Cost-Benefit Analysis 417

Aidan R. Vining and David L. Weimer
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page vii


25 Ethics and Public Policy 433

Gary Bryner

26 Performance and Performance Management 443

Geert Bouckaert and John Halligan

27 Argumentative Policy Analysis 461

Herbert Gottweis

28 Disciplinary Perspectives 481

Jon Pierre

Index 493
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page viii
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page ix

List of Contributors

Richard D. Bingham is a Professor of Public Administration and Urban Studies at

the Maxine Goodman Levin College of Urban Affairs, Cleveland State University,

Davis B. Bobrow is Professor at the Graduate School of Public and International

Affairs, University of Pittsburgh, USA.

Peter Bogason is Professor of Public Administration at the Department of Social

Sciences, Roskilde University, Denmark.

Geert Bouckaert is Professor and Director of the Public Management Institute,

Katholieke Universiteit Leuven, Belgium.

Graeme Boushey is a Graduate Research Fellow in the Center for American Politics
and Public Policy at the University of Washington, USA.

Gary Bryner is Professor at the Department of Political Science, Brigham Young

University, USA

Kenneth Button is Professor of Public Policy at the School of Public Policy, George
Mason University, USA

Walter Carlsnaes is Professor of Political Science at the Department of

Government, Uppsala University, Sweden.

Peter DeLeon is Professor at the Graduate School of Public Affairs, University of

Colorado at Denver, USA.

Helen Fawcett is Lecturer at the Department of Government, University of

Strathclyde, UK.

Herbert Gottweis is Professor at the Department of Political Science, University of

Vienna, Austria.

Wyn Grant is Professor in the Department of Politics and International Studies at

the University of Warwick, UK

John Halligan is Professor at the School of Management and Policy, University of

Canberra, Australia.
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page x


Michael Hill is Emeritus Professor, University of Newcastle upon Tyne, UK.

Peter L. Hupe is Associate Professor in the Department of Public Administration

at Erasmus University, Rotterdam, The Netherlands.

Bryan D. Jones is Donald R. Matthews Distinguished Professor of American

Politics, and Director of the Center for American Politics and Public Policy at the
University of Washington, USA.

Christoph Knill is Professor in the Department of Political Science and

Administration, University of Konstanz, Germany.

Christine R. Martell is Assistant Professor at the Graduate School of Public Affairs,

University of Colorado at Denver, USA.

Susan Marton is Assistant Professor at the Department of Political Science,

Karlstad University, Sweden.

Michael Moran is Professor of Government in the Department of Government

and International Politics, University of Manchester, UK.

Kevin V. Mulcahy is Sheldon Beychok Distinguished Professor at the Department

of Political Science, Louisiana State University, USA

Tim Newburn is Professor of Criminology and Social Policy and Director,

Mannheim Centre for Criminology, London School of Economics, UK.

B. Guy Peters is Maurice Falk Professor of Government in the Department of

Political Science at the University of Pittsburgh, USA.

Jon Pierre is Professor in the Department of Political Science at the University of

Gothenburg, Sweden.

Irene S. Rubin is Professor Emeritus of Political Science in the Division of Public

Administration at Northern Illinois University, Dekalb, USA.

Ian Thynne is in the Governance Program, Faculty of Law, Business and Arts,
Charles Darwin University, Australia.

John Uhr is Senior Fellow in the Political Science Program, Research School of
Social Sciences, at the Australian National University, Canberra.

David L. Weimer is Professor of Public Affairs and Political Science at the

University of Wisconsin-Madison, USA.
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page xi


Harold L. Wilensky is Professor Emeritus of Political Science at the Department of

Political Science, University of California at Berkeley, USA.

Søren C. Winter is Research Professor, Danish National Institute of Social

Research, Copenhagen, Denmark.

Samuel Workman is a Graduate Research Fellow in the Center for American

Politics and Public Policy at the University of Washington, USA.

Evert Verdung is Professor of Political Science at the Institute for Housing and
Urban Research (IBF) and Department of Government, Uppsala University,

Aiden R. Vining is the CNABS Professor of Government and Business Relations,

Faculty of Business Administration, Simon Fraser University, Vancouver, Canada.
Peter-3383-Prelims.qxd 6/8/2006 7:36 PM Page xii
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 1



The Handbook of Public Policy is an attempt the academic aspects of this area of inquiry.
to cover the area of public policy studies. That Practitioners generally want answers to their
is an ambitious goal, and perhaps a too ambi- day to day problems, are little concerned with
tious one. Both the real world of public policy theory, and think that academics are hopelessly
and the academic study of policy are sizeable, enisled in their ivory towers. On the other
complex and differentiated bodies of practice hand, however, academics often argue that the
and of knowledge. If anything the body of experts in particular policy areas are obsessed
information that any treatment of public with those policies and cannot see the bigger
policy must attempt to cover has been increas- political and economic picture about the
ing. On the one hand, governments have impact and meaning of those policies. We have
altered the manner in which they consider attempted to balance these various perspectives
policies substantively, changing some estab- within the Handbook as a whole, but relatively
lished fields such as economic policy to few of the individual chapters have been able to
become “competitiveness policy”, and/or “sus- strike the balance between these two poles.
tainable development policy”, and also are
attempting to integrate better the range of
policies that have existed in the past to gener-
ate more strategic approaches to governing.
The academic study of public policy has also
been expanding to include a wider range of The study of public policy is a very complex
academic disciplines and approaches. For topic, and any attempt to force policy into any
example, anthropology has become a more narrow theoretical frame should be considered
central player in understanding the processes with some skepticism. On the one hand there
by which policies are selected, as well as under- are some real virtues for policy as an area of
standing some substantive aspects of policy inquiry for the social sciences, given that it is
such as those directed toward immigrants. amenable to so many different perspectives.
Further, the range of theoretical approaches to On the other hand, however, this complexity
public policy has expanded as the limits of the requires bringing together a wide range of
more conventional ways of thinking about theoretical and analytical perspectives to gain
policy have become more apparent. For exam- any sort of understanding of what is happen-
ple, the chapter on constructivist approaches ing in any policy area. Both academic disci-
to policy (Gottweis, this volume) reflects the plines and substantive policy concerns tend to
emergence of a strong strand of theorizing that narrow the vision and to limit the ability of
has been developing to supplement, and to analysts to understand the underlying com-
contradict, to more rationalist perspectives on plexity of most policies.
policy.1 The typical manner of approaching public
The final major question that faces policy policy is to consider the various areas of gov-
studies is how to put together the practical and ernment activity one by one – health, defense,
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 2


taxation. Studies of that sort are certainly especially in Northern European countries in
valuable, but the first important aspect of the which the policy networks (Sorenson and
complexity of public policy is that focussing Torfing, 2003) and corporatism (Wiarda,
on the “single lonely policy” may vastly over- 1997) institutionalize the involvement of a
simplify the interactions of multiple policies in variety of social actors in making and imple-
producing outcomes for citizens. For example, menting public (sic?) policies. Even in the less-
if government wants to improve the quality of developed countries there is growing evidence
health for its population, the obvious area for of the use of social actors as a means of assist-
investment is in hospitals and other aspects of ing government to make and deliver public
the “health care industry”. On the other hand, policies, and that this involvement of non-
however, improving nutrition, or enhancing government actors has enhanced the legiti-
opportunities for exercise, may actually pro- macy of the State.
duce greater health benefits. Therefore, there is A final complexity concerns what do we
a need to think carefully about the interactions study when we attempt to study public policy?
of policies and means of coordinating policies Do we focus only on the decisions that are
to create more effective, if more complex, made and the processes that produce them, the
responses to policy problems. programs that governments develop to deliver
A second dimension of complexity in public services and to influence the society, or the
policy studies is the need to examine policy tools that governments use in their implemen-
questions from a range of theoretical perspec- tation? Ultimately, as Harold Lasswell (1935)
tives. For example, in his classic study of inter- pointed out some 70 years ago, that politics is
national politics and foreign policy in the about “who gets what”, and if that is true of
Cuban missile crisis, Graham Allison (1971) politics then it is certainly true for public poli-
discussed the virtues of “triangulation” and cies. Therefore, in the end we should be con-
the use of multiple theoretical perspectives to cerned with the impacts of policy choices on
understand decisions. His more comprehen- people, and the distribution of benefits in the
sive2 approach has rarely been used for other society. As numerous students of government
policy areas, but would make a major contri- performance have pointed out, however,
bution to understanding exactly what is hap- assessing those impacts is difficult. Some
pening in the policy area. The difficulty may be impacts, for example the benefits of education,
that few individuals are able to bring a wide- may not be fully apparent until many years
ranging theoretical grounding for the study of later, and some may be so diffuse (culture and
policy in this manner, so we all focus on what arts policy?) that measurement for other than
we know best and attempt to do it well. academic purposes may not be particularly
Scholars and practitioners may be able to build valuable.4
triangulated understandings of policies by
cumulating the work of individuals using their
own approaches, but this assumes that those
individuals will have examined exactly the
same policies and have made their findings
sufficiently transparent to do that sort of
comparison.3 To gain some understanding of public policy,
A third complexity involved in the study of which – as this volume will show – is both a
public policy arises from the word “public”. very heterogeneous concept and also a field of
There has been a tendency in the analysis of the social sciences that displays a number of
policy to consider primarily, or solely, the role different approaches, we must first remind
of the public sector and official actors in the ourselves of the significance of the substantive
process, and to ignore the role of private sector and analytical dimensions of public policy.
actors. In fairness, that tendency has been less Substantively, public policy has undergone
apparent in recent years than in the past, some rather fundamental changes over the
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 3


past couple of decades in terms of policy about what causes changes in public policy.
design, selection of policy instruments and the The growing interest in policy diffusion and
role of the state in society more broadly. As we policy learning depends in part on its claim to
will discuss later in this Introduction, public be a rewarding avenue into these issues. Most
policy in general has become less intervention- observers would, however, probably argue that
ist, controlling and obtrusive. Instead, there is such patterns of policy transfer among states
today a stronger emphasis of the regulatory offer only a limited explanation to the broader
role of the state. The Zeitgeist of much of the question of policy change. If anything, policy
1980s and 1990s has been that of rolling back transfer probably occurs once a changed
the state, unleashing markets, cutting taxes course of policy has been decided and is not so
and public spending and reducing control. much part of the drive for change per se.
Globalization – real or imagined – has contin- Instead, it appears as if the main sources of
ued to fuel this model of public policy. policy change are either economic or related to
Analytically, a number of approaches have voters’ demands. The economic drivers of
emerged, highlighting different elements of policy change – either in terms of changes in
public policy and offering different types of the tax base and the public budget or in terms
explanations to policy design and efficiency. of global pressures on tax levels and public
This volume presents the leading approaches in spending – represent a powerful pressure on
policy research but it does so without detailing policy makers. Voters’ preferences matter too,
the chronological development of this research but probably less in terms of policy instrument
field. Policy researchers have, rightly, been con- selection and policy design than in the rela-
cerned with process throughout most of the tionship between taxes and benefits for the
post-war period. Indeed, even when institu- main constituencies in society. People are
tional analysis has been the predominant theo- probably not very concerned with how services
retical perspective on public policy there has and programs are delivered as long as they are
been an undercurrent of studies calling for provided.
“process-tracing” to supplement the institu- Most importantly, there is plenty of proof
tional perspective. Process remains extremely that politics matters in shaping policy choice.
important; actors’ assessment of policy chal- That means that governments of different ide-
lenges and various strategies to ameliorate ological orientations tend to make different
those problems today are largely a reflection of choices with respect to how the state should
decisions and assessments made yesterday. allocate its resources and how those resources
Policy evolves through process, hence it is diffi- should be mobilized. These observations hold
cult for studies that ignore process to arrive at a true even in an era where it is often said that
deeper understanding of policy. We will return policy challenges are becoming more similar
to these issues later in this chapter. among different countries, whether because of
The tool kit of policy analysts includes an demographic patterns, economic moderniza-
impressing number of theories, models and tion or global economic pressures (see, for
frameworks. Unlike many other subfields of instance, Castles, 1998). National political and
the social sciences, however, these approaches institutional contexts still offer much in terms
seem to be complementary much more than of explanation of policy choice. Those choices
they are contending, mainly because of many are made in the context of national systems of
different types of questions which scholars structures and values that have proven quite
raise in policy analysis. As a result, much of the resilient in the midst of globalization and
controversy that has evolved in political economic deregulation.
science over rational choice versus other The second question we need to address has
theories never made much of an imprint in the to do with the development of policy analysis
field of public policy. as a social science research field. Public policy
These trends and developments raise two research has been concerned with changes in
sets of questions. First, we need to know more public policy across time and space. This
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 4


applies to the entire range of variables in policy somewhat malfunctioning rifle. The target is
research, from theories of societal problems to changing course frequently – or does not
selection of policy instruments. Students of change course when we expect it to – and our
public policy, like all social scientists, have fads weapon is far from perfect in composition and
and fashions with respect to what theories and robustness. However, it is safe to say that the
models are “in” and which are not. diversity of approaches to public policy that we
It could well be argued that institutional have seen develop has helped us understand
approaches, ideational approaches or, indeed, public policy to a greater extent than would
most or all of the approaches covered in this have been the case had the discipline been
Handbook are to a large extent more a reflec- more homogenous.
tion of changes in the preferences among the
observers of public policy than a reflection of
changes in policy itself. That having been said,
however, these linkages do exist between the
dimensions, i.e. between what is studied and
how it is studied. Thus, for instance, the grow-
ing interest in implementation followed fairly The debate among students of public policy
logically on the expansion of federal programs about how to best approach this topic has
in the mid- and late 1960s under the heading evolved alongside significant developments in
of the Strong Society. Similarly, there has been public policy itself. A quick look back in time
an increasing interest among scholars in regu- suggests that over the past several decades we
lation, or even in the emergence of a “regula- have seen profound changes in policy objec-
tory state”, partly as a result of states’ increasing tives and instruments. The 1960s and 1970s
use of regulation and less emphasis on using saw most Western governments expanding the
more costly policy instruments. public sector. The preferred policy instruments
Another example of the linkage between tended to be fairly coercive. New public pro-
substance and theory can be found in the study grams were launched to address societal prob-
of policy change. Adaptation to changing eco- lems. The growth in the economy generated
nomic preconditions is a surprisingly slow increasing incomes to the state which provided
process, and although politics matters in the the financial base for expanding programs in
aggregate, changes in individual programs can sectors like education, social welfare, housing
be a frustrating experience for incoming gov- policy and research and development. With
ernments because of the inertia in the public only slight exaggeration we could say that the
sector. This has drawn scholars’ attention to Western societies of the 1960s and 1970s were
the significance of different types of obstacles more state-centric compared to those of the
to policy changes. Such obstacles are typically 1980s and 1990s. This applies – with signifi-
believed to be institutions or processes which, cant variation – to jurisdictions on both sides
de facto if not de jure, lock in or protect areas of the Atlantic; in the case of the United States
of public spending from reassessment. Again, we need only compare the Strong Society pro-
it appears as if the increasing interest in insti- grams under Lyndon Johnson with the more
tutions has generated a broader and more market-embracing policy style of Reagan to
inclusive understanding of what shapes and note how much has changed in a rather short
sustains public policy. In some ways, it could period of time.
be that social scientists are just beginning to Although many policy objectives have not
discover what practitioners in the policy been abandoned, contemporary public policy
process have known for a long time; politics tends to draw on other instruments and insti-
matters but institutions mitigate the impact of tutional models compared to twenty years ago.
changes in the political leadership. Indeed, one could argue that in some cases the
Thus, the study of public policy could be social theory underlying public policy has been
described as shooting at a mobile target with a reassessed. In the early post-war period, the
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 5


predominant social theory of public policy beyond that role should be carefully assessed
emphasized the positive contribution of policy and, if possible, transferred to the market or to
and state action to resolve societal problems. In civil society. In the United States, the Reagan
the contemporary policy vernacular, there is a administration sought to rid the federal level
much greater awareness of the contingencies of all responsibilities and functions which
and complexities surrounding policy design could not be defined as part of the core role of
and implementation. government. Similarly, in Britain, Mrs. Thatcher
Looking more broadly at these issues, there embarked on a project aiming at drastically
seem to be three types of changes that summa- reducing the scope of government in society
rize the developments over the past couple of and to “unleash” the market. While these
decades. The first of these changes is a shift types of reform helped the state to curb budget
away from a policy style characterized by com- deficits and to stimulate the economy, they also
mand and control by the state towards a regu- caused considerable political friction as con-
latory state (Moran, 2003). The budgetary stituencies deprived of financial support took
cutbacks in welfare programs that represent a to political arms.
rather uniform pattern across the western The third and final change in policy style
hemisphere have been accompanied by a grow- which we will discuss here is in many ways
ing emphasis on the regulatory role of the derived from the former two; the so-called
state. This development has also been driven shift from government to governance. Some-
by the tendency to open up for market-based times labelled “the new governance”, this devel-
systems of resource allocations and the rela- opment refers to a tendency among most west
tionship between the providers and consumers European governments during the late 1980s
of public services. Thus, if the previously pre- and 1990s to produce and deliver public
dominant policy emphasized political solu- service in concert with the market and civil
tions to societal problems, the current strategy society. The role of the state in the governance
is more based on a notion of rolling back the model is not to produce all those services itself
state, allowing the market to play a greater role but rather to coordinate public and private
in society. action so as to ensure that those services and
Another overarching development in policy programs are delivered. Collective goals and
style could be described as “back to basics”. objectives remain defined by political institu-
During the first couple of decades following tions but the pursuit of those goals needs
World War II, most western states – as well as not be a matter solely for the state. Thus, the
subnational institutions – assumed roles and governance perspective highlights concerted
responsibilities that could not be said to action, shared resources and negotiation as an
belong to the core roles of the state in society. alternative to the state-centric model which
Thus, many states owned enterprises providing features a clear separation of state and society
collective good such as electricity, railway and a state strategy primarily based on com-
transportation, telecommunication services, mand and control.
and the like. Also, states provided financial The increasing interest in “new governance”
support to an array of functions and activities is part of a cluster of issues which we and many
in civil society and the culture sector. The others have discussed elsewhere (Pierre and
1980s and 1990s saw somewhat of a reversal of Peters, 2000, 2004; Pierre, 2000; see also Kjaer,
this expansion of the scope of the state. The 2004; Rhodes, 1997). The significance of the
previously mentioned stronger emphasis on governance model in the current analysis is
the regulatory role of the state is one element that it represents a very different foundation of
of this “purification” of the state (see Premfors, public policy compared to the traditional gov-
1999). The state should no longer own enter- ernment model. In a governance perspective,
prises, not even if the services produced were public policy is to a large extent the outcome of
collective goods. Instead, the state should bargaining among political institutions and
return to its core role in society; any function societal actors. Students of neo-corporatism
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 6


find little news in this observation and, to be designed and implemented. By studying the
sure, what has been labelled “new governance” different stages of the policy process – agenda
is not very new in many European national setting, decision making, legitimation, imple-
contexts. What does appear to be novel about mentation, and so on – observers develop an
the model of governance which evolved during understanding of why policy looks the way it
the 1990s is that it involves the market to a does (see, for instance, Dye, 1987; Kingdon,
larger extent than the neo-corporatist model, 1995). Each stage has its own set of actors; for
which mainly saw peak organizations engage instance, decision makers are rarely involved in
in deliberation with the state. If anything, the the implementation of policy. Also, each indi-
“new governance” is more contextual and vidual stage displays its own types of potential
messier than most previous models of gover- conflict and political opportunity structures.
nance. That development poses a significant The overall rationale of the stages approach is
challenge to observers of policy making and that what actors do at one stage of the policy
public policy. While the state remains the process is to a large extent framed by what
undisputed center of representation, account- other actors have done earlier in the process.
ability and coordination, it faces a more com- Historical institutionalists rightly point out
plex environment with which it interacts. that the first stages in any process have a major
These three developments in the role of the influence on what happens next. We should,
state have redefined the range and scope of however, recognize that, although there are
public policy and, more broadly, have refor- powerful lock-in factors in the policy process,
mulated the social theory which public policy there is some room for political and strategic
rests on. That development, in turn, redefines consideration at all the stages of the process.
policy objectives and influences the selection For instance, as the implementation literature
of policy instruments. Most importantly, per- tells us, identical policies can be implemented
haps, is that all models of state-society in a number of different ways – with different
exchanges become institutionalized; citizens outcomes – in different institutional settings.
and organized interests learn what to expect There is a great deal of logic in the process
from the state in terms of problem-solving and approach. What happens at a particular stage is
resource mobilization. In societies with “weak very much shaped by decisions and actions at
states” (Migdal, 1998), expectations on the the previous stages. Policy makers probably
state are low and societal actors develop non- assess policy alternatives and political strategies
political solutions to various problems. in a process stage model and, indeed, the formal
Similarly, in countries with “strong states”, procedure of policy making defines in great
people become socialized into turning to the detail the process. These formal rules of policy
state for help with almost any problem. This making shape actors’ behavior and it therefore
institutionalization makes a transition from a also makes sense for policy analysts to structure
government-centered model of governance to their observations according to that perspective.
a market-based or network-based governance On the other hand, however, individuals also
model difficult and politically quite complex. play a role in the process, and act as policy
entrepreneurs to attempt to have their own
policy preferences enacted into law. Thus, policy
making represents a complex interaction of
individuals, institutions, ideas and interests.
In addition to looking at policy, the process
A key purpose of this volume is to present the by which policies are formed and implemented,
multitude of approaches to public policy. In we can also ask a number of other questions
much of the public policy analysis of the past about policy. One of the more important com-
several decades, the focus has been on the ponents of the study of public policy is the use
process through which public policy is of analytic techniques that can be used to assist
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 7


decision-makers in choosing policies. These which makes competing policy options

techniques are often based on assumptions of unattractive because of high political or eco-
optimality and rationality derived from eco- nomic costs. Thus, institutional analysis has
nomics, and therefore often confront the politi- offered significant help in understanding lock-
cal realities that may make presumably optimal in effects in public policy and different types of
solutions impossible. Rationalist modes of obstacles to policy change.
analysis are not alone, however, and other meth- The institutional approach to policy analysis
ods (reflecting their theoretical underpinnings) thus appears to cut the proverbial pie in a dif-
such as bounded rationality, discourse analysis, ferent way than does the stages model. To be
and humanist approaches to policy, do contend sure, the increasing dominance of institutional
for attention, and offer alternative answers to analysis over the process approach has left
questions about public policy. some researchers looking for analytical models
Yet another means of thinking about policy which combine attention to institutions with
is to focus on the various policy areas and an understanding of the logics of the policy
types of policy. The most obvious way to process qua process (Sabatier, 1999). Although
engage in this form of analysis is to use the the institutional approach draws on a different
familiar functional names of policy – defense, logic than the stages model, there is some over-
agriculture, etc. These names are helpful, and lap between the two approaches; for instance,
there may be as much variance across policy scholars can investigate the role of institutions
areas than there is across countries in the man- in the policy process. That said, institutional
ner in which policies are processed. Still, there analysis has problems understanding agency,
is a great deal of internal variance within each something which the stages model is better
of these areas,5 so that a more analytic way of geared to conceptualize and explain.
thinking about policy types may be needed. Different academic disciplines conceptualize
Theodore Lowi (1972) has supplied the most public policy and policymaking differently.
commonly used of these analytic schemes, but Political science focuses on the processes, insti-
there are other ways of thinking about policy tutions and the power struggles among com-
types and their impact on policy analysis peting interests in having their wishes enacted
(Peters and Hoornbeek, 2005). into policy. Economics is concerned more with
The fourth significant dimension of public the economic effects of policies and with
policy research has been the rapidly growing designing optimal policy solutions. Philosophy
interest in the role of political institutions in is concerned with the profound but knotty
shaping and sustaining public policy. A couple normative questions involved in policy
of seminal studies published during the 1980s choices. This list could be extended, but the
(March and Olsen 1995, 1999; Hall, 1986) basic point is that a number of academic disci-
delivered a powerful argument that public plines bring something to the table when the
policy, state-society relationships and political discussion of public policy begins.
behavior more broadly are “shaped and con- The contribution of different academic dis-
strained” by institutions. Institutions should ciplines also can be assessed with regard to dif-
not only be understood in the narrow, struc- ferent stages of the policy process (see Pierres’s
tural sense; institutions are carriers and trans- chapter in this volume). More broadly, the
mitters of norms and values which define a influence of different disciplines plays out, not
“logic of appropriateness” for political and only in the scholarly debate on how to best
social behavior. Applied to public policy understand public policy, but also in the sub-
research, institutional analysis has highlighted stantive issues that the discipline helps to iden-
the significance of structure and norm in tify. Any unidimensional analysis of policy
defining policy objectives. Once a policy is in therefore should be somewhat suspect,
place, it stays in place. The institutionalization although each scholar will remain a prisoner of
of public policy creates a “path dependency” his or her academic training.
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 8


CONCLUSIONS: THE FUTURE OF 3. The increasing use of Bayesian methods to

PUBLIC POLICY RESEARCH bring together studies (Ragin, 1987) done in different
ways is
4. That having been said, however, arts organizations
The study of public policy is now a well- have been increasingly sophisticated in making arguments
for the economic benefits of their programs and the eco-
established component of several academic nomic utility of public sector subsidies, never mind the
disciplines, as well as having a literature, profes- real cultural benefits.
sional associations, journals, and theories of its 5. Think, for example, of the differences between
own. One of the questions about the future, kindergarten and higher educational policy issues,
therefore, is whether the eclecticism that has although they both are education policy in the functional
characterized the development of this area of
study in the past will persist, or whether there
will be the development of a paradigm for
policy studies. Our own preference would be to
continue with a more eclectic approach, empha-
sizing the range of alternative means of address- Allison, G. T. (1971) Essence of Decision (Boston:
ing the same concerns and learning from those Little, Brown).
multiple perspectives. Castles, F. G. (1998) Comparative Public Policy:
The future of policy studies will almost cer- Patterns of Post-War Transformation (Cheltenham:
tainly be more international and comparative. Edward Elgar).
Dye, T. R. (1987) Understanding Public Policy, 6th
The diffusion of policies among nations, and
Ed. (Englewood Cliffs, NJ: Prentice-Hall).
the importance of international regimes and Fischer, F. (2003) Reframing Public Policy (Oxford:
international organizations on national policy Oxford University Press).
choices, appears only likely to increase. That Hall, P. A. (1986) Governing the Economy: The
having been said, however, we must remain Politics of State Intervention in Britain and
cognizant of the persistence of national policy France (Oxford: Oxford University Press).
styles and national policy problems that can Kingdon, J. W. (1995) Agendas, Alternative and
not be subsumed under a large international Public Policies, 2nd. Ed (New York: Harper
umbrella. Collins).
We hope that this volume will contribute Kjaer, A. M. (2004) Governance (Cambridge: Polity
both to the understanding of public policy of Press).
Lasswell, H. (1935) Politics: Who Gets What, When,
those who come to it as novices, and to the
How (New York: McGraw Hill).
increased understanding of veterans in the Lowi, T. J. (1972) Four Systems of Policy, Politics
area. The coverage is extensive, but could and Choice, Public Administration Review 32,
always have been larger. The range of scholars 298–310.
and the range of issues should make clear that March, J. G. and J. P. Olsen (1989) Rediscovering
this is a multi-faceted and international enter- Institutions (New York: Free Press).
prise, and further that the study of public March, J. G. and J. P. Olsen (1995) Democratic
policy has importance, not just as an academic Governance (New York: Free Press).
enterprise, but also has real impacts on the Migdal, J. S. (1988) Strong Societies and Weak States
lives of citizens. (Princeton, NJ: Princeton University Press).
Moran, M. (2003) The British Regulatory State
(Oxford: Oxford University Press).
Peters, B. G. and J. A. Hoornbeek (2005) The
NOTES Problem of Policy Problems, in P. Eliadia, M. M.
Hill and M. Howlett, eds., Designing Government
1. This approach to policy is not really that new. See (Montreal: McGill-Queens University Press).
Fischer (2003) for an excellent restatement. Pierre, J. (ed.) (2000) Debating Governance
2. Although Allison did use three approaches there may (Oxford: Oxford University Press).
well be still others that could have been used to illuminate Pierre, J. and B. G. Peters (2000), Governance,
these decisions even further. Politics and the State (Basingstoke: Macmillan).
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 9


Pierre, J. and B. G. Peters (2004) Governing Rhodes, R. A. W. (1997) Understanding Governance

Complex Societies: Trajectories and Scenarios (Buckingham: Open University Press).
(Basingstoke: Palgrave). Sabatier, P. A. (1999) Theories of the Policy Process
Premfors, R. (1999) “Organisationsförändringar (Boulder, CO: Westivew).
och förvaltningspolitik – Sverige”, in P. Laegreid Sorenson, E. and J. Torfing (2003) “Network
and O. K. Pedersen (eds), Fra opbyning till Politics, Political Capital and Democracy”,
ombygning i staten: Organisationsforandringer i International Journal of Public Administration
tre nordiske lande (Copenhagen: Jurist-og 26, 609–43.
Ökonomforbundets Forlag), 145–68. Wiarda, H. (1997) Corporatism and Comparative
Ragin, C. (1987) The Comparative Method: Moving Politics: The Other Great “ism” (Armonk, NY:
Beyond Qualitative and Quantitative Strategies M. E. Sharpe).
(Berkeley: University of California Press).
Peter-3383-Introduction.qxd 6/8/2006 7:44 PM Page 10
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 11

Section One
Making Policy
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 12
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 13

The Three Action Levels of
Governance: Re-framing the Policy
Process Beyond the Stages Model


1. INTRODUCTION (c) within a policy domain there are normally

dozens of different programmes involv-
The state of the art ing multiple layers of government;
(d) a variety of debates about the policy
Mountain islands of theoretical structure, intermingled involved, partly of a highly technical
with, and occasionally attached together by foothills of
character and held in different fora;
shared methods and concepts, and empirical work, all of
which is surrounded by oceans of descriptive work not (e) the high stakes involved give rise to ‘poli-
attached to any mountain of theory. tics’ and power political behaviour in and
around a policy process.
Thus Schlager (1997: 14) characterises the
present landscape of the study of the policy These complexities of the object form the
process1. This varied landscape partly mirrors background to the fact that the contemporary
the complexities of the object of study itself. A study of the policy process appears to go in
key element in many approaches to describing various directions. In his textbook Public Policy
this landscape is the stages model of the policy Wayne Parsons gives a broad overview of the
process. In this chapter we will review the conceptual wealth present in the field of policy
functionality of that so called ‘model’. analysis. He calls this field ‘rich in different
Sabatier (1999: 1–2) specifies the elements approaches, academic disciplines, models
of the policy process as ‘an extremely complex (heuristic and causal), metaphors and maps’
set’, consisting of: (Parsons, 1995: 64). One of the striking aspects
of the almost 700 pages of Parsons’ book is not
(a) a multiplicity of actors (both individual only the scope of the insights about the policy
and corporate) each of which have differ- process available, but also the variety of labels
ent interests, values, perceptions and policy these insights are presented under. Starting
preferences; from the first pages the reader encounters terms
(b) a time span of a decade or more; like ‘approach’ (p. xv), ‘frames of analysis’
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 14


(p. xvi), ‘frameworks and methods or Meanwhile, on the other hand, we see
approaches’ (p. xvii), ‘analytical frameworks’ theoretical work which suggests that the policy
(p. 32, heading), ‘theoretical frameworks’ process departs so far from rationality that the
(p. 32, text). Under the heading ‘Models, maps most appropriate model is a ‘garbage can’
and metaphors’ Parsons explicitly addresses (Cohen, March and Olsen, 1972) or that at
the question of how to label the different ways least parts of the policy process, such as agenda
of organizing one’s ideas in the study of the setting, are ‘dynamic, fluid and loosely jointed’
policy process. As an overarching and most (Kingdon, 1995: 230). Post-modernist theory
general term he chooses ‘frameworks’, ‘within goes even further, not merely to unmask the
which and through which we can think and rationalist discourse, but indeed to suggest the
explain’ (p. 57). Parsons distinguishes here impossibility of rational processes at all (Fox
between explanatory, ‘ideal-type’ and norma- and Miller, 1995; see also Fischer, 2003). The
tive frameworks (pp. 57–8). argument we will develop is obviously closely
We begin with theories, models, mental maps, meta-
linked with the former of the alternative per-
phors; and to think analytically about public policy spectives just mentioned. At the same time we
we have to be sensitive to the existence of ‘reality’ as a consider there can be a reconciliation between
construction within a multiplicity of frameworks. The these apparent extremes.
activity of theorizing about public policy is, therefore, Our point of departure is the view that in
like drawing a map… (p. 58).
Western countries in many fields public poli-
Given the varied landscape as pictured there is cies are working. Such policies are only given
more than reason to have adequate maps. For, public attention when there is a crisis, or even
since the founding fathers of the study of a disaster. For academics there has been a ten-
public administration, the discourse has devel- dency to take their leads from publicity about
oped in diverging ways that in many respects the negative aspects of government perfor-
can be described as a ‘dialogue of the deaf ’. On mance. Since Selznick published TVA and the
the one hand, the rational assumptions about Grass Roots in 1949, many social science stud-
public decision making, as once postulated by ies, certainly ones concentrating on implemen-
authors like Wilson (1887) and Simon (1945), tation, have found disappointing results of
seem to remain persistent. While implying public policies in practise. Often this dis-
‘stagist’ relationships between sets of activities appointment about what happened to good
within the realm of politics and administra- intentions has been expressed in more or less
tion, these assumptions above all refer to a straightforwardly gloomy titles. Setting the
clear division of labour. The way in which tone were New Towns in Town: Why a Federal
practising politicians and administrators in Program Failed (Derthick, 1972) and Imple-
many countries, for example, have adopted mentation: How Great Expectations Are Dashed
contracting out and similar meta-policies, in Oakland (…) (Pressman and Wildavsky,
aiming at the separation of ‘implementation’ 1973; see Hill and Hupe, 2003, for a further
from ‘policy’, resembles those assumptions. At discussion of the issues about ‘implementation
the same time, reporting on ‘policy fiascos’, failure’). In this context Linder and Peters speak
journalists tend to confront what is happening of the ‘horrors of war’ approach to the study of
at the implementation level of a policy public policy (1987: 460). This approach, with
straightforwardly with what has been agreed its emphasis on ‘failures’, has turned many of
upon in the legislative stage. Moreover, many these studies into what Rothstein calls ‘misery
academics are quite happy to do work which research’ (1998: 62–5).
would seem to share assumptions of this kind. Before we move to look at the stages model
We have in mind here the substantial evalua- more precisely, it is important to highlight a
tion literature, particularly the current concern problem about the study of the policy process,
to explore ways to make policy more evidence which has made the debate about that model
based. In the UK this trend is visible under the both complex and intense. As indicated, the
slogan ‘What works?’ (Davies et al., 2000). systematic study of the policy process has its
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 15


roots in the classical literature on public plants, aircraft carriers and submarines’ (2003:
administration. The founding elements in that 80). Quoting Perrow (1999) on ‘normal acci-
literature had two contentious characteristics. dents’ they say: ‘There will be failures and there
One was the notion that there could be a clear will be accidents, simple probability demon-
separation between politics and administra- strates that it is so … But every day we enjoy
tion, making the latter the careful formulation the modern miracle of high-reliability systems’
of specific activities based upon the policy (Ibid.: 81). We could similarly refer to social
goals set by democratically accountable politi- policy systems. All over the world complex sys-
cians (Wilson, 1887). The other was that the tems have been designed to deliver social ben-
policy process could, or should, be a rational efits, in accordance with explicit rules, which
one, in the sense in which that term was used function uncontroversially much of the time.
by Simon (1945). This would mean that a care- Debate focuses upon either their occasional
ful and thorough examination of the relation- failures or upon those (often marginal) parts
ship between ends and means is carried out by of the systems where rules are hard to operate.
those required to translate policies into action. If we try to take a balanced view between the
Both of these notions, of course, came under positive and the negative image of policy
attack. It was argued that Wilson’s ideal of the processes we need a way of theorising about
separation of administration from politics was both its more systematic aspects and about the
unattainable, that administration is inevitably phenomena that may undermine that. In our
a political process (Waldo, 1946) and that the stance we join the many students of the policy
translation of policies into action is a very process who want to try to sort out approaches
much more haphazard process than Simon to specifying the issues at stake and the scope
suggested (Lindblom, 1959). for their systematic study. One key element in
debates about how to do this has involved
sharp differences of view about the extent to
Systematic study requires which it is possible to separate the policy
a general map process into ‘stages’. The so called ‘stages
model’ has been offered as a map, around
We put aside here the arguments, particularly which a consensus about theory building may
used on the ‘rationalist side’, that their models be developed. A question then is whether this
represent ways policy process should occur. map building can be a shared process, as is
Rather our stance is based upon a view that implicit in the positivist model of social
both sides in the argument tend to distort real- science, or whether we have a series of compet-
ity. The case against the rational actor can be ing maps (or at worst each our own map)
easily made: indeed much of the work of its which cannot be reconciled with each other.
opponents deploys good evidence of haphaz- Our objective here is first to explore the
ard and irrational decision making processes nature, functions and limitations of what has
and of the manipulative deployment of dis- been called the stages model, and then to see
courses. The case that needs to be made more whether an alternative general framework for
clearly, however, is on the other side, that there the analysis of policy processes may be built
is a tendency for contemporary analyses of upon it. The central questions are:
policy processes to focus upon mistakes and
disasters and to disregard the large number of • What functions does the so called ‘stages
examples of stable, successful policy processes, model’ fulfil in the study of the policy
involving deliberate design work worthy of process?
Simon’s model. Frederickson and Smith speak • Which limitations to its use can be
of ‘the high-reliability systems; the best exam- identified?
ples include commercial air travel, the provi- • And how may these limitations be over-
sion of electricity, gas, and cable television come in an alternative framework for
services; and the operation of nuclear power analysis?
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 16


Hence, in the following discussion we first look • intelligence

at various versions of the stages model and at • promotion
criticisms of it (second section). Next, an • prescription
assessment is given: Is it a ‘model’ or some- • invocation
thing else? – while also the methodological • application
pitfalls arising out of an unreflecting use get • termination
attention (third section). From there we pro- • appraisal
ceed to set out an alternative analytical frame-
work built upon notions of alternative aspects In doing this Lasswell argues that he is framing
of governance, aiming to offer suggestions for a ‘conceptual map (that) must provide a guide
dealing with the identified limitations of the to obtaining a generalised image of the major
model (fourth section). We then go on to phases of any collective act’ (Lasswell, 1971:
explore the analytical gains that may be made 28). Lasswell was by no means the only scholar
by approaching policy processes in terms of to see the policy process as involving stages;
our framework (fifth section). The chapter indeed Wilson’s politics/administration dicho-
ends with some conclusions (sixth section). tomy can be seen as containing the seeds of
such an approach. In 1945 Herbert Simon
(1945) had formulated the logic of the way
people reach decisions in terms of three suc-
2. THE STAGES MODEL OF THE cessive stages: intelligence, design and choice.
POLICY PROCESS Since Lasswell’s specification of seven stages
there have been many variants, altering the
Variants in the literature number of stages or specifying them a little dif-
ferently (Mack, 1971; Rose 1973; Brewer, 1974;
The study of the policy process owes a lot to Jenkins, 1978; Hogwood and Gunn, 1984).
Harold Lasswell, the American political scien- Probably the most complex elaboration of the
tist who was active during the second half of stages heuristic is provided by Dror (1989:
the twentieth century. Parsons sees Lasswell 163–4). He distinguishes three major stages: the
as the originator of what the latter himself called metapolicy-making stage, the policy-making
‘the policy orientation’ (1951a). Overviewing stage, and the post-policy making stage. Within
Lasswell’s work (1951b, 1968, 1970, 1971) each of those Dror specifies four to seven sub-
Parsons gives the following characterisation of stages; all adding up to a total number of 18
that ‘policy orientation’: ‘multi-method, multi- successive (sub-) stages. In numerous policy
disciplinary, problem-focused, concerned to textbooks published since the 1970s, the stages
map the contextuality of the policy process, heuristic has been used to structure accounts of
policy options and policy outcomes; and the policy process (see, for example, Jones, 1970;
whose goal is to integrate knowledge into an Anderson, 1975; May and Wildavsky (eds.),
overarching discipline to analyse public 1978; Hoogerwerf (ed.), 1978; Kuypers, 1980;
choices and decision making and thereby con- Brewer and DeLeon, 1983; Hogwood and Gunn,
tribute to the democratization of society’ 1984; Palumbo, 1988; Van de Graaf and Hoppe,
(Parsons, 1995: xvi). Lasswell (1956) was one 1989; Howlett and Ramesh, 2003).
of the first to approach the overall process of Even amongst contemporary writers, eager
the making of public policy explicitly in terms to stress the complexity of the policy process,
of ‘phases’ or ‘stages’. He uses that term to refer we still find that the idea of stages is influential.
to a set of separate and successive steps, Thus Sabatier, in his edited book on the policy
thought of as in principle taken in a chrono- process, provides the following definition:
logical order, from initiative via formulation
and decision to evaluation and termination. The process of public policy making includes the manner
in which problems get conceptualized and brought to gov-
More specifically, Lasswell (1956) distinguishes ernment for solution; governmental institutions formulate
between what he calls the seven ‘stages’ of ‘the alternatives and select policy solutions; and those solu-
decision process’: tions get implemented, evaluated, and revised (1999: 3).
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 17


Criticisms what should it do? One of its proponents of the

first hour who has stayed an advocate, is Peter
Despite the perfectly ‘stagist’ definition just DeLeon. He outlines a few characteristics of
quoted, Sabatier argues: the heuristic, and the functions it fulfils (1999:
The conclusion seems inescapable: The stages heuristic 20–3). First, stages ‘offer a way to think about
has outlived its usefulness and needs to be replaced with public policy in concept and (…) in operation’
better theoretical frameworks (1999: 7). (Ibid.: 21). Hence, by giving attention to the
Nakamura (1987) criticizes the stages model of characteristics of related sets of activities the
the policy process, portraying it as unrealistic, stages notion not only gave rise to textbook
calling it the ‘textbook approach’. Lindblom knowledge of the policy process, but also
(1968) was one of the first to emphasise the directed an entire generation of theoretical-
sometimes blurred borders between the stages; empirical research. DeLeon points out a
later he and Woodhouse stress that ‘policy sequence of ‘policy classics’ that focused upon
making is (…) a complexly inter-active process crucial stages in the policy process rather than
without beginning or end’ (1993: 11). Jenkins- on specific issue areas. He argues that these
Smith and Sabatier make the following points works brought a new richness to the policy
in a more detailed critique of what they call the sciences, because they emphasized the com-
stages ‘heuristic’: plexities of policy processes and researched
these in greater depth than political scientists
(a) That it is not a causal model. and economists, using more rigorous models,
(b) That it ‘does not provide a clear basis for had done before. Even John (1998: 36), a critic
empirical hypothesis testing’. of what has been called the stages model, con-
(c) That it is descriptively inaccurate. cedes that ‘Researchers can apply it because it
(d) That is has a ‘built-in legalistic, top-down imposes some order on the research process’.
focus’. DeLeon goes on to argue that analysing the
(e) That it gives an inappropriate emphasis to policy process, while doing such in terms of
‘the policy cycle as the temporal unit of stages, has enabled a move away from the study
analysis’. of legal-juridical institutions as practised in
(f) That it ‘fails to provide a good vehicle for public administration and from the study of
integrating the roles of policy analysis and quasi-markets exercised by economists. This
policy-oriented learning throughout the innovative emphasis ‘helped to rationalize a
public policy process’ (1993: 3–4, italics new problem oriented perspective’ (Ibid.: 22).
from the original). In offering such a defence of the stages heuris-
Similar arguments had been set out in Sabatier tic DeLeon considers it important to make
(1991) and in Stone (1989). clear that he does not regard it as a ‘theory’, in
the sense implied by the first three points
against it mentioned by Jenkins-Smith and
Sabatier. DeLeon argues:
THE STAGES MODEL Brewer and DeLeon (and, by implication, Lasswell) (…)
realized that it was not suitable to formal hypothesis
testing or prediction with much precision (…). Rather,
An assessment of the they viewed the policy process as a device (a heuristic, as
stages heuristic it were) to help disaggregate an otherwise seamless web
of public policy transactions. They proposed that each
In order to find out if these seemingly defini- segment and transition were distinguished by differen-
tiated actions and purposes (DeLeon, 1999: 24).
tive judgements can be justified it is worth-
while to look more precisely at the nature of Accordingly, DeLeon argues that frameworks
the stages heuristic or stages notion – neutral like the ‘advocacy coalition framework’ of
terms that, indeed, beforehand seem more Sabatier and Jenkins-Smith and the ‘punctu-
appropriate. What is the ‘stages model’ and ated equilibrium framework’ of Baumgartner
what not? What can it do and what not? And and Jones (1993) can be positioned as referring
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 18


to policy initiation, an early part in the stages decisions and actions that occur around a
heuristic (though that hardly seems fair to the policy’ (Schlager, 1999: 239). But if it is a typol-
former). DeLeon gives more examples of per- ogy, what then are the parameters? In our view
ceived alternative approaches to the stages the key to understanding the nature of the
heuristic (for instance Fischer, 1995, on evalu- stages heuristic must be sought in its origins:
ation), which, in fact, relate well to it instead of Lasswell’s ‘policy orientation’ (1951a). Focusing
replacing it. If this is so, may it be then perhaps on what Dewey (1927) once described as ‘the
that the stages heuristic is something of a dif- public and its problems’, public policy is an
ferent order? object of scholarly inquiry with a scope going
It seems useful to note here how carelessly beyond scientific theory. This implies that the
notions of theory are used in policy studies. stages heuristic can be seen as part of a wide-
Elinor Ostrom (1999) argues that it is impor- spread, in fact pre-scientific, general conception
tant to distinguish within theoretical work of public policy. The term ‘public policy’ seems
three levels of specificity, which are often con- to imply a set of underlying assumptions of
fused: frameworks, theories and models. which, in an ideal-typical form, the following
Hence she identifies these as follows: construction can be made.
Within the trias politica the executive power
• A framework helps to identify the elements
executes what the legislative power has formu-
needed for more systematic analysis, provid-
lated and decided upon. Actors other than the
ing a list of variables and ‘metaphorical lan-
designer/decision maker execute public poli-
guage that can be used to compare theories’
cies according to the intentions legitimately
(Ibid.: 40).
laid down in laws or other official documents.
• Theories ‘enable the analyst to specify
Preconditions for effective policy are adequate
which elements of the framework are
knowledge and information, a solid basis of
particularly relevant to certain kinds of
power and cooperation, and an obedient
questions and to make general working
field of application (Van Gunsteren, 1980).
assumptions about these elements. Thus,
Through the implementation process the
theories focus on a framework and make
intentions of a public policy are, literally,
specific assumptions that are necessary for
realised. Overall, the realisation of these inten-
an analyst to diagnose a phenomenon,
tions takes the form of what is called a policy
explain its processes, and predict outcomes.
process: a series of orderly staged successive
Several theories are usually compatible
sequences, going from articulation via forma-
with any framework’ (Ibid.).
tion and implementation to evaluation (and
• Models ‘make precise assumptions about a
back). A policy is seen as ‘a hypothesis contain-
limited set of parameters and variables’
ing initial conditions and predicted conse-
quences. If X is done at time t1, then Y will
Therefore, perhaps we may argue that the result at time t2’ (Pressman and Wildavsky,
stages heuristic is a framework rather than a 1973: xiii). Despite the use of the term ‘policy
theory. This seems the obvious position to take cycle’, actually the more appropriate metaphor
about the value of the heuristic as a device to is that of a chain: both in time and often in
facilitate research, and, as John (1998) adds, space, vertically linking various ‘clearances’.
teaching. Public servants, especially those with an imple-
Schlager, however, offers a slightly different mentation task, fulfil their jobs within a hier-
approach to this issue. She sees the stages notion archical setting, with fixed competences, led by
as a ‘useful categorization of behaviour and documents and guided by rules. If what is
action within entire policy processes’ but not a achieved is not what was expected, shortcom-
framework because ‘general classes of variables, ings in implementation are to blame, particu-
or “universal elements”, and general relationships larly insufficient rule compliance.
among them’ are lacking. She thus describes it as For policy researchers a few consequences
‘a typology that completely describes policy follow (for an elaboration see Hill and Hupe,
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 19


2003): Policy determines action, while policy model which denies that policy making goes
intentions predict policy performance in a 1:1 on at later stages, nevertheless this is the impli-
relationship. cation of its casual or uncritical use. It may,
With the general conception of public policy, indeed, be ‘implementation’; but it could also
as constructed here, we obviously are back to the be ongoing ‘policy formation’. While it may
idea of the administrative process described in have been originally inherent to the stages
our introduction as the traditional model linked heuristic that there was a fixed, a priori locali-
with the work of Wilson, Simon and many sation of specific stages to certain administra-
others. We accept then the way in which tive layers, this empirically needs not be the
researchers have in various respects argued that case. Second, related to this problem is what
this ideal-typical ‘image’ does not represent the has been designated as ‘the fallacy of the wrong
real world of public policy. Yet people continue layer’ (Hill and Hupe, 2003). If at a ‘lower’ layer
to hold this conception of public policy, founded the stage of ‘policy formation’ de facto can be
as it is in institutionalised normative views observed, this must not be seen as implying a
on democracy and the rule of law, as highly judgement about the legitimacy of that partic-
attractive. The conception seems so widespread, ipation in the policy process. The latter is a
not only because it is rooted in the political- normative question that should be distin-
administrative culture and institutions of the guished from the empirical observation. The
Western Rechtsstaat, but also because it appeals to stages heuristic seems to imply for each stage a
both a general quest for control (Van Gunsteren, specific, presupposed set of actors, a third
1976) and a psychological desire for a rational methodological limitation. Where empirical
order of things2. In this context there is no need and normative matters thus are confused, one
to be surprised about the persistency of the stages may fall into a pitfall which could be called the
heuristic. If that is the perspective, is the heuristic ‘control trap’ (Hill and Hupe, 2003).
then suited for the study of the policy process as Our position is that the stages heuristic is
it is, after all; or should it be adapted? rooted in the logically defensible assumption
DeLeon, though an advocate, acknowledges that decisions are followed by actions, which
that the negative aspect of the invitation from require a cumulative process if anything is to
the stages heuristic to researchers to look at ‘just occur at all. But such a view does not preclude
one stage at a time’ is the neglect of the policy the idea that there will be subsequent decisions
process as a whole (1999: 23). Furthermore, the that may undermine the original one. Then
heuristic may lead to a view of the policy does this imply a need to abandon the stages
process as ‘disjointed’ and ‘episodic’, taking place notion or to use it with care?
in the relatively short term (a key aspect of Despite his criticism of the stages heuristic
Sabatier’s critique from the point of view of his Sabatier observes ‘Given the staggering com-
‘advocacy coalition framework’). Besides, the plexity of the policy process, the analyst must
picture of stages inappropriately implies a cer- find some way of simplifying the situation in
tain linearity to many, instead of ‘feedback order to have any chance of understanding it’
actions and recursive loops’ (Ibid.). Also, when (1999: 4). Parsons sees the value of the stages
one takes the stages heuristic as a general map heuristic to cope with ‘multi-framed activity’
for the analysis of policy processes, as indicated (1995: 80), but at the same time underlines the
above, it potentially gives rise to some misun- necessity to look beyond it towards
derstandings that may have consequences for
the mapping of the wider contexts of problems, social
research findings. process, values and institutions within which policy-
First, the picture of stages seems to reify the making and policy analysis (take) place (p. 81).
scope of each separate stage. This conflicts with
a view taken for a long while, and particularly Hence he argues
since Lipsky’s seminal work (1980), that policy Given the existence of a complex policy reality framed
making goes on even at the street-level. While, by a range of theories, models, explanations, values and
in fact, there is nothing implicit to the stages ideologies, the problem is not with the policy cycle
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 20


[stages may be substituted in this context – the authors] doing so, it is important to recognise that
per se, as with the need to incorporate or include in some respects these still involve a ‘stages’
models and approaches which are or may be deployed in
policy analysis (Parsons, 1995: 81).
perspective, albeit a modified one.
Lynn (1981: 146–9; 1987) uses the concept
In other words, given the need for a relatively of ‘nested games’ to assign the various parts of
‘empty’ framework general enough to com- the policy process. There is the ‘high game’, in
prise multiple theoretical approaches, Parsons which it is decided whether or not a policy will
sees a remaining function for the, perhaps be made. Then, in the ‘middle game’, the direc-
amended, stages heuristic. tion of the policy is determined. The ‘low
On the basis of this overview we suggest the game’ is about the practical side of the policy
following criteria for a ‘framework’ as a ‘general making; implementation is central here.
map’. First, the latter must have the capacity to Parsons (1995: xvii) distinguishes ‘three
encompass conceptually the ‘multiple multiplic- broad and overlapping levels or dimensions of
ity’ character of public policy processes. This analysis’. Each of those ‘may be seen through a
involves both a variety of actors and very often a variety of different frameworks and approaches’
variety of administrative layers, and horizontally (p. 82). The way in which issues and problems
linked organisations. The scholarly analysis of are defined and policy agendas are set Parsons
these uses a variety of disciplinary perspectives, calls ‘meso-analysis’. As a level of analysis cut-
‘lenses’, and focuses specifically on various parts ting through the various phases of the policy
of the policy process. Second, a ‘general map’ or process, it is ‘meso’ because it:
framework must enable specified (‘localised’)
explores approaches which link the input side of the
theory formation around selected sets of vari- policy-making process with the policy/decision-making
ables, rather than compel the building of one and output process focusing on the relationship
grand theory. Third, while the first criterion between the ‘pre-decisional’ dimensions of policy-
requires the framework to be as comprehensive making and its decisional and post-decisional contexts
(Parsons, 1995: 82).
as possible, it at the same time needs to be open
in the empirical sense. Rather than implying a The analysis of ‘how decisions are taken and
‘managerialist’ or ‘top-down’ view of the policy policies are made and how analysis is used
process (Parsons, 1995: 81), a general framework within the decision-making process’ Parsons
for analysis must facilitate systematic and nor- calls ‘decision analysis’ (p. 82). Then ‘delivery
matively open empirical research. analysis’ refers to ‘how policies are adminis-
Assessing the stages heuristic according to tered, managed, implemented, evaluated and
these criteria we may conclude that it has diffi- terminated’ (p. 82).
culty in meeting the first and third criteria. Finally, there is the ‘institutional analysis and
DeLeon’s reference to the way in which it has development’ (IAD) framework developed by
facilitated some classic pieces of research indi- Elinor Ostrom and her colleagues, presented
cates that it meets the second one. Overall, we as such for the first time in 1982 by her and
may characterise the stages heuristic as a multi- Larry Kiser. Stemming from micro-institutional
dimensional framework for the analysis of policy analysis Kiser and Ostrom (1982: 184) specify
processes, of which not all relevant dimensions three related but distinct levels of analysis.
have been articulated. Can we build on it to pro-
The operational level, which explains the world of
duce a more helpful overarching framework? action. The collective choice level, which explains the
world of ‘authoritative decision-making’. The third is
the constitutional level, explaining ‘the design of collec-
Alternative analytical frameworks tive choice mechanisms’ (p. 184). It may be noted that
as ‘general maps’ these are listed in the original formulation in an order
that reverses the conventional order of the stages model.
Individuals at the operational level ‘either take direct
Though the stages heuristic seems a widely
action or adopt a strategy for future actions, depending
used general framework for the study of the on expected contingencies.’ They are often ‘authorized
policy process, some alternatives have been to take a wide variety of actions at this level without
developed. We will look at these here. But in prior agreement with other individuals’ (pp. 207–8).
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 21


Collective decisions are made ‘by officials (including Governance ‘is a descriptive label that is used to high-
citizens acting as officials) to determine, enforce, continue, light the changing nature of the policy process in recent
or alter actions authorized within institutional arrange- decades. In particular, it sensitizes us to the ever-increasing
ments’. These collective decisions are plans for future variety of terrains and actors involved in the making of
action. ‘Unlike individual strategies, collective decisions public policy. Thus, it demands that we consider all the
are enforceable against nonconforming individuals … actors and locations beyond the “core executive” involved
The authority to impose sanctions is a key attribute of the in the policy making process’ (2002: 3).
collective choice level of decision-making’ (p. 208).
On the third level ‘Constitutional decisions are collec- While for Milward and Provan
tive choices about rules governing future collective deci- ‘Governance (…) is concerned with creating the condi-
sions to authorize actions. Constitutional choices, in tions for ordered rules and collective action, often
other words, are decisions about decision rules’ (p. 208). including agents in the private and nonprofits sectors, as
In the framework institutional arrangements are ‘link- well as within the public sector. The essence of gover-
ing each level of decision making to the next level. nance is its focus on governing mechanisms – grants,
Constitutional decisions establish institutional arrange- contracts, agreements – that do not rest solely on the
ments and their enforcement for collective choice. authority and sanctions of government’ (1999: 3; for
Collective decisions, in turn, establish institutional arrange- other definitions see also Kooiman, ed., 1993; 1999;
ments and their enforcement for individual action’ (p. 209). 2003; Pierre and Peters, 2000; Heinrich and Lynn, eds,
2000; Lynn et al., 2001).
In this framework it is only at the operational
level where an action in the physical world A contribution towards developing a gover-
flows directly from a decision. nance perspective to the study of policy
In an updated discussion of the framework, processes, founding and grounding their con-
Ostrom (1999: 36–9) elaborates some ‘key dif- ceptualisation in relation to the state of the art
ficulties’ in the study of institutions. They of implementation theory and research, was
involve the multiple meanings of the term made by Hill and Hupe (2002). What follows in
institution, the invisibility of institutions, the this section is partly based on that but offers a
multiplicity of inputs coming from various further elaboration of the insights presented
disciplines and the corresponding need to there. The concept of governance is designed ‘to
develop a specific ‘language’, and the configu- incorporate a more complete understanding of
ration character of relationships. An additional the multiple levels of action and kinds of vari-
matter Ostrom mentions is the multiplicity of ables that can be expected to influence perfor-
levels of analysis. Particularly important here is mance’ (O’Toole, 2000: 276). That makes the
the analytical treatment of the nested structure concept well suited to incorporation into a gen-
of the framework (1999: 38–9). eral framework for the multi-dimensional
Further to these ways of specifying a frame- analysis of policy processes. Thus, such analysis
work for policy analysis we think there is room can be seen as ‘governance research’ (see
for an additional framework as another alterna- Heinrich and Lynn, eds, 2000; Lynn et al., 2001).
tive for the stages heuristic, driven by the aim of Hill and Hupe (2002: 15) sketch the conse-
all three but particularly inspired conceptually quences of that. First, a clear distinction is made
by Ostrom’s framework. In our view for that between the how and the what of scholarly
purpose some amending of that framework is attention. Focusing upon governing as action,
necessary. Specifically, there is a need to make an rather than as government as institution, leaves
explicit link with the concept of ‘governance’. empirically open who is the acting actor (may
be a public, may be a private one). Second, dif-
ferentiating between administrative layers is
4. THE POLICY PROCESS AS important. Third, the act of management is
MULTIPLE GOVERNANCE taken seriously; in principle it can be observed
in all the loci of political-societal relations.
Governance Therefore levels of analysis have to be specified.
Summarising, the stucture of a policy process
Various authors have given definitions of the can be seen as consisting of various elements:
concept of ‘governance’. Richards and Smith, actors, sets of activities, action situations, and
for instance, say: layers. Each concept will be elaborated here.
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 22


Actors decision making about collectively desired

outcomes. Facilitating the conditions for the
Ostrom’s definitions of an actor (‘a single indi- realisation of these situations belongs to this
vidual or a group functioning as a corporate part of governance. Operational governance con-
actor’) and of action (‘those human behaviors cerns the actual managing of that realisation
to which the acting individual attaches a sub- process.
jective and instrumental meaning’) cannot be Respectively these three action levels refer
improved (1999: 43). Taking the concept of to structure-oriented, content-oriented, and
governance seriously means that the answer process-oriented sets of activities. Ostrom’s
to the question who is the ‘governing actor’ is illustration of a shift of action levels is worth-
empirically open. This implies that the number while quoting here.
and legitimacy of actors actually involved in a (W)hen a ‘boss’ says to an ‘employee’, ‘How about
particular policy process may differ from what changing the way we do X?’ and the two discuss options
on normative grounds may be expected. and jointly agree upon a better way, they have shifted
from taking actions within previously established rules
to making decisions about the rules structuring future
actions (Ostrom, 1999: 47).
Action levels
In the terminology of the Multiple Governance
Taking our lead from Kiser and Ostrom’s Framework presented in this chapter, what
(1982) ‘three worlds of action’, outlined above, is at stake here is a shift from the level of
we see the policy process as governance as con- operational governance to the level of directive
sisting of three broad sets of activities that we governance. Speaking of ‘a policy (or collective–
call constitutive, directive and operational gov- choice) tier’ (Ibid.: 41) and even of ‘policy-
ernance. The notion of constitutive governance making (or governance)’ (Ibid.: 59), it looks as
derives from Kiser and Ostrom’s notion of if Ostrom, though only casually, unintendedly
constitutional choice, which they define as the anticipates the kind of explicit policy process-
framing of rules that ‘affect operational activi- as-governance conception proposed in this
ties and their effects in determining who is eli- chapter. At one place she calls policy subsys-
gible’ together with rules ‘to be used in crafting tems ‘multiple linked action arenas at all three
the set of collective-choice rules that in turn levels of analysis’ (Ibid.: 58). What Ostrom
affect the set of operational rules’ (Ostrom, then states about the nested character of
1999: 59). This somewhat ambiguous formula- her framework also goes – of course with an
tion, defined elsewhere, as noted above, as amended terminology – for ours. Ostrom thus
‘decisions about decision rules’ seems to specifies the way institutional rules (a central
embrace both fundamental decisions about concept in the substance of her framework)
the content of policy and about the organiza- cumulatively affect the actions taken and out-
tional arrangements for its delivery. This dis- comes obtained in any setting.
tinction is important; for example, a major Though the centrality of the concept of
policy innovation in the field of health care ‘rules’ is specific, the working of the nesting
delivery will contain both rules on who is to be mechanism could not be better pictured,
entitled to new health benefits together with and is very similar to the consequences of the
rules about how those benefits are to be deliv- nested character of the Multiple Governance
ered. In that respect it may be a bit misleading Framework. Schlager (1999: 238) interprets
to speak of ‘constitutional choice’, directing the this as follows: ‘(T)he rules-in-use that struc-
reader’s attention to the latter when that has ture the operational level originate from the
little significance without the former. Hence other two levels’. In the same paragraph she
our preference for the term ‘constitutive’3. concludes: ‘Although the analyst can choose
The direction in directive governance, our to keep the analysis focused on a single level,
alternative to Kiser and Ostrom’s ‘collective the other two levels are always implicitly
choice’, stands for the formulation of and included’.
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 23


Table 1.1 The Multiple Governance Framework

Action levels
Scale of action situations Constitutive Governance Directive Governance Operational Governance
System Institutional design General rule setting Managing trajectories
Organisation Designing contextual relations Context maintenance Managing relations
Individual Developing professional norms Situation bound rule application Managing contacts
Adaptation of: Hill and Hupe, 2002, p. 183.

Action situations normatively cut out are one or more layers.

The term ‘layers’ refers to a specific kind of
The three sets of activities generally distin- action locations: the formal, legitimate political-
guished above as action levels get a specific form administrative institutions, including repre-
dependent on the locus observed. When we sentative organs, with certain territorial
speak of ‘locus’ our definition is comparable to competences. While the term ‘locus’ concerns
(though perhaps less refined than) the one action situations in political–societal relations
Ostrom gives of her concept ‘action situation’: designated in the threefold aggregated distinc-
An analytic concept that enables an analyst to isolate the tion mentioned above, it refers to a series of
immediate structure affecting a process of interest to the spots on a line of vertical public administra-
analyst for the purpose of explaining regularities in tion from which ‘real’ actors participate in a
human actions and results, and potentially to reform particular policy process. Related to each other
them (Ostrom, 1999: 43).
in a nested configuration, like Russian dolls,
Empirically, both the number of acting actors both the number of real persons actually
and that of the potential action situations in involved in a given policy process and the vari-
which they act – as Elinor Ostrom has pointed ety of action situations in which they act, usu-
out – can be thought of as infinite. Since the ally is larger than implied by the specific
latter category does not coincide with the for- formal administrative layer. When we examine
mer, can a usable taxonomy be devised? Here, the latter we are (only) looking at legitimate
too, a threefold distinction can be made: irre- constitutional settings within a specific political-
spective of the kind of formal administrative administrative system, while the study of the
layer looked at, an actor there can be seen policy process needs a broader perspective on
performing specific activities in action situa- relevant variables.
tions on a scale that can vary from action of We call this ‘the Multiple Governance
and between individuals (in practice called ‘the Framework’. Its nested character implies that,
street-level’), via action of and between organ- conceptually, one action level is not necessarily
isations, to action on the system-scale. This confined to one administrative layer. Whether,
varying degree of aggregation can be labelled for instance, in a given policy process at the
in a summarising way as, respectively, the locus layer of local government just ‘implementa-
of the Individual, of the Organisation, and of tion’ or rather ‘policy co-formation’ is prac-
the System. tised, is, to begin with, an empirical question,
resting upon an interpretation of the extent of
change. Drawing the line between goal setting
Administrative layers and goal realisation seems at stake here. Any
judgement about whether the specific empiri-
Going ‘downwards’ in a system of vertical cally observed action is desirable is a norma-
public administration, a policy process tive matter. Similarly, there are various acts of
encounters a range of actors and loci as action ‘operational governance’ – consisting of manag-
situations. Within that empirical range as a spe- ing trajectories, managing inter-organisational
cific assembly of such actors and actor situations relations, and managing external and internal
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 24


Table 1.2 Alternative general analytical frameworks for the study of the policy process
Institutional rational choice
Nested games Kiser and Ostrom Multiple stages Multiple governance
Lynn (1981) (1982; Ostrom, 1999) (Parsons, 1995) (Hill and Hupe, 2002)
High game Constitutional level Meso analysis Constitutive governance

Middle game Collective choice level Decision analysis Directive governance

Low game Operational level Delivery analysis Operational governance

contacts – which can be regarded as sublevels themes of public administration. Thus, a

of action and are not confined to one spe- framework has been devised which is particu-
cific layer of government. Thus, the connec- larly suitable for the study of the policy
tions between actors, acts and action spots are process. Fourth, Ostrom’s formulation has a
of an empirical instead of an a priori nature. strong institutional emphasis whilst, as we
Rather than supposing them, they are to be pointed out above to explain our adaptation of
investigated4. the concept of ‘constitutional choice’, we see
‘content’ issues in respect of policy innovations
as of great importance in structuring subse-
Positioning the framework
quent decisions.
We set out in Table 1.2 our categories to
We acknowledge that the Multiple Governance
describe levels of analysis, comparing them
Framework as presented here is partly intellec-
with some of the alternative approaches.
tually derivative from the IAD-framework. At
the same time, however, there are some major
differences that amount to more than the
terminological amendments and justify the
development of a separate framework. First,
the framework links the study of the policy
process explicitly with the concept of gover- When we leave the concept of actors aside the
nance. Essential to this is the shared focus on singularity of the notion of stages has been dif-
action rather than (only) on institutions in the ferentiated into three concepts: action levels,
traditional-legalistic sense of the word; the action situations and layers in a politico-
combination of a ‘vertical’ and ‘horizontal’ administrative system. If a stages perspective
orientation; the stress on specifying levels of were adopted each of them might be seen to
analysis and on distinguishing empirical from designate alternative approaches to exploring
normative matters. Second, a specific charac- the development of policy. Indeed, the more
teristic of the framework outlined here is the simplistic approaches to the stages perspective
localisation of the various governance activi- go further to equate and confuse these three
ties, in the variety of action situations within a concepts. Our approach instead is to suggest
range of political-societal relations. The gen- that, while of course what we have called nest-
eral ‘employee’ of Elinor Ostrom becomes, for ing occurs, one cannot presume either any
instance, a street-level bureaucrat, may be a fire explicit uni-directional progress through the
officer, while his or her ‘boss’ may be the head designated categories or a taken for granted
of the local Fire Brigade. Third, the Multiple equation between successive levels, loci and
Governance Framework draws the micro- layers. In a given situation one may hypothe-
economically rooted assumptions of the IAD- size that either or both of these relationships
framework into mainstream social science, do occur, and accordingly test whether this is
making links with the classical scholarly true. Furthermore, and this is a particular
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 25


source of confusion in many discussions of management arrangements and rules affecting

public administration, one may encounter discretion at the street-level. But not only is
views, often strongly held, that the supposed there likely to be great variation around these
relationships should occur. themes, there will also be controversy about
By using a matrix form any reference of ver- the appropriateness or inappropriateness of the
ticalisation seems to have been avoided. And structuring involved. The latter will take the
yet, despite the grid character, even this frame- form of arguments about professional preroga-
work could be read in a ‘stagist’ way. Such read- tives and about the rights of patients, pupils and
ing would imply that constitutive governance parents to influence the system. Such arguments
establishes the structural dimensions, that will not be merely about discretion at the street-
directive governance determines the detailed level but also about the adequateness of ‘prior’
contents and that operational governance is structuring decisions.
concerned with the process side of public poli- Hence, we suggest, our alternative analytical
cies. Readers may react to this, saying that this framework offers two contributions to the study
is indeed what they would expect to find. But if of the policy process while avoiding the metho-
they do so, are they providing a hypothesis, or dological limitations of the stages heuristic. We
a statement about what they should find? set these two out below: enhancing ‘localised’
This question highlights the relationship theory formation and helping the identifica-
between the various activities in a policy process tion of action possibilities.
taking place at different moments, at different
spots, by different actors. Leaving aside the
Enabling contextual theory formation
empirical question about who are the acting
actors on one hand, and the similar question
The primary function of the Multiple
which administrative layer is looked at, on the
Governance Framework is to provide a con-
other, the framework as pictured in Table 1.1 in
ceptual (meta-) basis for contextual theory-
the boxes, shows the kind of activities that may
building in the study of the policy process. At a
occur as a result of specified action level/action
meta-level it designates organising concepts
situation combinations. Could it, indeed, be,
that enable the formation of specific low- or
that an activity cluster (‘stage’), identified here
middle range theories of a more ‘localised’, in
as one specific level of action and usually asso-
the sense of locus-related, character. Like
ciated with one particular layer, along the line of
DeLeon (1999: 29), Hill and Hupe (2003)
vertical administration, is going on not only
advise against reaching for a ‘grand unifying
(legitimately or not) practised by actors at other
theory’. Rather, we suggest that for mainstream
layers but in a variety of action situations as
comparative-empirical research projects it is
well? It is then crucial, repeating our key obser-
desirable to look at a subject at one adminis-
vations about hypotheses and ‘ought’ state-
trative layer and no more than two action
ments, to recognise both that there is great
levels at a time. After all, in the study of the
variation in the extent to which these activities
policy process, many more hypotheses have
as a result of certain action level/action situation
been formulated than tested.
combinations do occur and that there can be
In the specification of a research question
debate about the extent to which policy outputs
the framework guides the selection of vari-
should be affected by these activities. Clear
ables. Looking at the dependent variables used
examples of this can be found in controversies
in policy research so far it is possible to distin-
about the determination of health or education
guish the aim of explaining:
policy. Reality in these systems will, we hypoth-
esize (following our own recommendation for • policy change within a given political
caution in this respect), be a combination of system;
an explicitly and not easily amended system, • adoption of a policy or set of policies;
conveying certain expectations of the system • variation in policy outputs or outcomes
as a whole, together with locally determined (Sabatier, ed., 1999).
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 26


The framework provides flexibility to choose argue that public policy processes need to be
the appropriate unit of analysis. On the side able to hold in tension multiple ‘accountabi-
of the independent variables, too, the Multiple lities’. In this respect, as in the cases of public
Governance Framework offers a way of struc- policy for health and education mentioned
turing these rather than prescribing what they above, it may be important to try to develop
should be. O’Toole’s (1986) scan of the imple- policies so that they follow complex pathways
mentation literature suggested a massive list of across the items in Table 1.1. The issues about
variables, attracting the criticism that here is a how, and by whom, the overall structure and
subject in need of parsimony (see Matland, funding should be determined may differ from
1995; also Meier, 1999). The identification of those about local arrangements for policy
the framework within which action may occur delivery, whilst each again may differ from the
brings this list down to a limited range; Hill concerns about the discretionary behaviour of
and Hupe (2002: 123) have suggested about practitioners. All are connected, but there are
seven categories of independent variables5. many options about how those connections
This narrower range makes it possible to select may be made: this goes for policy research as
a small number of variables thought relevant well as in policy practise.
to the research question at hand.

Aiding the identification of
action choices
Hardly any other insight from public adminis-
A great deal of the implementation literature tration or political science has been so gener-
has been preoccupied with a normative argu- ally adopted by practitioners as the so-called
ment between ‘top-down’ and ‘bottom-up’ stages model of the policy process. This seems
perspectives. Both stem from deeply held views to exemplify Lasswell’s stress on the extended
about democratic accountability. Faced with scope of what he called the ‘policy orientation’,
what they regard as defects in the implementa- going beyond the borders of science. The idea
tion process both perspectives concern them- of policy making as proceeding in successive
selves with efforts to increase the capacity to stages influenced Pressman and Wildavsky in
steer the policy processes, either from the formulating their ‘chain’ hypothesis, implying
top or the bottom. By contrast, an analytical the likelihood of ‘implementation gaps’
perspective that recognises that influencing (Bowen, 1982). By consequence it invited the
the policy process can involve adjustments to a kind of ‘misery research’ we mentioned above.
complex nested system of levels, loci and The stages notion is an important element in a
layers, may help actors to identify alternatives broader conception of public policy that is
for action. Those with a strong top-down per- both persistent and, in empirical research, to a
spective may be assisted by recognising that certain extent misleading. That it is persistent
they have choices between fundamental has to do, as we have shown, with three factors.
restructuring, the adjustment of specific First, the conception is normatively attractive
arrangements, or new ways of curbing street- because it has been founded on the principles
level discretion. Conversely, from a bottom-up of democracy and the trias politica. Second,
perspective there will be questions about it appeals to what Van Gunsteren (1976)
whether what is crucial is to devise new ways of describes as a general quest for control in public
making street-level decisions, or whether there affairs. Third, the orderly neatness of particu-
are institutional and/or structural modifica- larly the stages notion seems psychologically
tions that would need to be made before these attractive.
would be feasible. At the same time, especially inasmuch as this
Between those strong ideological positions conception of public policy points automati-
many will, following Day and Klein (1987), cally to the ‘black box’ of the implementation
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 27


of a specific policy as ‘explaining’ why the studies, empirical questions about its validity
results of that policy are judged as disappoint- are to be answered. Apart from enabling con-
ing or worse, it is misleading. In this respect textual theory formation we have indicated
this conception may keep away from open and that it may aid contextual reflection by practi-
systematic research into what has happened, tioners. Furthermore, the Multiple Governance
how and why. This ‘control trap’ seems to lay at Framework may also fulfil institutional func-
the basis of much of the failure discourse tions; one could think of a research program-
around public policy processes. It is particu- matic use or a function as a map for public
larly this restricted explanatory power that agendas. In fact, the framework is designed to
forms the heart of the criticisms frequently combine two essential tasks: one within social
expressed with regard to the stages ‘model’. science, and one in a broader societal context.
We accept the case for the stages heuristic The first task is providing what Ostrom (1999:
functioning as a general map for the analysis of 41) calls a ‘multi-tier conceptual map’. The
policy processes. As such it is a very general other task goes beyond that: to function as a
map – not a ‘model’ and certainly not a causal multiple heuristic in an institutional sense.
‘theory’ – that has been used to good effect in Then, after all, we are back again with the
some classic policy studies. At the same time, essence of Lasswell’s ‘policy orientation’; double
according to the standards of a moderate bound as it is.
positivism in social science, it is conceptually
neither multi-dimensional nor empirically
open enough to enable and enhance systematic NOTES
theoretical-empirical research in the study of
the policy process. Given the varied landscape 1. Following up Implementing Public Policy that Michael
of the study of the policy process, our stance Hill and Peter Hupe published in 2002, this chapter elabo-
is that such research is both necessary and rates on some of the insights developed in that book. The
possible. major work on it was done during a stay of the latter as a
visiting fellow, March and April 2003, at the Public
Therefore, building further upon Elinor
Management Institute, Katholieke Universiteit Leuven. He
Ostrom’s Institutional Analysis and Devel- thanks Geert Bouckaert, the director of the Institute, and
opment framework, we have linked that Rudolf Maes for their generosity. Wim Hafkamp, the Dean
framework into the mainstream of the politics- of the School of Social Sciences, Erasmus University
and-administration focused policy sciences con- Rotterdam, and Victor Bekkers and Jan Hakvoort, of the
Department of Public Administration, are thanked for
necting it to an explicit governance orientation.
their support of Peter Hupe’s two months stay in Leuven.
As such the Multiple Governance Framework On the 10th of June 2003 Peter Hupe presented a draft of
has a multidimensional as well as a nested char- this chapter in the Centre for Public Governance, research
acter. It provides organising concepts that can group of the Department of Public Administration,
assist low- or middle-range theory formation Erasmus University Rotterdam. He thanks the participants
in that session, particularly Menno Fenger, for their com-
and systematic empirical research with a
ments. On a later version comments were given by Elinor
‘localised’ character. Among other things this Ostrom and George Frederickson, which were highly
means that one level of action, such as the ‘stage’ appreciated by the authors.
of implementation, is not seen as a priori con- 2. It is against this background that we could formulate
fined to one administrative layer. the implementation follows formulation and decision theorem.
This says nothing about who and where, but is a matter of
The framework presented here does not
logic (Hill and Hupe, 2002: 4).
break altogether with the concerns of the many 3. Originally (Hill and Hupe, 2002) we used the term
scholars who have recognised the logical ‘nest- ‘constitutional governance’ here. Now we think the possi-
ing’ or ‘institutional pathways’ affecting many ble reference to ‘the constitution’ – though not intended –
decision processes. In that respect it can still be may make the connotation of this term too fixed and
formal. Essential is the general notion of the Latin verb
said to embody an idea of stages, though a
‘constituere’: forming, building, designing. Therefore we
loose one. When the framework is put forward propose here, instead, to speak of constitutive governance.
as a conceptual device to assist with the fram- 4. Acknowledging two things seems appropriate here.
ing of subsequent theoretical-empirical First, the labels used in this chapter purposely are made
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 28


more ‘empty’ or abstract than in the book (2002) version. Dewey, J. (1927) The Public and its Problems.
Aiming at maximal analytical applicability, in the designa- New York: Holt.
tion of the dimensions of the analytical framework any Dror, Y. (1989) Public Policymaking Reexamined,
connotation with real world epitheta (cf. ‘street-level’) has second edition (first edition: 1968). New
been avoided. Eliminating connotations with real layers
Brunswick, N.J.: Transaction Publishers.
within the framework makes the latter more applicable in
research, in principle, at any layer. Second, at the same time
Fischer, F. (1995) Evaluating Public Policy. Chicago:
the nested character is reconfirmed: the distinguished loci Nelson Hall.
and actors can be observed at each empirical layer in the Fischer, F. (2003) Reframing Public Policy, Oxford:
real world of public administration, formal or not. Oxford University Press.
5. On the basis of an updated extensive literature scan Fox, C.J. and H.T. Miller (1995) Postmodern Public
Hill and Hupe (2002: 123) identified seven categories of Administration: Towards Discourse. Thousand
independent variables seen as having an impact on the Oaks Calif.: Sage.
results of policy processes. These categories are the follow- Frederickson, H.G. and K.B. Smith (2003) The
ing: a) variables dealing with policy characteristics; b) vari- Public Administration Theory Primer. Boulder,
ables on the policy formation sub process; c) variables
Colorado: Westview Press.
about the kind of ‘vertical public administration’ con-
cerned (number of layers; their legitimate competence,
Graaf, H., van de and R. Hoppe (1989) Beleid en
actual relationships, etc.); d) variables regarding the politiek. Bussum: Coutinho.
response from implementers to the policy involved; Gunsteren, H.R. van (1976) The Quest for Control:
e) variables dealing with the horizontal inter-organizational A Critique of the Rational Central Rule-Approach
relationships; f) variables on the response from those in Public Affairs. London: John Wiley and Sons.
affected by the policy, and g) variables regarding the con- Gunsteren, H.R. van (1980) ‘Planning in de verzorg-
figuration of wider macro-environmental factors. ingsstaat: Van chaotisch naar systematisch falen’
in J.K.M. Gevers and R.J. in ’t Veld (eds.), Planning
als maatschappelijke vormgeving Deventer: Van
REFERENCES Loghum Slaterus: 15–32.
Heinrich, C.J. and L.E. Lynn, Jr (eds.) (2000)
Governance and Performance; New Perspectives.
Anderson, J.E. (1975) Public Policy-making. Washington, D.C.: Georgetown University Press.
New York: Holt, Praeger. Hill, M.J. and P.L. Hupe (2002) Implementing Public
Baumgartner, F.R. and B.D. Jones (1993) Agendas Policy: Governance in Theory and Practice. London:
and Instability in American Politics. Chicago: Sage.
University of Chicago Press. Hill, M.J. and P.L. Hupe (2003) ‘The multi-layer
Bowen, E.R. (1982) ‘The Pressman-Wildavsky problem in implementation research’. Public
paradox’. Journal of Public Policy, 2 (1): 1–21. Management Review, 5 (4): 469–488.
Brewer, G.D. (1974) ‘The policy sciences emerge: To Hogwood, B.W. and L. Gunn (1984) Policy Analysis for
nurture and structure a discipline’. Policy Sciences, the Real World. Oxford: Oxford University Press.
5 (3) (September): 239–44. Hoogerwerf, A. (ed.) (1978) Overheidsbeleid: Een
Brewer, G.D. and P. DeLeon (1983) The Foundation inleiding in de beleidswetenschap. Alphen aan den
of Policy Analysis. Monterey, California: Brooks/ Rijn: Samsom.
Cole. Howlett, M. and M. Ramesh (2003) Studying Public
Cohen, M.D., March, J.G. and Olsen, J.P. (1972) Policy: Policy Cycles and Policy Subsystems. Second
‘A garbage can model of organizational choice’, edition, Ontario: Oxford University Press.
Administrative Science Quarterly, 17 (1): 1–25. Jenkins, W.I. (1978) Policy Analysis: A Political and
Day, P. and R. Klein (1987) Accountabilities. London: Organisational Perspective. London: Martin
Tavistock. Robertson.
Davies, H.T.O., S.M. Nutley and P.C. Smith (eds.) Jenkins-Smith, H. and P.A. Sabatier (1993) ‘The
(2000) What Works? Bristol: Policy Press. study of the public policy process’ in P.A. Sabatier
DeLeon, P. (1999) The Stages Approach to the Policy and H.C. Jenkins-Smith (eds.) (1993) Policy
Process: What Has it Done? Where Is It Going? Change and Learning: An Advocacy Coalition
in: P.A. Sabatier (ed.) (1999) Theories of the Approach. Boulder, Colorado: Westview Press.
Policy Process. Boulder, Colorado: Westview Press, John, P. (1998) Analysing Public Policy. London:
19–32. Pinter.
Derthick, M. (1972) New Towns in Town: Why a Jones, Ch. (1970) An Introduction to the Study of
Federal Program Failed. Washington, D.C.: Urban Public Policy (first edition). Belmont, California:
Institute. Wadsworth.
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 29


Kingdon, J.D. (1995) Agendas, Alternatives and Mack, R. (1971) Planning and Uncertainty.
Public Policies (second edition). Boston: Little, New York: John Wiley.
Brown and Company. Matland, R.E. (1995) ‘Synthesizing the implemen-
Kiser, L.L. and E. Ostrom (1982) ‘The Three Worlds of tation literature: The ambiguity-conflict model
Action: A Metatheoretical Synthesis of Institutional of policy implementation’, Journal of Public
Approaches’ in E. Ostrom (ed.) (1982) Strategies of Administration Research and Theory, 5 (2): 145–74.
Political Inquiry. Beverly Hills, California: Sage. May, J. and A.B. Wildavsky (eds.) (1978) The Policy
Kooiman, J. (ed.) (1993) Modern Governance: New Cycle. Beverly Hills, California: Sage Publications.
Government-Society Interactions. London: Sage. Meier, K.J. (1999) ‘Are we sure Lasswell did it this
Kooiman, J. (1999) ‘Social-Political Governance: way? Lester, Goggin and implementation
Overview, Reflections and Design’, Public research’, Policy Currents, 9 (1): 5–8.
Management (Review), 1 (1): 67–92. Milward, H.B. and K.G. Provan (1999) ‘How net-
Kooiman, J. (2003) Governing as Governance. London: works are governed’, unpublished paper.
Sage. Nakamura, R. (1987) ‘The textbook policy process
Kuypers, G. (1980) Beginselen van beleidsontwikkel- and implementation research’. Policy Studies
ing (two volumes). Muiderberg: Coutinho. Review, 7 (2) (Autumn): 142–54.
Lasswell, H.D. (1951a) ‘The policy orientation’ in Ostrom, E. (1999) ‘Institutional Rational Choice: An
Lerner, D. and H.D. Laswell (eds.) (1951) The Assessment of the Institutional Analysis and
Policy Sciences. Stanford, California: Stanford Development Framework’ in P.A. Sabatier (ed.)
University Press. (1999) Theories of the Policy Process. Boulder,
Lasswell, H.D. (1951b) ‘Democratic character’ in Colorado: Westview Press, 35–71.
The Political Writings of Harold D. Lasswell. O’Toole, L.J., Jr (1986) ‘Policy recommendations for
Glencoe, Illinois: Free Press. multi-actor implementation: An assessment of
Lasswell, H.D. (1956) The Decision Process: Seven the field’. Journal of Public Policy, 6 (2): 181–210.
Categories of Functional Analysis. College Park, O’Toole, L.J., Jr (2000) ‘Research on Policy Imple-
Maryland: University of Maryland. mentation: Assessment and Prospects’. Journal
Lasswell, H.D. (1968) ‘The policy sciences’ in The of Public Administration Research and Theory,
Encyclopedia of the Social Sciences, vol. 12. 10 (2): 263–88.
New York: Macmillan/Free Press. Palumbo, D. (1988) Public Policy in America. New
Lasswell, H.D. (1970) ‘The emerging conception of York: Harcourt, Brace, Jovanovich.
the policy sciences’. Policy Sciences, 1: 3–14. Parsons, W. (1995) Public Policy: An Introduction to
Lasswell, H.D. (1971) A Pre-view of Policy Sciences. the Theory and Practice of Policy Analysis.
New York: Elsevier. Cheltenham, UK: Edward Elgar.
Linder, S.H. and B.G. Peters (1987) ‘A design per- Perrow, C. (1999) Normal Accidents: Living with
spective on policy implementation: The fallacies High Risk Technologies. Princeton: Princeton
of misplaced prescription’. Policy Studies Review, University Press.
6 (3): 459–75. Pierre, J. and B.G. Peters (2000) Governance, Politics
Lindblom, C.E. (1959) The science of muddling and the State. New York: St Martin’s Press.
through, Public Administration Review, 19: 78–88. Pressman J. and A. Wildavsky (1973, first edition)
Lindblom, C.E. (1968) The Policy-making Process. Implementation: How Great Expectations Are
Englewood Cliffs, N.J.: Prentice-Hall. Dashed in Oakland; Or, Why It’s Amazing that
Lindblom, C.E. and E.J. Woodhouse (1993) The Federal Programs Work at All, This Being a Saga of
Policy-making Process, third edition. Englewood the Economic Development Administration as Told
Cliffs, N.J.: Prentice-Hall. by Two Sympathetic Observers Who seek to Build
Lipsky, M. (1980) Street-level Bureaucracy: Dilemmas Morals on a Foundation of Ruined Hopes.
of the individual in public services. New York: Berkeley: University of California Press.
Russell Sage Foundation. Richards, D. and M.J. Smith (2002) Governance and
Lynn, L.E., Jr (1981) Managing the Public’s Business. Public Policy in the UK. Oxford: Oxford University
New York: Free Press. Press.
Lynn, L.E., Jr (1987) Managing Public Policy. Boston, Rose, R. (1973) ‘Comparing public policy: An
Mass.: Little, Brown and Company. overview’. European Journal of Political Research,
Lynn, L.E., Jr, C. Heinrich and C.J. Hill (2001) 1: 67–94.
Improving Governance: A New Logic for Empirical Rothstein, B. (1998) Just Institutions Matter: The
Research. Washington DC: Georgetown University Moral and Political Logic of the Universal Welfare
Press. State. Cambridge: Cambridge University Press.
Peter-3383-Chapter-01.qxd 6/8/2006 7:31 PM Page 30


Sabatier, P.A. (1991) ‘Towards better theories of the Schlager, E. (1999) ‘A Comparison of Frameworks,
policy process’. Political Science and Politics, 24 (2) Theories, and Models of Policy Processes’ in
(June): 147–56. P.A. Sabatier (ed.) (1999), Theories of the Policy
Sabatier, P.A. (1999) ‘The need for better theories’ in Process. Boulder, Colorado: Westview Press, 233–60.
P.A. Sabatier (ed.) (1999) Theories of the Policy Selznick, P. (1949) TVA and the Grass Roots.
Process. Boulder, Colorado: Westview Press, 3–17. Berkeley: University of California Press.
Sabatier, P.A. (ed.) (1999) Theories of the Policy Simon, H.A. (1945) Administrative Behavior. First
Process. Boulder, Colorado: Westview Press. edition. New York: Free Press.
Sabatier, P.A. and H. Jenkins-Smith (1999) ‘The Stone D.A. (1989) ‘Causal stories and the formation
Advocacy Coalition Framework: An Assessment’ of policy agendas’, Political Science Quarterly,
in P.A. Sabatier (ed.) (1999) Theories of the Policy 104: 281–300.
Process. Boulder, Colorado: Westview Press, Waldo, D. (1946) The Administrative State: A Study of
117–66. the Political Theory of American Public Adminis-
Schlager, E. (1997) ‘A response to Kim Quaile Hill’s tration. New York: The Ronald Press Company.
In Search of Policy Theory’, Policy Currents, Wilson, W. (1887) ‘The study of administration’,
7 (June): 14–5. Political Science Quarterly, 2: 197–222.
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 31

The Policy Sciences: Past,
Present, and Future


INTRODUCTION frustrations, and offer some areas for potential

future growth in light of their past.
The Policy Sciences orientation has primarily
been attributed to Harold D. Lasswell, writing
in the late 1940s and early 1950s, most pro-
minently articulated in his essay ‘The Policy
Orientation’ as the opening chapter to The If the study of public policy and providing
Policy Sciences (1951a; also see Lasswell 1971).1 advice to policymakers has a relatively short
The Policy Sciences approach was explicitly academic lineage, from a practitioner stand-
focused on the rigorous application of a point, it reflects a storied legacy. Rulers have
variety of science endeavors (hence, the plural been the recipient of policy advice since at least
usage of ‘sciences’) to issues affecting the the recording of history (see Goldhamer 1978
processes of governance; along these lines, for details); advisers to whomever ruled were
Lasswell wrote of the knowledge ‘in and of ’ the rarely lacking for reasons easy to imagine.
act of governing; that is, the process and sub- However, there is a clear distinction between
stance of governance (respectively). In addi- the earlier purveyors of policy advice and what
tion, there was a clear understanding of the later came to be known as the policy sciences,
necessity of democratic processes or what he namely that advisers to rulers rarely relied on
defined as the ‘policy sciences of democracy’ extensive policy research nor carefully crafted
(e.g., Lasswell 1951b). reports. Their advice, whatever its merits,
Since this time, however, the policy sciences, was usually shaped by their ‘power behind
as both an academic discipline and an applied the throne’ experience. For this reason, policy
craft, have experienced a checkered pattern of advisers were invariably members of the
growth, application, and contraction. This essay royalty or the ruler’s personal attendants; there
will briefly delineate the initial purposes of the is scant record of laypersons serving in an
policy sciences, indicate their development and advisory capacity.2
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 32


In contrast, the American university at the later articulated by Brewer (1974) and, subse-
turn of the 20th century housed a number of quently, Brewer and DeLeon (1983) (also see
disciplinary approaches, such as political Anderson 1975/1979 and Jones 1970/1984)
science, anthropology, geography, law, psy- includes Policy Initiation, Policy Estimation,
chology, sociology, and public health, that were Selection, Program Implementation, Program
the natural precursors to the study of public Evaluation, and Policy Termination. Providing
affairs in general and the activities of govern- a conceptual breakdown of policy formulation
ment in particular. Heineman et al. 2002 (also and execution, with each stage possessing
Fischer 2003) have singled out public adminis- unique characteristics, the stages approach
tration and political science as progenitors in (referred to as a ‘heuristic’ by Sabatier 1999
this particular focus. However, the policy and Fischer 2003) offers a mechanism to
sciences approach and their authors have care- achieve a multidisciplinary and value-oriented
fully distinguished themselves from these early approach to policy. In practice, however,
disciplinary contributions by offering three researchers have broken the stages into dis-
defining characteristics of the approach: jointed units, as we shall see below, betraying
the holistic intent of Lasswell’s process and
1. The policy sciences are explicitly problem- resulting in an implied linear rationality
oriented, rejecting the study of a specific devoid of idea and value (DeLeon 1999).
phenomenon per se; the societal or politi- Paul Sabatier (1993 and 1999), along with
cal question of ‘so what’ has always been Robert Nakamura (1987) and others, have been
integral to the policy sciences. Likewise, very critical of the stages process, noting,
problems occur in particular contexts that among other things, that it neglects ‘the role of
must be considered in terms of both the ideas – particularly ideas involving the rela-
analysis and later recommendations. tively technical aspects of the policy debates –
2. The policy sciences are distinctively multi- in policy evolution’ (Sabatier 1993: 15). He
disciplinary in their intellectual and practi- (and co-author Hank Jenkins-Smith) have
cal approaches; virtually every social or severely criticized the policy process frame-
political problem has components tied to work for its theoretic shortcomings, specifi-
varying academic disciplines without cally (Jenkins-Smith and Sabatier 1993: 3–4;
falling clearly into any one discipline’s emphases in original):
exclusive domain.
3. The policy sciences’ approach is explicitly • ‘The stages model is not really a causal
value oriented; in many cases, the central model at all.’ That is, it did not lend itself to
theme deals with the democratic ethos and prediction or even how one stage transi-
human dignity, thus denying the strictures tioned to another.
of logical positivism that were so prevalent • ‘The stages model does not provide a clear
in the American social sciences in the 20th basis for empirical hypothesis testing.’ That
century.3 This value orientation recognizes is, it is not amenable to amendment, con-
that no social problem is without a value firmation, or verification.
component. As such, in order to under- • ‘The stages heuristic suffers from descriptive
stand a problem, one must acknowledge inaccuracy in posing a series of stages …’
its value components. Similarly, no policy • ‘The stages metaphor suffers from a built-
analyst is without her/his own values, in legalistic, top-down focus.’
which also must be addressed (Amy 1984; • ‘The stages metaphor inappropriately
Stone 1998).4 emphasizes the policy cycle as the temporal
unit of analysis.’
The policy sciences have been operational- • ‘The stages metaphor fails to provide a
ized as a process delineated in terms of discrete good vehicle for integrating the roles of
stages in the policy process. The decision policy analysis and policy-oriented learning
process originally proposed by Lasswell (1956),5 throughout the public policy process.’
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 33


Many of Sabatier’s observations are correct Lindblom and Cohen 1979), or, if it is, to what
when one views the policy process approach effect or purpose, as David Kirp (1991) talks
(or what Sabatier labels a ‘metaphor’ or semi-facetiously (one hopes) about the ‘end of
‘heuristic’) but there is little evidence that such policy analysis’? Heineman and his colleagues
a set of criteria was ever intended (DeLeon (2002, 1, 9) speak of this concern:
1999: 24 and Brunner 1991) or even appropri- … despite the development of sophisticated methods of
ate. Rather, the stages approach is designed to inquiry, policy analysis has not had a major substantive
feature different stages of the policy process, impact on policymakers. Policy analysts have remained
highlighting their distinct functions and fea- distant from power centers where policy decisions are
made… . In this environment, the values of analytical
tures, ranging from Policy Initiation to Policy rigor and logic have given way to political necessities.
Termination, and provide the necessary guide-
lines. For example, different mechanisms Radin (2000) provides a counter to these
attend policy estimation compared to policy charges of despair, arguing that policy analysis
implementation. A review of the policy litera- has not only produced excellent and effective
ture over the past forty years indicates that the policy research but has had a distinct effect on
stages approach has done precisely that (see policymaking, although not as much as a pro-
DeLeon 1999: 22). In that sense, Lasswell’s ponent would have preferred. And the policy
model continues as a beacon, although, as we ‘market’ place would be supportive, in terms of
will see below, the stages’ particular roles have the number of policy analysts employed in a
been amended by lessons drawn from various lengthy list of policy agencies.
political events. We need not necessarily agree with all of the
claims of the demise of the policy sciences and
certainly not the utility of policy research in
THE APPLICATION OF THE general. Still, one can assert that the Lasswellian
POLICY SCIENCES charge for the policy sciences in either applica-
tion or concept has not been universally real-
ized. Let us take, then, a moment to chronicle
Moving the policy sciences from the halls of the political events that have had a noticeable
academe to the offices of government largely effect on the policy sciences to better appreci-
occurred on the federal level during the 1960s ate their evolution.
(see Radin 2000), such that, by the 1980s,
virtually every federal office had an analytic
office. Since then, many states (including
memberships in inter-state consortia, such as
the National Conference of State Legislatures)
have built up policy analysis shops to the In general, two paths have been proposed to
extent budgets permit. In addition, for-hire outline the development of the policy sciences.
‘think tanks’ of most every political orientation Beryl Radin (2000) has characterized the insti-
have proliferated. Every public sector official tutional growth of the policy approach, largely
would agree that more information on which relying on the (fictional) histories of an ‘old
to base decisions and policies is better than school’ economist cum policy analyst juxta-
less. To serve that demand, virtually every uni- posed with a ‘younger,’ university-trained policy
versity has a graduate program in public affairs analyst. Through them, she casts an institutional
(or has re-tooled its public administration framework on the policy sciences, indicating
program) to fill the apparent need for sophis- their march from a limited analytic approach,
ticated policy analysis. Yet the turn of the 21st practiced by relatively few practitioners, to a
century has hardly ushered in a Golden Age of growing number of government institutions.
Policy Advice. One needs to ask why policy Specifically, Radin notes the emergence of ana-
scientists increasingly voice the perception that lytic studies from the RAND Corporation to the
their work is not being utilized (Weiss 1980; US Department of Defense (DoD) in the early
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 34


1960s (under the guise of ‘systems analysis’ and campaign. However, after the war, while the
a Programmed Planning and Budget System, ‘supply’ side of the policy equation was ready,
PPBS). From its apparent success in the Defense there was little on the ‘demand’ side. Policy-
Department, President Lyndon Johnson man- makers, perhaps tired of the wartime exigen-
dated the government-wide diffusion of PPBS, cies or perhaps enveloped with a return to
most visibly in the Department of Health, peacetime ‘normalcy,’ did not take these newly
Education, and Welfare, in the mid-1960s. honed skills into consideration.
Although success in the DoD of PPBS’s DoD Still, these wartime activities established an
was never duplicated elsewhere (see Wildavsky important illustration of the ability of the
1984; Schick 1973), the analysis orientation social sciences to direct problem-oriented
soon was adopted by a number of federal analysis to urgent public issues, in this case
offices, state agencies, and a number of analytic assuring victory over the Axis powers. As a
consultant groups (see Fischer 1993 and Ricci point of interest, Lasswell and Kaplan spent the
1984). Thus, Radin views the growth of the war with the Library of Congress studying how
policy sciences as a ‘growth industry,’ in which a to best utilize (and protect against) propa-
few select government agencies first adopted an ganda. These realizations led directly to the
explicitly innovative analytic approach, others formation of the National Science Foundation
adopted similar approaches, and a correspond- and the Council of Economic Advisors (see
ing industry developed. Polsby 1984), as well as research facilities such
DeLeon (1988) has cast the growth of the as The RAND Corporation (Smith 1966). Yet,
policy analysis in a consonant but more com- as a result of the imbalance of supply and
plicated manner, in which he ties the growth of demand, the policy analytic approach was
specific analytic ‘lessons learned’ to given more or less quiescent until the 1960s, when
political events. In his view, political condi- the assassination of President John Kennedy
tions effectively supplied analysts with particu- and the succession of President Lyndon Johnson
lar scenarios and data to which they could turn conspired to declare a War on Poverty.
their skills, thus impressing both their imme- The policy sciences faced another opportu-
diate policymaking clients and the larger pop- nity to practice their skills during the War on
ulation, as well with the perspicacity of the Poverty, namely the need to confront social
approach. In particular, he suggested that the complexity and identify the central problem,
resulting policy initiatives (which he termed namely, the pervasive poverty – largely fueled
‘supply’) and policymakers’ requirements by the emerging civil rights demonstrations –
(‘demand’) collaborated to define the specific afflicting what Harrington (1960) called ‘the
development of the policy sciences. The two, other America.’ Even though poverty had
he posits, must be synchronous for a synergis- always been a part of the American fabric, US
tic relationship to develop. policymakers found that they were remark-
DeLeon (1988) initially set forth five politi- ably uninformed about the conditions and
cal conditions that articulated the policy extent of poverty in America. Social scientists
sciences: the Second World War; Lyndon moved aggressively into this knowledge gap
Johnson’s ‘War on Poverty;’ America’s involve- with unbridled enthusiasm, if not always rele-
ment in the Vietnam War; the ‘Watergate vant insights, producing what Moynihan
Affair’ and the ensuing impeachment of (1969) called ‘maximum feasible misunder-
President Richard Nixon; and the 1970’s standing.’ Policymakers proved to be inher-
Energy Crisis. ently limited in their views by their unique set
During the Second World War, the United of experiences.
States marshaled an unprecedented array of To engage the campaign against poverty, a
social scientists – economists, political scien- vast (if not necessarily coordinated) number
tists, psychologists, etc. – to support the war of social programs (Model Cities, VISTA,
effort, ranging from managing the domestic Headstart, and a host of programs out of the
economy to coordinating the strategic bombing Office of Economic Opportunity, OEO) was
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 35


initiated, with important milestones being Vietnam from the early 1960s until the early
achieved, especially in the improved statistical 1970s. The decade-long Vietnam war brought
measures of what constituted poverty and the detached policy analysis instruments,
evaluation measures to assess the various anti- including applied systems analytic techniques,
poverty programs (Rivlin 1970), but poverty to the intimate horrors of combat, with politi-
was insistently endemic. Walter Williams cal conditions exacerbated by the growing
(1998), looking back on his days in the OEO, domestic civil unrest as to its conduct and, of
has suggested that these were the ‘glory days’ of course, the loss of lives. The war was closely
policy analysis. Other OEO veterans, such as monitored and managed by the Secretary of
Robert Levine (1970), were more reserved, Defense’s office, with close oversight from a
while some, such as Murray (1984), indicated succession of presidents. It became increas-
that, with the advent of the anti-poverty pro- ingly obvious that analytic rigor – specified in
grams, the American poor was actually ‘losing terms such as ‘body counts,’ sorties flown, and
ground.’ At best, policy analysts were forced to hamlets ‘pacified’ – and ‘rational’ decisionmak-
confront the immense complexity of the social ing were not indicative of the growing rancor
condition. Later, DeLeon was to ask ‘if ten of the war. There was repeated evidence that
years and billions of dollars had produced any ‘hard and fast’ numbers were being manipu-
discernible, let alone effective, relief ’ (DeLeon lated to serve political purposes. Moreover, sys-
1988: 61). tems analysis was not intellectually able to
The ‘policy lessons’ ascribed to the War capture analytically the almost daily changes in
on Poverty were three-fold. In the first place, the war’s activities, occurring on both the
the policy sciences were thwarted by policy- international and domestic arenas (see Gelb
makers’ inability to understand and respond to and Betts 1979).
the complexity of poverty as presented to To return to DeLeon’s metaphor, during the
them. There was an inability to formulate per- Vietnam War and its domestic ramifications, the
suasive arguments in the policy initiation policy ‘supply’ could not square with the politi-
stage. Second, policymakers and analysts ‘dis- cal ‘demand.’ In terms of policy estimation, sys-
covered’ the vagaries of implementation (see tems analysis, one of the apparent US advantages
Pressman and Wildavsky 1984 for a particularly of defense policymaking, was surprisingly
cogent example); in retrospect, this ‘imple- myopic and was a partial contributor to the
mentation blinder’ could have been foretold by ultimate US failures in Vietnam (Gray 1971).
the public administration scholars had they Department of Defense analysts could not
been engaged, but the difficulties encountered reflect the required (and respective) political
eviscerated many programs (see Derthick wills necessary to triumph, as The Pentagon
1972). Finally, and arguably the most success- Papers (Sheehan 1971) subsequently showed. On
ful learning experience, policymakers learned the other hand, Frances FitzGerald’s Fire in the
to demand thorough evaluations of a variety of Lake (1972) foretold the inevitable American
policy programs, even though, at that time, military disaster, as the North Vietnamese were
evaluators were still methodologically naïve; willing to incur whatever losses were needed in
more to the point, however, both parties failed what they saw as the defense of their nation.
to realize the political nature inherent in these Even if the war effort itself had been well con-
exercises. In effect, policy soldiers in the War ducted (surely an arguable point), the manner in
on Poverty failed to frame the right questions, which the war was visualized and projected by
account for the factors that affected implemen- the analytic community left much to be desired,
tation, and were unable to evaluate the pro- a shortcoming widely noted in the domestic
grams with discernment. In short, the War on anti-war community and, ultimately, in policy-
Poverty served as an annealing agent for policy making circles.
research. The policy sciences learned that, in spite of
The disappointment of the policy sciences the best analysis, good policy analysis is insepa-
recurred with the unfortunate experiences in rable from values. The events surrounding the
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 36


re-election of President Richard Nixon in 1968, considerations (e.g., untapped petroleum

his Administration’s heavy-handed attempts to reserves and complex technical modeling; see
‘cover up’ the incriminating evidence, and his Commoner 1979 and Greenberger et al. 1984)
willingness to covertly prosecute Vietnam war but the basic decisions were decidedly political
protester Daniel Ellsberg, combined to lead the (that is, not driven by analysis), as President
Congress to the potential impeachment of a Nixon declared ‘Project Independence,’ Presi-
sitting American, averted only because the dent Carter intoned that energy independence
President chose to resign in ignominy rather represented the ‘moral equivalency of war,’ and
than face certain impeachment proceedings President Ford created a new Department of
(Olson, 2003). The overwhelming evidence of Energy. Policy estimation was particularly
foul play in the highest councils of the US gov- found to be lacking, partially because of the
ernment brought home the idea that moral inherent technical demand of the problem,
norms and values were central to the activities but also because of the imputed political posi-
of government. The Ethics in Government Act tions implied by the various energy options.
(1978) was only the most manifest recognition There was seemingly a convergence between
that normative standards were central to gover- ‘analytic supply’ and ‘government demand,’
nance processes, validating, as it were, one of the yet no policy consensus was forthcoming, a
central tenets of the policy sciences. Regardless, condition that did little to enshrine the policy
however, few will ever forget the President of sciences approach with either its immediate
the United States protesting, ‘I am not a crook,’ clients (government officials) or its ultimate
and its effect of the public’s trust in its elected ones (the citizenry). The experience of the
government. energy crisis highlighted the lack of negotia-
The energy crises of the 1970s provided a tion and resolution.
virtual test bed for the best analytic efforts the Since DeLeon (1988) first posed that these
country could offer. Partially as a result of an historical events shaped the development of
Arab boycott on petroleum production, the policy sciences, there have been two addi-
record-high gasoline prices spiraled through- tional ‘events’ that, arguably, have been equally
out the nation. As a result, the public was inun- definitive: The impeachment of President
dated with analyses and formulae regarding the Bill Clinton (in the late 1990s); and the hor-
level of petroleum reserves (domestic and rific terrorist attacks on the United States
world-wide), competing energy sources (e.g., in September 2001, with the subsequent deci-
nuclear vs. petroleum vs. coal) over differing sions by President George W. Bush to declare
(projected) time frames, with a backdrop of war on world-wide terrorist movements, and,
threatened national security. With this plethora in keeping with this charge, to go to war
of technical data, the analytic community was against Afghanistan and Iraq. Without pre-
seemingly prepared to bring light out of the tending to offer exhaustive histories of these
darkness (for example, see Landsberg 1972). actions (indeed, the latter is on-going and, as
But, this was not to be the case; as Weyant was such, impossible to put into perspective), let us
later to note, ‘perhaps as many as two-thirds of indicate their broad outlines and effects.6
the [energy] models failed to achieve their The policy sciences approach was seemingly
avowed purposes in the form of direct appli- well represented in the Clinton White House as
cation to policy problems’ (quoted in Weyant President Clinton was widely (and glowingly)
1980: 212). Aaron Wildavsky and Ellen acknowledged to be the presidential prototype
Tannenbaum (1981) poignantly referred to this of a ‘policy wonk’ (Woodward 1994). But, as
period as ‘the politics of mistrust.’ Faced with a events transpired over the course of his Presi-
chorus of demands to ‘fix the problem,’ the dency, politics, pure and simple, trumped analy-
research community was unable to provide an sis during the administration, as partisan agenda
acceptable supply. and personal norms assumed more importance
The contrast was both remarkable and than policy development. For whatever reasons,
apparent: Energy policy was awash in technical President Bill Clinton was a political lightning
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 37


rod, an almost constant target of personal presence of mounting information, due to

calumny, from the day of his inauguration in prior belief preferences and misunderstand-
1992 until he left office in January 2000. Minor ings, possibly multiplied by what Janis (1983)
peccadilloes (like the furor over the so-called earlier labeled ‘groupthink.’ The September
White House ‘travelgate’ incident, in which 11th air-borne attacks on New York City and
Clinton staffers bureaucratically hijacked the Washington DC, followed shortly thereafter by
travel arrangements for the White House press; the subsequent retaliatory American attacks on
see Drew 1994), unfortunate tragedies (the sui- Afghanistan and (less directly) Iraq, can be
cide of White House Counselor Vincent Foster), viewed in terms of shortcomings in the ‘ratio-
and the curse of a few proposals that were nal’ policy advice, representing a failure to pre-
downright ill-fated (the Clinton Health Care sciently see looming disasters and to act
initiatives; see Skocpal 1997) constantly plagued accordingly. Thousands of lives were lost, lead-
the Administration. These tended to over- ing citizen to later disagree over whether the
shadow some very major successes (e.g., presid- intelligence communities should have been
ing over the reversal of the Federal deficit to a able to ‘follow the dots’ that would have
position of surplus) and the continued faith revealed the plans of Al Qaeda. Others have
accorded him by the voting public. Taken argued whether US spokespersons callously
in combination, they all contributed to a rich misrepresented (or seriously politicized) the
political legacy. evidence at hand, or, more probably, ignored
Regardless of these political comings and the possibility that the evidence was sketchy at
goings, his Administration will mostly be best (Woodward 2004 and Clarke 2004). Core
remembered for his intemperate acts with a values once held dearly (e.g., habeas corpus) or
young White House female intern and his sub- acknowledged by international conventions
sequent attempts to hide these acts from the (such as the Geneva Conventions regarding
American members of his administration, his prisoner of war treatment) are being chal-
own family, and the voting public. Especially in lenged as little more than bothersome inconve-
light of his vehement initial denials, Clinton’s niences when confronted with possible issues
remarkable parsing concerning the meaning of of national survival.
‘is’ and his later admissions of these events will At the very least, the United States finds
become the unfortunate signature moments of itself engaged in a second war in the Persian
his second term in office. Gulf in a decade with serious loss of life, with
President Clinton was the first American goals that are increasingly questioned, restated,
President in well over 100 years to be brought and debated. The situation has been exacer-
before the Congress in an act of impeachment. bated by the partisanship expected in a presi-
Although he was eventually found by the US dential election, making an ‘objective’ reading
Senate to be innocent of the charges leveled for of the ‘facts’ more nettlesome than usual. The
impeachment (see Johnson 2001 and Baker one clear lesson, however, from this com-
2001 for particulars), the proceedings were as mitment resonates from Vietnam, namely, an
politically value laden as few events in recent understanding of the ‘human element,’ i.e.,
American history. Thus, mirroring in many nations are more susceptible to being ‘liber-
ways Nixon’s Watergate scandal, the impeach- ated,’ not ‘occupied,’ and it is their (rather than
ment of President Clinton, although conducted the CIA or DoD’s) reading of the visceral tea
under a full canopy of legal proceedings and leaves that makes the difference. The Selection
media attention, was a reminder that moral stage seems especially vulnerable to this
considerations can dictate seemingly analytic episode, as earlier positions seemed to color
decisions, as Clinton’s subsequent autobiogra- the evidence at hand. In many ways, policy
phy (2004) indicates. analysis failed to marshal the evidence that
An inherent shortcoming for most policy policymakers could have gleaned from past
recommendations has been the inability to policy failures, suggesting a bigger human
prepare for future contingencies, even in the challenge to the policy sciences.
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 38


These constellations of events have mani- context always matters. And policymakers
fested themselves in a worrisome position for must be willing to see and accept the products
the policy sciences, that is, a general disillusion of the policy sciences and their proponents.
in the way in which the American people view Confronting complex problems not only
their government and its processes and, as a requires a multidisciplinary approach. Many
result, the role of the policy sciences. From the have held that the key to the policy process
immense national pride that characterized the basically must accommodate a broad under-
victory over totalitarian forces in the Second standing with accurate problem framing
World War, the American voter has suffered a (Schön and Rein 1994). In spite of a set of
series of on-going disappointments, ranging strong analytic skills, the policy sciences are
from what many consider to be a failed War on inseparable from values, normative concepts,
Poverty to a failed war in Southeast Asia, to the and political ideology. Yet, the policy sciences
unprecedented (in living memory) attacks on have not regularly integrated complexity and
Washington DC and New York City, to the fail- values with policymaking. One needs to ask:
ure of US troops to be treated as ‘liberators’ in Why should the nominal recipients of the
Iraq. Watergate cast a darkening pall on the policy sciences subscribe to them if they do not
American body politic; the Clinton Adminis- manifest the values and intuitions of the client
tration did little to dispel those clouds; and policymaker? To this question, one needs to
President Bush – who was elected in 2000 on a add the question of democratic procedures, a
platform of lowering political dissonance in tenet virtually everybody would agree upon
Washington – has not been able to reduce the until the important issues of detail emerge (see
partisan tensions. Thus, scholars like E.J. DeLeon 1997; Barber 1984; Dahl 1970/1990),
Dionne writes Why Americans Hate Politics e.g., does direct democracy have a realistic
(1991) or Joseph Nye (and colleagues) edit a place in a representative democracy?
book Why Americans Don’t Trust Their Govern-
ment (1997) disparaging the American body
politic and, with it, the policy advice industry.
Most damaging to the policy sciences’ tradi-
tion is Christopher Lasch’s pointed and hardly
irrelevant question: ‘does democracy have a Let us submit the following proposition: that,
future? … It isn’t a question of whether democ- on face value, the policy sciences approach has
racy can survive … [it] is whether democracy inherent strengths, both in ‘knowledge of ’ gov-
deserves to survive’ (Lasch 1995: 1 and 85; ernment (i.e., understanding the processes),
emphases added). and ‘knowledge in government’ (what they
To be sure, political activities are not syn- offer policymakers in terms of substance). But
onymous with the practice of the policy what we have seen above is that the juncture of
sciences. But the two indisputably reside in the the Lasswellian vision of the policy sciences
same policy space. For the policy sciences to with workaday policymaking has not been
meet the goals of improving the processes and realized, often because the analytic ‘supply’ has
results of government through a rigorous not coincided with the policymaking ‘demand.’
application of its central themes, the failures of So how can one best prescribe the policy ana-
the body politics naturally must be at least par- lytic skills that policymakers request from their
tially ascribed to the policy advice industry, advisors and how can the policy sciences best
which includes the policy sciences. Historical respond with integrity? Inherent in this ques-
examples have shown that the supply and tion is a principal assumption: policy scien-
demand conditions for the policy sciences are tists, in the words of Aaron Wildavsky (1979),
necessary but not sufficient for good policy. must ‘speak truth to power.’ Without access to
Supply and demand for policy analysis needs and trust from policymakers, the policy
to be coordinated around the right issue at the sciences lose their sine quo non. They are, from
right time to the right person/agency. As such, their earliest iteration, an applied discipline; if
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 39


the policy sciences become irrelevant through identity, economic instability, environmental
lack of application or, to borrow another degradation, militaristic ideals, and a history
metaphor, if (policy) advice does not match and fears of marginalization, or what Samuel
(political) consent, then the policy sciences Huntington (1996) has referred to a as ‘clash of
will have failed to meet the challenges spelled civilizations and the remaking of world order.’
out by its earliest advocates. The ideological debates between public and pri-
Of course, we should not necessarily abide vate sectors continue, with the relatively new
by a counsel of despair. As policy scientists, infusion of the nonprofit sector (both nation-
we need to recognize that a variety of condi- ally and internationally) assuming an increas-
tions have changed (witness, for one, the ingly important role in service provision. The
revolution in technologies that directly affect resulting process makes it clear that no specific
the productivity levels of the American sector has a monopoly on the processes or prod-
worker) and, moreover, that no one has ever ucts of governance. Moreover, the bifurcation
suggested that the policy sciences must of the American body politic along largely par-
remain constant to their original vision; tisan lines often makes agreement on specific
mutatis mutandis in terms of context and policy issues problematic.
processes must be part of the policy sciences.
In this section, then, let us outline a postpos-
itivist approach to enhance the policy Postpositivism and the
sciences and a few relatively new methodolo- Policy Sciences
gies and approaches (e.g., social network
analysis, participatory policy analysis, and Q- The policy sciences community has never been
methodology) that the policy sciences might blind to the presence of and competition
wish to apply to a changing world. But, as we between competing values, but perhaps values
will show, none of these (or the combination) have been under-represented in policy research
will act as an analytic Rosetta Stone; the policy taken as a whole. This amendment, then, is to
community, its issues, and its membership are find conceptual approaches and tools that
too diverse and, in some cases, oppositional. accommodate the diversity. To that end,
There are, in short, no easy answers. numerous authors (Fischer 1998; Schneider
Let us first offer a few milestones from which and Ingram 1997; Bobrow and Dryzek 1987;
to view the landscape as a means to assess in Schön and Rein 1994; Dryzek 1990; Hajer and
what ways conditions have changed since Wagenaar 2003; Forester 1999) have advanced
Lasswell. The early twenty-first century brings a a postpostivist perspective. Although not uni-
world that is increasingly interdependent, where versally accepted (see Sabatier 1999 and Lynn
regional issues have global reach. Economies 1999), its advantages warrant discussion.
and social systems are inextricably connected In response to the shortcomings of framing
and interdependent, as transnational economic and issue understanding, as well as presenting
activities spur local responses; for instance, basi- a more encompassing epistemological per-
cally invisible (and all but invulnerable) life spective, the postpositivist perspective – which
forms called ‘prions’ can result in ‘bovine includes a variety of differing methodologies,
spongiform encephalopathy’ (a degenerative such as ‘deliberative’ (Forester 1999), policy
and fatal brain disease, or Creutzfeldt-Jacob discourse (Hajer 1993), argumentation (Fischer
syndrome in cattle, or what is popularly referred and Forester 1993), interpretative, and narra-
to as ‘mad cow’ disease), a local event that affects tive (Roe 1994) – provides a more thorough
the food supply of nations oceans away, upset- prescription for dealing with diverse, intercon-
ting international commerce and threatening nected, value-laden policy issues.7 We will deal
public health regimes. While in general the end with two arguments in order, the first being a
of the Cold War has brought a shift from com- rejection of the positivist orientation, the
munistic to democratic thought, the realization second being more constructive (dare we say
of these benefits is challenged by issues of ethnic ‘positive’?) in nature.
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 40


Regarding the first assessment, Fischer The policy researcher must recognize that
(1998: 143; also see Fischer 2003) provides a not all empirical research is to be discarded in
model postpositivist rationale when he writes the postpositivist dustbin of policy analysis. An
that ‘Postpositivism … not only offer[s] a immense corpus of analytic research over the
theory of the social sciences that is identifiable past half-century, as well as highly skilled
readily in our existing practices, it also consti- analysts, have added greatly to our knowledge;
tutes an incorporation of new methods and in many cases, we should be loathe to relin-
approaches rather than a simple rejection of quish these contributions and to deny them,
old ones.’ While empirical or behavioral in Lynn’s (1999) words, a ‘place at the table.’
approaches which have been lumped under a Edward Lawlor (1996) poses the hard ques-
positivist label identify causal relationships as a tions that policy scientists need to ask them-
means of deriving an aggregated, predictive selves before professionally signing on the any
relationship, postpositivist approaches identify new research orientation: ‘What separates the
causal mechanisms as a means understanding a analyst from the journalist or consumer advo-
relationship (DeLeon 1998). Ann Chin Lin cate under this new argumentative turn? What
suggests that ‘[I]nterpretive work reconstructs separates the policy analyst from literary theo-
categories that are organic to the context it rist and critic in the case of narrative policy
studies, and thus is much less likely to be led analysis?’ He continues:
astray by preconceived notions that stem from To disconnect policy analysts from their disciplinary
inappropriate generalizations’ (Lin 1998: 164). roots and charge them with the general communicative
Other authors – Frank Fischer (2003), John functions espoused by the new argumentative school
Dryzek (1990 and 2000), Ronald Brunner would not only remove ‘tools’ as a defining feature of
the field, it would further undermine the already shaky
(1991), Maarten Hajer (with Wagenaar 2003) –
intellectual identity of the field. Postpostivism and
are more strident in their criticisms and have so-called postmodernism in policy analysis is a swamp
identified what they describe as serious episte- of ambiguity, relativism, and self-doubt … creating
mological shortcomings of the positivist more problems for the policy analysis business than it
approach, assumptions, and results, offering solves (Lawlor 1996: 120).
historical examples that attest to its deficien- Thus, without abandoning the positivist tools
cies. Dryzek (1990: 4–6) has been particularly that are appropriate for specific situations –
outspoken in his assessments of positivism, remember, context always counts (see, for
especially of what he (and others) call ‘instru- example, DeLeon 1998) – let us explore some
mental rationality,’ which he claims, research tools that move the policy sciences
destroys the more congenial, spontaneous, egalitarian, forward in the postpositivist direction.
and intrinsically meaningful aspects of human associa-
tion … represses individuals … is ineffective when con-
fronted with complex social problems … makes Social Network Theory, Participatory
effective and appropriate policy analysis impossible … Policy and Q-Methodology
[and, most critically] is antidemocratic.

In a more constructive sense, a postpositivist In response to the shortcomings of complexity,

approach, particularly in terms of ‘policy dis- multidisciplinary nature of issues, social net-
course’ (see Hajer 1993), provides a more com- work theory offers a conceptual approach
plete understanding of values and relationships for understanding interconnectivity among
(among both organizations and personal per- various actors. In many ways, networks have
ceptions). As we will argue below, the postposi- succeeded governmental units as an appropri-
tivist orientation is more conductive to the ate unit of analysis, because, for most, any con-
democratic strain in policy research. Just as temporary policy issue – be it criminal justice,
important, policy discourse is more conducive to human rights, education, or health care –
understanding the policy process results because specific policy problems are attended by social
research findings ‘only have meaning if they have congeries or a network of concerned actors
a theory attached to them’ (Coleman 1991: 432). representing the public, private, and nonprofit
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 41


sectors, working (hopefully but not always) attention to the structural context and content
cooperatively towards a consensus resolution. of exchange, in favor of research on transac-
As such, networks reflect interconnectedness tions. Previous research shows a separation of
among actors, issues, or groups and provide the connection of networks and the process
a mechanism to conceptualize the complex and outcome variables (Scharpf 1990: 161).
relationships among these societal elements Atkinson and Coleman (1992: 160) pose the
(Coleman and Skogstad 1990; Heclo 1978; question: ‘Are there relationships of power and
Rhodes 1990). Castells (1996: 468) summa- dependency that transcend and color individ-
rizes: ‘Networks constitute the new social mor- ual transactions?’ Answers to this and similar
phology of our societies and the diffusion of questions would go far to build a predictive
network logic substantially modifies the oper- theory of policy outcomes.
ation and outcomes in processes of produc- As all theories are refined over time, network
tion, experience, power, and culture.’ theory will continue to be developed. A future
As with other approaches, social network research agenda would address some outstand-
analysis imposes a semblance of order on a ing issues. The first is how policy networks
chaotic reality. It provides form and identity to change and how that change affects the policy
relationships and analysis, yet recognizes the outcomes, whereby research would address the
dynamic nature of boundaries. Networks variables of boundary shifts and inclusion and
strive to address who will participate in which exclusion (Atkinson and Coleman 1992). In this
event. That is, whereas other policy analysis vein, a similar agendum considers the connec-
approaches tend to focus on the hierarchical tions between policy communities and policy
processes that have characterized the process networks, where the former represent a variety
in the past, a network approach examines the of actors and potential actors who are interested
policy process in terms of the horizontal rela- in the policy issue and share interest and beliefs,
tionships that increasingly describe policy though not necessarily concordant, about the
issues. As Heclo (1978: 104) notes, ‘it is policy solution. Policy networks are a subset of
through networks of people who regard each policy communities.8 They are formed on the
other as knowledgeable, or at least needing to basis of exchange, of primarily information and
be answered, that public policy issues tend to resources, and influence and represent the body
be refined, evidence debated, and alternative of actors that interacts regularly (Fischer 2003).
options worked out – though rarely in any Howlett and Rayner (1995) hypothesize that
controlled, well-organized way.’ policy change occurs most readily when the
To address Sabatier’s (1999) criticisms that policy community and networks are unified.
the stages associated with the policy sciences A second area of study is to understand the
are disjointed, networks provide a more fluid ways in which network and community actors
view of the policy process and the contingent develop, ascribe, and share meaning of back-
actors, as well as addressing the complexity ground assumptions, ideas towards scientific
created by conditions of reciprocal interdepen- knowledge, and their role of involvement
dence (Atkinson and Coleman 1992; Scharpf (Fischer 2003). This area is premised on an
1990). Hajer and Wagenaar (2003: 13) place interpretive community, where knowledge is
network analysis at a key juncture of future created not just by the relationships among
policy research when they ask ‘what kind of selected data and variables, but by the inter-
policy analysis might be relevant to under- pretation and situational context in which
standing governance in the emerging network those variables are applied (Innes 1998).
society.’ The third deals with the impact of political
While the identification of network actors ideas and theories about policies. Particularly,
and relationships is, in and of itself, important, attention must shift to ‘the dominant values
its real value to the policy sciences is the iden- guiding public policy, the knowledge base
tification of the content (if not the intensity) of available to policymakers, and the norms that
relationships. Policy research has paid less legitimize various approaches to policy’
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 42


(Atkinson and Coleman 1992: 174). Norms and identified and meaning ascribed to the artifacts
values may change over time or differ per actor. is discerned for different members of the
As such, ‘analysts must seek to ascertain the community so that points of conflict, reflect-
more general principles and norms underlying ing different interpretations, can be identified
interpretation of the policy field’ (Atkinson (Fischer 2003; also see Stone 1998). In this way,
and Coleman 1992: 175). stakeholders help analysts understand the con-
As a conceptual and methodological tool, text of analysis (Durning 1999), in theory, on a
networks offer promise. Interestingly enough, more equitable, discursive basis, or what
the injection needed to remedy and develop Habermas (1983) refers to as an ‘ideal speech
this approach is consistent with a postpositivist situation.’ In a complementary basis, Emery
prescription. The value of postpositivism over Roe (1994) has indicated that ‘narrative’ policy
positivistic approaches is the ability to discern analyses can examine competing narratives
context, power, and the answer to ‘why’. Thus, and help frame problems differently, especially
structural interpretations, change processes, in the context of a highly uncertain, charged,
and norms and values of networks can possi- or power-laden issue. The methodology places
bly be handled with postpositivist approaches. an emphasis on learning from ambiguity,
Within the policy tool kit should be mecha- but one needs to carefully extrapolate to other
nisms to enhance participatory policy analysis policy situations (Schram 1995).
(DeLeon 1997), discourse analysis (Torgerson Still, one needs to caution that deliberative
2003; Hajer 1993), and policy learning. Fischer democracy must be treated with the same
(1998: 143) posed the issue directly: ‘Holding skepticism that underlies all policy methodol-
out the possibility of redeeming or realizing a ogy (see Lynn 1999), especially from its propo-
policy science of democracy, [postpositivism] nents. Certainly greater public participation
calls for participatory institutions and prac- will threaten the extant bureaucracies and their
tices that open spaces for citizen deliberation proponents, but the democratic promises of
on contextual assumptions, empirical out- the initiative and referenda have been sorely
comes, and the social meaning of conclusions’ undercut by highly organized political groups
(also see Fischer 2003). But just how one oper- (Broder 2000; Ellis 2002). Torgerson (2003:
ationally reaches the saddle point between par- 119) cautions: ‘The institutionalization of a
ticipation and governing has been a question discursive design … has the potential to influ-
for eternities. Still, participatory policy at least ence the power context from which it emerges
explicitly addresses the issue. Indeed, Torgerson and may … be opposed because of this. Both
(2003) makes the argument that increased the feasibility of discursive designs and their
efforts in participatory policy analysis would co-optive tendencies thus need to be consid-
narrow the perceived gap between politics and ered in terms of the power relationships found
policy, since both parties will be able to address in a particular setting.’ Moreover, few would
their preferences more openly and delibera- advocate for a policy discourse in which sheer
tively. A participative approach has the poten- volume and empty rhetoric outweighed delib-
tial to provide better information that may eration (DeLeon 1997), but the prevalence of
ultimately open the door to broader and more that behavior is seen daily in the public media
appropriate policy solutions. Thus, a parti- and its effects on the body politic.
cipatory discourse or inquiry allows for Of the postpositivist tools, Q-methodology
decreased conflict, increased trust, multiple is consistent with narrative analysis that gets
viewpoints, and normative interpretations, beyond the polarized framing of issues by
and portends towards a greater democratic reframing policy issues in ways not initially
underpinning, along the lines of a ‘one person, perceived, largely by subverting the assump-
one vote’ direct democracy (Barber 1984). tions of objectivism. The method provides
A participatory approach opens the door to information on public values and positions,
interpretive policy analysis, whereby the artifacts providing policy makers with a broader range
that carry meaning to the policy community are of policy options (Durning 1999; van Eeten
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 43


2001). ‘Q-methodology is useful in the The history of the policy sciences naturally
prescription context because it can allow a affects the present day analyst/client relation-
public dialogue to take place regarding values ships. We can further say that a continuation of
and can then contribute to the stabilization of the status quo will not be to the benefit of
expectations needed to achieve prescription either party. The ‘promise’ then of the policy
outcomes’ (Steelman and Maguire 1999: 365). sciences is that, to gain in stature and accep-
Q-methodology identifies patterns of subjec- tance, some important new directions must be
tive perspectives across individuals. It groups grafted to the policy sciences’ approach.
individuals with like views to discern how that Moreover, most observers can agree that the
subgroup perceives an issue (Durning 1999). policy sciences cannot surrender their dedica-
Important to its application is to conceptu- tion to a democratic ethos. As Hajer and
alize the relationship between the policy Wagenaar (2003: 15) state, ‘Whatever we have
analyst and the decisionmaker – as consulta- to say about the nature and foundation of
tive or participative – and the stage of the the policy sciences, its litmus test will be that
policy process for which information is gath- it must ‘work’ for the everyday reality of mod-
ered (Steelman and Maguire 1999; Durning ern democracy’ (also see Torgerson 2003).
1999; Roberts 1995; Lasswell 1971). The tech- Given the three defining characteristics of
nique offers policymakers a tool to increase the policy sciences – problem oriented, multi-
the knowledge base and understanding of a disciplinary, and normative – we propose that
situation’s context and meaning. The chal- these directions will more clearly articulate the
lenge to Q-methodology will be to aggregate value-orientation vector. It is important that
preferences to survive a collective process of the value component of policy decisions be
deliberation (Lynn 1999). understood and made more transparent for all
To these ends, we have suggested ways in
which this avenue can be accommodated most
readily, through the use of a more ecumenical
postpositivist approach, a more participatory
The policy sciences are unquestionably an set of guidelines, the application of social net-
approach in transition, but it is not clear what work analysis, and perhaps a greater use of
the end point will resemble. The past has been tools such as Q-methodology for specific pur-
promising in many ways, but the policy poses. The common denominator among these
sciences have not achieved the prominence in approaches is that all encourage a less reduc-
policymaking circles that its early proponents tionist, more democratic voice in the policy
might have wished; in addition, at times, the process. These will not come particularly easy,
means to these ends have been problematic. as they require skill sets somewhat different
Part of this was due to the policy-defining con- from those currently practiced. For instance,
ditions that have proven to be remarkably participatory policy analysis requires a certain
complex and analytically unwieldy, a situation adroitness at group processes and mediation.
only worsened by the wholesale movement of We would not want to suggest that these are
the policy analysis community to economics the only new avenues for the policy sciences to
and behavioralism, or what we have called here consider; surely there is a world of alternatives
‘positivism.’ But part of this dyspepsia is that to ponder. However, we do wish to make the
the policy sciences’ offerings were consistently point that, if the policy sciences are to legiti-
not timely nor resonant with the requirement mately aspire to their original claims, featuring
of the policymaking community. Thus, neither ‘knowledge of and knowledge in’ government,
the policy advising nor the policymaking then the ‘same old same old’ will only lead to the
communities have been satisfied with the marginalization of the policy sciences, a fate
applications of the policy sciences, hardly an Lasswell and succeeding generations of policy
encouraging condition. scientists and policymakers can only regret.
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 44


NOTES Barber, Benjamin (1984). Strong Democracy:

Participatory Politics for a New Age. Berkeley:
University of California Press.
1. The attribution is not universal. Beryl Radin (2000)
Bobrow, Davis B., and John Dryzek S. (1987). Policy
traces the development of policy analysis with Yehezkel Dror
(see Dror 1971) as the principal early contributor to the field.
Analysis by Design. Pittsburgh, PA: University of
2. An anecdote to this effect. In 1707, a fleet of British Pittsburgh Press.
ships of the line was returning to England; during a par- Brewer, Garry D. (1974). ‘The Policy Sciences
ticularly foggy night, a seaman, who knew the area, sug- Emerge: To Nurture and Structure a Discipline.’
gested to the British admiral that the fleet was approaching Policy Sciences. Vol. 5, No. 3. Pp. 239–244.
dangerous shoals. The seaman was immediately hung for ——, and Peter DeLeon (1983). The Foundations
his impertinence; shortly thereafter, the fleet was destroyed of Policy Analysis. Belmont, CA: Wadsworth
when it ran aground of the shoals and thousands of lives Publishing.
were lost (Sobel 1995). Broder, David S. (2000). Democracy Derailed: The
3. Lasswell and Kaplan (1950, xii and xxiv) dedicate the
Initiative Movement and Power of Money. New
policy sciences to provide the ‘intelligence pertinent to the
integration of values realized by and embodied in inter-
York: Harcourt.
personal relations,’ which ‘prizes not the glory of a deper- Brunner, Ronald D. (1991). ‘The Policy Movement
sonalized state of the efficiency of a social mechanism, but as a Policy Problem.’ Policy Sciences. Vol. 24, No. 1
human dignity and the realization of human capabilities.’ (February). Pp. 295–331.
4. A moment should be set aside to distinguish ‘policy Castells, Manuel (1996). The Rise of the Network
analysis’ from the ‘policy sciences.’ Many (e.g., Radin 2000, Society. Oxford: Blackwell.
Dunn 1981, and Heineman et al. 2002) reference the former. Clarke, Richard A. (2004). Against All Enemies.
DeLeon (1988: 9) indicated that ‘Policy analysis is the most New York: The Free Press.
noted derivative and application of the tools and method- Clinton, Bill (2004). My Life. New York: Alfred
ologies of the policy sciences’ approach.’ For the purposes of
A. Knopf.
this essay, the terms are largely interchangeable.
5. Lasswell’s original stages are intelligence, promotion,
Coleman, David A. (1991). ‘Policy Research—Who
prescription, invocation, application, termination, and Needs It?’ Governance: An International Journal of
appraisal (1956). Policy and Administration.Vol. 4, No. 4. Pp. 420–455.
6. Nor is this to suggest that there have not been other Coleman, William D. and Grace Skogstad (1990).
significant political events, such as the complete collapse of ‘Policy Communities and Policy Networks: A
Communism in 1989, the 1996 Welfare Reform Act, or the Structural Approach.’ In William D. Coleman and
globalization/telecommunication phenomena. Grace Skogstad (eds.), Policy Communities and
7. Fischer (2003), among others, has used the terms ‘post- Public Policy in Canada. Toronto: Copp Clark
positivism,’ ‘postmodernism,’ and ‘postempiricism’ as syn- Pitman.
onyms. In this essay, we will use the first as a collective term.
Commoner, Barry (1979). The Politics of Energy.
8. For more work on policy subsystems, see Milward
and Walmsley (1984).
New York: Alfred A. Knopf.
Dahl, Robert A. (1970/1999). After the Revolution.
New Haven, CT: Yale University Press.
REFERENCES DeLeon, Peter (1988). Advice and Consent: The
Development of the Policy Sciences. New York:
Russell Sage Foundation.
Amy, Douglas J. (1984). ‘Why Policy Analysis —— (1997). Democracy and the Policy Sciences.
and Ethics are Incompatible.’ Journal of Policy Albany NY: SUNY Press.
Analysis and Management. Vol. 3, No. 4 (Summer). —— (1998). ‘Models of Policy Discourse: Insights
Pp. 573–591. vs. Prediction.’ Policy Studies Journal. Vol. 26, No. 1
Anderson, James E. (1975/1979). Public Policy (Spring). Pp. 147–161.
Making. New York: Holt, Rinehard, and Winston. —— (1999). ‘The Stages Approach to the Policy
2nd edition. Process: What Has it Done? Where Is It Going?’ In
Atkinson, Michael M. and William D. Coleman Paul A. Sabatier (ed.), Theories of the Policy
(1992). ‘Policy Networks, Policy Communities and Process. Boulder, CO: Westview Press.
the Problems of Governance.’ Governance: An Derthick, Martha (1972). New Towns In-Town.
International Journal of Policy and Administration. Washington DC: The Urban Institute.
Vol. 5, No. 2. Pp. 154–180. Dionne, E.J. (1991). Why Americans Hate Politics.
Baker, Peter (2001). The Breach: Inside the New York: Simon & Schuster.
Impeachment and Trial of William Jefferson Dunn, William N. (1981). Public Policy Analysis.
Clinton. New York: Berkeley Books. Englewood Cliffs, NJ: Prentice-Hall.
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 45


Durning, Dan (1999). ‘The transition from tradi- Hajer, Maarten A., and Hendrik Wagenaar (2003).
tional to postpositivist policy analysis: A role for ‘Introduction.’ In Maarten A. Hajer and Hendrik
Q-methodology.’ Journal of Policy Analysis and Wagenaar (eds.), Deliberative Policy Analysis:
Management. Vol. 18, No. 33. Pp. 389–410. Understanding Governance in the Network Society.
Drew, Elizabeth (1994). On the Edge: The Clinton Cambridge, NY: Cambridge University Press.
Presidency. New York: Simon & Schuster/ Pp. 1–33.
Touchstone. Hajer, Maarten A., and Hendrik Wagenaar (eds.)
Dror, Yezekhel (1971). Design for the Policy Sciences. (2003) Deliberative Policy Analysis: Understanding
New York: American Elsevier. Governance in the Network Society. Cambridge,
Dryzek, John S. (1990). Discursive Democracy: NY: Cambridge University Press.
Politics, Policy, and Political Science. Cambridge, Harrington, Michael (1960). The Other America:
UK: Cambridge University Press. Poverty in the United States. New York: Penguin.
—— (2000). Deliberative Democracy and Beyond. Heclo, Hugh (1978). ‘Issue Networks and the
Oxford, UK: Oxford University Press. Executive Establishment.’ In The American
Ellis, Richard J. (2002). Democratic Delusions: The Political System, Anthony King (ed.). Washington:
Initiative Process in America. Lawrence: University American Enterprise Institute.
of Kansas Press. Heineman, Robert A., William T. Bluhm, Steven A.
Fischer, Frank (1993). ‘Policy Discourse and the Peterson, and Edward N. Kearny (2002). The World
Politics of Washington Think Tanks.’ In Frank of the Policy Analyst. Chatham, NJ: Chatham
Fischer and John Forester (eds.), The Argumentative House Publishers. Third Edition.
Turn in Policy Analysis and Planning. Durham, NC: Howlett, Michael and Jeremy Rayner (1995). ‘Do
Duke University Press. Pp. 21–42. ideas matter? Policy network configurations and
—— (1998). ‘Beyond Empiricism: Policy Inquiry in resistance to policy change in the Canadian forest
Postpositivist Perspective.’ Policy Sciences Journal. sector.’ Canadian Public Administration. Vol. 38,
Vol. 26, No. 1 (Spring). Pp. 129–146. No. 3. Pp. 382–410.
—— (2003). Reframing Public Policy. Oxford: Huntington, Samuel P. (1996). The Clash of Civiliza-
Oxford University Press. tions and the Remaking of World Order. New York:
—— and John Forester (eds.) (1993). The Argumen- Simon & Schuster.
tative Turn in Policy Analysis and Planning. Innes, Judith J. (1998). ‘Information on Communica-
Durham, NC: Duke University Press. tive Planning.’ Journal of the American Planning
FitzGerald, Frances (1972). Fire in the Lake. Boston: Association, Vol. 64, No. 1. Pp. 52–63.
Little, Brown. Janis, Irving (1983). GroupThink. Boston: Houghton
Forester, John (1999). The Deliberative Practitioner: Mifflin. 2nd Edition.
Encouraging Participative Planning Processes. Jones, Charles (1970/1984). An Introduction to the
Cambridge, MA: The MIT Press. Study of Public Policy. Belmont, CA: Wadsworth.
Gelb, Leslie, with Richard K. Betts (1979). The Irony 3rd Edition.
of Vietnam: The System Worked. Washington DC: Jenkins-Smith, Hank C., and Paul A. Sabatier
The Brookings Institution. (1993). ‘The Study of the Policy Process.’ In Paul
Goldhamer, Herbert (1978). The Adviser. New York: A. Sabatier and Hank C. Jenkins-Smith (eds.),
American Elsevier. Policy Change and Learning. Boulder, CO:
Gray, Colin (1971). ‘What Has Rand Wrought?’ Westview Press. Chap. 1.
Foreign Policy. No. 4 (Fall). Pp. 111–129. Johnson, Hayes (2001). The Best of Times: America
Greenberger, Martin, Garry D. Brewer, and Thomas in the Clinton Years. New York: Harcourt.
Schelling (1984). Caught Unawares: The Energy Kirp, David L. (1991). ‘The End of Policy Analysis:
Decade in Retrospect. Cambridge, MA: Ballinger. With Apologies to Daniel (The End of Ideology)
Habermas, Jürgen (1983). The Theory of Communica- and Frank (The End of History) Fukuyama.’
tive Action: Reasons and the Rationalization of Journal of Policy Analysis and Management. Vol. 11,
Society. Thomas McCarthy (Trans.). Boston: Beacon No. 4 (Fall). Pp. 693–696.
Press. Landsberg, Hans (1979). Energy: The Next Twenty
Hajer, Maarten A. (1993). ‘Discourse Coalitions and Years. Cambridge: Ballinger for Resources for the
the Institutionalization of Practice: The Case of Future.
Acid Rain in Great Britain.’ In Frank Fischer and Lasch, Christopher (1995). The Revolt of the Elites and
John Forester (eds.), The Argumentative Turn in the Betrayal of Democracy. New York: W.W. Norton.
Policy Analysis and Planning. Durham, NC: Duke Lasswell, Harold D. (1951a). ‘The Policy Orienta-
University Press. Pp. 43–76. tion.’ In Daniel Lerner and Harold D. Lasswell
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 46


(eds.), The Policy Sciences. Palo Alto, CA: Stanford Radin, Beryl A. (2000). Beyond Machiavelli: Policy
University Press. Chap. 1. Analysis Comes of Age. Washington, DC: Georgetown
—— (1951b). The World Revolution of Our Time: A University Press.
Framework for Basic Policy Research. Palo Alto, Rhodes, R.A.W. (1990).’Policy Networks: A British
CA: Stanford University Press. Reprinted in Perspective.’ Journal of Theoretical Politics. Vol. 2.
Harold D. Lasswell and Daniel Lerner (eds.), Pp. 293–317.
World Revolutionary Elites. Cambridge, MA: The Ricci, David M. (1984). The Transformation of
MIT Press 1965. Chap. 2. American Politics: The New Washington and
—— (1956). The Decision Process. College Park: the Rise of Think Tanks. New Haven, CT: Yale
University of Maryland Press. University Press.
—— (1971). A Pre-View of Policy Sciences. New Rivlin, Alice M. (1970). Systematic Thinking for
York: American Elsevier. Social Action. Washington DC: The Brookings
—— and Abraham Kaplan (1950). Power and Institution.
Society. New Haven, CT: Yale University Press. Roberts, R. (1995). ‘Public Involvement: From
Lawlor, Edward F. (1996). ‘Book Review.’ Journal of Consultation to Participation.’ In F. Vanclay and
Policy Analysis and Management. Vol. 15, No. 1 D.A. Bronstein (eds.). Environmental and Social
(Winter). Pp. 110–121. Impact Assessment. New York: John Wiley & Sons.
Levine, Robert A. (1970). The Poor Ye Need Not Roe, Emery (1994). Narrative Policy Analysis.
Have With You: Lessons from the War on Poverty. Durham, NC: Duke University Press.
Cambridge, MA: The MIT Press. Sabatier, Paul A. (1993). ‘Policy Change over a
Lin, Ann Chih (1998). ‘Bridging Positivist and Decade or More.’ In Paul A. Sabatier and Hank C.
Interpretivist Approaches to Qualitative Methods.’ Jenkins-Smith (eds.), Policy Change and Learning.
Policy Studies Journal. Vol. 26, No. 1 (Spring). Boulder, CO: Westview Press. Chap. 2.
Pp. 162–180. —— (ed.) (1999). Theories of the Policy Process.
Lindblom, Charles E. and David K. Cohen, (1979). Boulder, CO: Westview Press.
Usable Knowledge: Social Science and Social Problem ——, and Hank C. Jenkins-Smith (eds.) (1993).
Solving. New Haven: Yale University Press. Policy Change and Learning. Boulder, CO:
Lynn, Laurence E., Jr. (1999). ‘A Place at the Table: Westview Press.
Policy Analysis, Its Postpositive Critics, and the Scharpf, Fritz (1990). ‘Games Real Actors Could
Future of Practice.’ Journal of Policy Analysis Play: The Problem of Connectedness.’ Köln:
and Management. Vol. 16, No. 3 (Summer). Max-Planck-Institute für Gesellschaftsforschung,
Pp. 411–425. Paper 90/8.
Milward, H. Brinton and Gary L. Walmsley (1984). Schick, Allen (1973).‘A Death in the Bureaucracy: The
‘Policy Subsystems, Networks and the Tools of Demise of Federal PPB.’ Public Administration
Public Management.’ In R. Eyestone (ed.), Public Review. Vol. 33, No. 2 (March/April). Pp. 146–156.
Policy Formation. Greenwich, CT: JAI Press. Schneider, Anne Larason and Helen Ingram (1997).
Moynihan, Daniel P. (1969). Maximum Feasible Policy Design for Democracy. Lawrence, KS:
Misunderstanding: Community Action in the War University of Kansas.
on Poverty. New York: The Free Press. Schön, Donald and Martin Rein (1994). Frame
Murray, Charles (1984). Losing Ground. New York: Reflection. New York, NY: Basic Books.
BasicBooks. Schram, Stanford F. (1995). ‘Against Policy Analysis:
Nakaruma, Robert (1987).‘The Textbook Polic Process Critical Reason and Poststructural Resistance.’ Policy
and Implementation Research.’ Policy Studies Sciences. Vol. 28, No. 4 (November). Pp. 375–384.
Research. Vol 7, No. 2 (Autumn). Pp. 142–154. Sckopal, Theda (1997). Boomerang: Health Care
Nye, Joseph S., Jr., Philip D. Zelikow, and David C. Reform and the Turn Against Government. New
King (1997) (eds.). Why People Don’t Trust Govern- York: W.W. Norton.
ment. Cambridge, MA: Harvard University Press. Sheehan, Neil (1971). The Pentagon Papers. New
Olson, Keith W. (2003). Watergate: The Presidential York: Bantam and the New York Times.
Scandal that Shook America. Lawrence, KS: The Smith, Bruce L.R. (1966). The RAND Corporation.
University Press of Kansas. Cambridge, MA: Harvard University Press.
Polsby, Nelson W. (1984). Political Innovation in Sobel, Dava (1995). Longitude. New York: Walker
America. New Haven: Yale University Press. and Co.
Pressman, Jeffrey, and Aaron Wildavsky (1984). Steelman, Toddi A. and Lynn A. Maguire (1999).
Implementation, Berkeley: University of California ‘Understanding Participant Perspectives: Q-
Press. 3rd Edition. Methodology in National Forest Management.’
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 47


Journal of Policy Analysis and Management. Weiss, Carol H. (1980). ‘Knowledge creep and
Vol. 16, No. 3 (Summer). Pp. 361–388. decision accretion.’ Knowledge. Vol. 1, No. 3.
Stone, Deborah (1998). Policy Paradox. New York: Pp. 341–404.
Norton. Wildavsky, Aaron (1979). Speaking Truth to Power.
Torgerson, Douglas (2003). ‘Democracy through Boston: Little, Brown.
Policy Discourse.’ In Maarten A. Hajer and Hendrik —— (1984). The Politics of the Budgetary Process.
Wagenaar (eds.), Deliberative Policy Analysis: Boston: Little Brown.
Understanding Governance in the Network Society. ——, and Ellen Tanenbaum (1981). The Politics of
Cambridge: Cambridge University Press. Chap 4. Mistrust. Beverly Hills: Sage.
van Eeten, Michel J. G. (2001). ‘Recasting Intractable Williams, Walter (1998). Honest Numbers and Demo-
Policy Issues: The Wider Implications of The cracy. Washington DC: Georgetown University
Netherlands Civil Aviation Controversy.’ Journal Press.
of Policy Analysis and Management, Vol. 20, No. 3. Woodward, Bob (1994). The Agenda. New York:
Pp. 391–414. Simon & Schuster.
Weyant, John P. (1980). ‘Quantitative Models in —— (2004). Plan of Attack. New York: Simon &
Energy Policy.’ Policy Analysis. Vol. 6, No. 2 (Spring). Schuster.
Pp. 221–234.
Peter-3383-Chapter-02.qxd 6/8/2006 7:31 PM Page 48
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 49

Behavioral Rationality and the Policy
Processes: Toward A New Model of
Organizational Information Processing

B R YA N D . J O N E S , G R A E M E B O U S H E Y , A N D

INTRODUCTION deductive power. Bounded rationality insists

that any model of choice be based in scientific
Behind any study of public policy processes is analysis of the cognitive architecture of
a theory of organizations. Policy is made by humans, even if that scientific analysis implies
organizations, but organizations are made up less parsimony.
of interacting human decision-makers. As a In the second part, we survey recent empiri-
consequence, any theory of organizations har- cal assessments of behavioral rationality—how
bors a theory of individual choice. people actually behave in experimental and
This chapter considers the analysis of orga- observational situations where comprehensive
nizations from the competing perspectives of rationality makes precise predictions about
rational choice theory and bounded rational- outcomes. Findings across disciplines of deci-
ity. Our essay is divided into four parts. The sion making not only point to the theoretical
first section surveys the broad characteristics inadequacies of the rational choice framework,
of these behavioral models in the study of but also provide the important tools for con-
public policy. Although both begin with a structing more realistic models of human
common goal of connecting individual choice choice.
to macropolitical outcomes, they disagree fun- The third part of this essay examines policy
damentally over how individual behavior implementation and administration in light of
should be understood. Rational choice theory studies of principal-agent dilemmas in organiza-
claims analytical rigor and theoretical parsi- tions to highlight key differences between ratio-
mony, holding that individuals behave as if nal choice theory and bounded rationality.
they were pure utility maximizers to deduce While rational choice theory has evolved to
patterns of outputs from social systems provide a more accurate portrayal of human
(Friedman 1953). There is no scientific analy- behavior and preferences (Ostrom 1999a;
sis to support this assumption; it is simply an Ostrom 1999b; Levi 1997), research in this para-
assumption, presumably validated by its digm nonetheless continues to overemphasize
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 50


the problems of delegation and control in evolved from the efforts of such diverse
bureaucracy. The obsession with control is theorists as Herbert Simon, Mancur Olson,
symptomatic of the larger pathologies of ratio- and Anthony Downs to link individual human
nal choice theory – the core assumptions of indi- decision-making with broader macropolitical
vidual utility maximization simply do not outcomes (Simon 1947; Olson 1982; Downs
capture the complexities of human decision- 1957). Although the competing models of
making. By focusing narrowly on questions gen- rationality disagreed about the fundamental
erated from the control trap, rational choice motivations behind individual choice, they
theory ignores a wide range of intriguing ques- were unified in their belief that the processes
tions about information processing and policy and policy outcomes be most powerfully
dynamics in bureaucratic decision-making. understood through exploring the role of indi-
In the final part of this essay, we point toward vidual behavior in collective decision-making.
an information-processing model of the policy A theory of rationality which anticipated indi-
process that avoids the obsession with control that vidual choices in the context of the larger
permeates rational theories of policymaking. We political process would present a significant
argue that it is both more scientifically sound and advancement in the study of government –
fundamentally more parsimonious than models such a set of assumptions would act as a theo-
based on rational choice. retical tool-box that could be used, not only to
cut through the complexity of public decision-
making, but also to locate areas of conflict
within organizations, and potentially predict
future political outcomes. Both rational choice
theory and bounded rationality pushed the
study of public policy away from case studies
and atheoretical descriptions of public admin-
In the behavioral sciences, such as psychology istration toward a generalized theory of public
and behavioral biology, the major topic of policy.
inquiry is individual behavior. In the social Despite these common aims, the theories of
sciences, such as political science and econom- rationality are deeply divided over the most
ics, the aim is to understand social systems of basic assumptions of individual choice.
interacting individuals. Social scientists need Rational choice theory borrows heavily from
both a model of individual behavior and an economic assumptions of individual prefer-
understanding of how individuals interact ences, and believes that a sufficient behavioral
to produce social outcomes, such as those model could be drawn from deductions of an
produced by free markets or policymaking individual’s self interested utility maximiza-
processes in government. As a consequence, tion. Bounded rationality began as a critique of
social scientists cannot dally with the nuances comprehensive rationality, and grew from an
of human behavior that do not impact on the effort to reconcile the reductionist economic
interactions of people in organizations. Any assumptions of rational choice with observed
model of individual choice in policymaking psychological constraints on human decision-
processes must be parsimonious, and they making. The following sections outline the dis-
must link individual level actions to policy- tinguishing characteristics of both rational
making outcomes. choice theory and bounded rationality as
The behavioral models of comprehensive applied to the study of public policy. This cri-
rationality and bounded rationality, which today tique intends not only to cast light on the
divide the theories of public policy, were driven major differences between the two theories,
by a common desire to improve the rigor of but also calls attention to exciting empirical
political analysis. Applications of both rational findings which we believe are fueling increas-
choice theory and bounded rationality in ing theoretical convergence around a single
the study of public policy and administration positive theory of human choice.
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 51


COMPREHENSIVE RATIONALITY AND scholars reduce it to a question of self-interest.

THE THEORY OF POLICY PROCESSES Their behaviors reflect only their effort to max-
imize the utility of their choice.
In recent decades rational choice theory has Even the staunchest proponents of rational
been widely applied to the study of govern- choice theory regard this characterization of
ment. This dominance extends in no small part the model as more of an ideal type than a real-
from the model’s theoretical parsimony. istic portrayal of human behavior. Decisions
Rational choice theorists have explored some made under conditions of complete certainty –
of the most complex aspects of politics by when specific strategic choices are known to
making relatively few core assumptions about lead to explicit outcomes – are rarely found in
individual behavior. As in neoclassical eco- political or economic life. More powerful
nomic models, the theory of comprehensive insights have evolved from rational choice
rationality posits that political decision- theory when theorists have examined decision-
makers are self-interested utility maximizers making under risk – where a strategy might
who hold stable preferences and objectives, lead to several different outcomes with known
and make strategic decisions to maximize the probability, or conditions of uncertainty –
personal benefits of a given choice. To under- where outcomes are known but the probabili-
stand politics at the aggregate level, researchers ties associated with those outcomes are not,
need only to understand ordered preferences and must be estimated.
of individual actors who populate a specific Rational choice theorists use expected utility
institution or political sphere and the formal theory to approximate how individuals make
rules by which these fixed preferences are com- calculations that rank alternatives by their
bined. In this approach, preferences + rules = expected value (under risk or uncertainty)
policy outcomes. rather than their known value. Under condi-
Important distinctions can be drawn tions of risk, individuals form strategic prefer-
between rational choice models in public policy, ences probabilistically – they compare the
political science, and economics; however, all probability that their most preferred outcome
rational choice models share common charac- will occur against the probability that their less
teristics. First, decision-makers hold stable preferred outcome will occur, and both against
ranked and ordered preferences for outcomes. the cost of making a decision.
Given three possible alternatives – options A, B, It is not enough that individuals calculate
and C, a rational chooser will form clear prefer- the probabilities that outcomes will occur if
ences between each of the three given alterna- they take a particular action; they must also
tives. These preferences are transitive, meaning calculate the likelihood that their choice will
that if an individual prefers option A to B, and yield the outcome against the probability that
prefers option B to C, he also prefers option A to their choice will have no bearing on the out-
C. Second, a decision-maker possesses necessary come. Individuals must not only predict the
information to connect choices to outcomes. probability that an event will occur, but also
With this information, individuals then opti- the chance that an event would occur without
mize when making decisions – they make strate- their participation, or that their less preferred
gic choices in order to achieve their most outcome would occur in spite of their partici-
preferred outcome. “Thick” rationality adds the pation. A classic example is rational voter
assumption of individual self-interest in deter- models, where people are predicted to vote
mining preferences. For example – preferences only if their expectation that their vote will
for a higher tax rate to fund public education make a difference exceeds the marginal cost of
over a lower overall tax rate suggest that an indi- voting.
vidual will receive greater personal utility for Comprehensive rationality holds great
increased funding in public education than appeal as a model of choice for three basic rea-
from lower taxes. Regardless of the underlying sons. First, rational choice promises a parsimo-
reasons for this preference, rational choice nious method for studying complex political
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 52


behavior. By assuming that actors are that is not the point. It was clearly and cleanly
singularly motivated to maximize gains avail- stated in a generalized manner that facilitated
able in political decisions, a wide range of con- testing against empirical observation.
siderations that have long complicated the study This leads to the third advantage. The ratio-
of political science – social class, partisanship, or nal choice method is a comparably rigorous
cultural values – become peripheral to a deci- approach to the study of political processes.
sion-maker’s preferences. The challenge of Rational choice researchers derive formal mathe-
rational choice theory is to characterize accu- matical models from a set of assumptions
rately the payoffs available to political actors in about individual preferences. These models
a given arena. A classic example of this approach derive a set of hypotheses of anticipating a spe-
can be drawn from David Mayhew’s study of cific outcome in politics, and then test them
congressional voting behavior in Congress: the against the self interested behavior of actors
Electoral Connection. In his analysis, democratic put forward in the theoretical assumptions of
representation is motivated singularly by a con- rational choice theory—in this regard, rational
gressman’s desire to retain political power choice hypotheses are verified or falsified. This
(Mayhew 1974). Ideology and policy-making approach forces researchers to advance well-
matter only insomuch as they instruct a politi- specified models of political choice. Rational
cian as to where his source of political power is choice models focus only on those most crucial
located, and how to make appropriate votes to elements of the political process that are neces-
retain electoral support. sary to explain outcomes.
The second advantage of the rational choice Both the simplicity and the promise of theo-
method is its broad theoretical generalizability. retical generalizablity have made rational
Unlike descriptive studies of political behavior, choice theory a popular tool for the study of
the baseline assumptions of rational choice government. The rational choice approach has
methods are characterized by a ‘universalism been applied to such subfields as organiza-
that reveals generalizable implications beyond tional behavior (Bendor and Moe 1985; Moe
those under immediate investigation’ (Levi 1984) congress (Mayhew 1974; Arnold 1990),
1997, 20). Deductions from rational choice parties and elections (Downs 1967) and collec-
theory are not specific to a particular time and tive action problems (Olson 1982). The
case, but should offer insights wherever similar breadth of this research is suggestive of the
conditions can be observed. Mancur Olson theoretical power of the model. By exploring
argues that ‘the persuasiveness of a theory the strategic behavior of self-interested individ-
depends not only on how many facts are uals, rational choice theorists have produced a
explained, but also how diverse are the kinds of rich and theoretically unified body of research
facts explained’. In Olson’s view, the rational in a discipline once marked by methodological
choice approach has the double benefit of pre- eclecticism.
serving the parsimony of a theory by removing
‘any inessential premises or complexities that
ought to be removed from an argument,’ while
retaining a high degree of explanatory power
(Olson 1982: 12–13). Drawing upon a relatively
small set of assumptions about human behav-
ior, Olson is able to advance a series of predic- Perhaps because rational choice theory takes a
tions that serve as the basis of a theory of how decidedly reductionist approach to the study
the formation of interest groups in democratic of government—one which is ‘willing to sac-
societies detract economic growth. That demo- rifice nuance for generalizability, detail for
cratic nations have sustained high levels of logic’ (Levi 1997: 21)—dissenting researchers
growth, even as interest groups have prolifer- have long charged that these sparse assump-
ated, means that Olson’s theory was wrong, but tions of individual utility maximization distort
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 53


the complexity of both individual behavior moving between individual decisions and
and organizational decision-making. Field organizational outcomes. Bounded rationality
researchers complain that observations of retains the hallmarks of a theoretical model—
individual behavior rarely match the calculat- it captures only those aspects of human
ing self-interested actor posited in rational behavior needed to understand collective
choice theory (Brehm and Gates 1997; Lipsky decision-making.
1980). The behavioral norm of individual util- As with so many researchers who followed
ity maximization simply does not seem to him, Simon noticed that the assumptions of
reflect accurately the actions of politicians, expected utility analysis failed to match his own
bureaucrats and voters, whose choices so often observations of real world economic decision-
seem to be motivated by risk aversion, sense making (Simon 1999). Looking at budgeting in
of mission, identity, fairness, or altruism. For Milwaukee, Simon observed that relatively little
these critics, rational choice theory is at best individual behavior matched ‘substantive or
idiosyncratic—applicable only to discrete objective rationality, that is, behavior that can be
institutions such as the US congress where adjudged to be optimally adapted to the situa-
political behavior can be safely matched to tion’ (Simon 1985, 294). He instead found that
self-interest (Rockman 2000). At worst rational the processes of both individual and organiza-
choice theory is misleading, as it reduces tional decision-making were a good deal
potentially interesting social behavior such messier than rational expectations would have
as altruism to an individual’s self-interest us believe. Bureaucratic budgets were often
(Monroe 1996). adjusted incrementally, using the prior year’s
Researchers who actually study individual budgeting as a benchmark for future spending
decision making find the rational choice needs (Thompson and Green 2001; Simon
approach curious. Cognitive psychologists find 1991). Organizational decisions were made
the economic assumptions of stable ordered more through horse trading and bargaining
preferences, transitivity, and utility maximiza- than a process even remotely resembling fully
tion to be a strange abstraction of human rational decision-making (Simon 1999). Orga-
decision-making. Findings in psychology indi- nizations and individuals proved poor at
cate that people are poor at forming prefer- generating complex alternatives or making trade-
ences, generating alternatives, and making offs. Environmental factors, such as issue
decisions (Sniderman et al. 1991; Tetlock 2000; salience, individual attention, and time con-
Jones 2001). Studies show that individuals often straints, shaped the depth of solution searches
lack even the most basic tools with which to in organizations.
make informed rational decisions. Preferences Simon (1985) believed the most glaring
and choices are bounded more by emotion and problem with comprehensive rationality was
environmental context than by rational analysis its focus on outcomes rather than the process
(Jones 2001). Research in cognitive psychology of individual decision-making. By ignoring the
makes one wonder how people are able to form procedures of choice, rational choice theory
preferences or make decisions at all. blithely accepted that ‘rational’ outcome
Herbert Simon (1947) developed bounded emerged from self-interested behavior. Such an
rationality as an effort to reconcile the strict approach is especially dangerous when
economic assumptions of comprehensive ratio- attempting to understand the broader mecha-
nality with actual decision-making revealed by nisms of collective policy-making, where the
the empirical study of organizations. Preferences motivations for preference formation and
and choices seemed bounded by cognitive and political behavior are at least as important as
emotional constraints that interfered with the the outcome itself. Simon (1985: 294) explains:
process of purely rational decision-making.
Much like comprehensive rationality, bounded There is a fundamental difference between substantive
rationality offers an efficient method for and procedural rationality. To deduce substantively, or
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 54


objectively, rational choice in a given situation, we need research have improved our understanding of
to know only the choosing organism’s goals and the the behavioral model of choice, these four
objective characteristics of the situation. We need know
absolutely nothing else about the organism, nor would
central tenets remain largely valid.
such information be of any use to us, for it could not
affect the objectively rational behavior in any way.
To deduce the procedurally or boundedly rational The Principle of Intended Rationality
choice in a situation, we must know the choosing organ-
ism’s goals, the information and conceptualization it has The principle of intentionality suggests that we
of the situation, and its abilities to draw inferences from
look at the goal directed behavior of people,
the information it possesses. We need know nothing
about the objective situation in which the organism and investigate the manner in which their cog-
finds itself, except insofar as that situation influences the nitive and emotional constitutions either help
subjective representation. or hinder their goal-seeking. This distinguishes
bounded rationality from psychological
Simon argued that a theoretical model ori- theories, which generally focus only on the
ented toward understanding procedurally or limitations of individual choice-makers. While
boundedly rational decision-making would pro- comprehensive rationality assumes single-
vide a more realistic bridge between individual minded maximization, the principle of
and collective choice. A ‘behavioral’ model of intended rationality allows researchers to
rationality would take a decidedly scientific and distinguish between careful cost benefit analy-
inductive approach to understanding decision- sis that closely approximates utility maximiz-
making. In order to make efficient generaliza- ing decision-making, quick decisions based on
tions connecting individual psychological heuristic cues, unthinking reliance on past
processes to collective political and economic strategies, or even spontaneous decisions that
choice, research needed to follow strict scientific seem to make no reference to potential gains or
guidelines. Underlying theoretical assumptions losses. Given the time, costs, and demands of a
of human behavior needed to be tested and specific decision, humans may rely on hard-
retested. Those assumptions that were verified wired biological responses, generalized decision-
through scientific research would be preserved in making strategies, or full information searches.
the scientific model, while others would be mod- ‘Cognitive architecture is most obvious when
ified or abandoned. What bounded rationality action occurs at short time scales. As one
lost in parsimony it would gain in accuracy. moves toward actions that take longer times,
Unlike the artificial behavioral assumptions of cognitive architecture is less and less evident,
comprehensive rationality, bounded rationality and the nature of the task takes on more and
would capture the biological, emotional, and more importance in explaining action’ (Jones
environmental constraints that constrained the 2001, 56). Humans may intend to be rational,
procedures of decision-making. but their decision-making capabilities break
As with rational choice theory, bounded down under time constraints or very high
rationality has been widely applied in the study information costs. Moreover, there is no good
of public policy. While bounded rationalists evidence that people are more rational when
might emphasize different elements of behav- the stakes are high, as some rational choice
ioral model in their research, virtually all theorists have maintained. In direct defiance of
research in bounded rationality draws from that claim, state lotteries sell more tickets when
four core principles: the Principal of Intended the pot is large, lowering the probabilities and
Rationality, the Principal of Adaptation, the the expected return.
Principle of Uncertainty and the Principle of
Trade-offs (Jones 2003). From these principles,
modern researchers have advanced a vigorous Principle of Adaptation
research program that explores how both
people and institutions behave. As we will The principle of adaptation is closely related to
see, even though experimental and empirical the notion of intentionality. Much of human
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 55


behavior is explained by the nature of the ‘task on procedures, and may explain instances of
environment’ surrounding a decision. With extremely risk adverse behavior.
time, human decision-making adapts to the
specific nature of the problems they face in
The Principle of Trade-Offs
a specific circumstance. The more time and
learning an individual invests in a specific
The final notion central to models of bounded
problem, the less constrained they become by
rationality is the principle of trade-offs. Unlike
environmental or biological constraints. This
comprehensive rationality, which suggests
notion of adaptation accounts for changes in
that individuals are able to move seamlessly
decision-making efficiency in a single problem
between ranked goals, bounded rationalists
space over time. When a problem is iterated
argue that people find it difficult to trade off
over time, people learn or develop coping
one goal against another when forming prefer-
strategies. Even more intuitively, the principle
ences and making choices (Slovak 1990;
of adaptation may explain why organizations
Tetlock 2000). This critique is a major shift
encourage specialization in areas of complex-
away from the transitivity assumption in ratio-
ity, and routinization in decisions under severe
nal choice theory; however it is important to
time constraints.
capture the volatile shifts in preferences
A rich tradition of research in political
observed by both behavioral economists
science and psychology explores the use of
(Kahneman and Tversky 1986) and public
heuristics in political decision-making—
opinion researchers (Zaller 1992). Preferences
attempting to identify cost-cutting cues indi-
are determined by emotional and cognitive
viduals rely upon in order to limit investment
cues, and are rarely as stable as rational choice
costs from making complex decisions in
theorists have us believe. Because of these
low-information environments. The key issue
trade-off difficulties, Simon argued that indi-
is whether these heuristics are maximally
viduals, and by extension organizations, ‘satis-
adaptive (Lupia, McCubbins and Popkin 2000;
fice,’ quickly choosing an option that is ‘good
Gigerenzer et al. 1999). While some have
enough’ rather than searching for one that
argued that heuristics invariably follow what a
weighs the payoff of every possible choice.
fully rational individual would choose, that is
Because bounded rationality is concerned
quite clearly not the case. ‘Buy a lottery ticket
with the procedures of individual choice, the
when the stakes are high’ is a classic heuristic
school of thought tends to approach the study
that leads to lower expected returns than not
of politics by looking to how individuals
buying at all, or buying when the stakes are
and organizations respond to changes in their
low. It is likely that some heuristics are adap-
problem environments. Because individual
tive calculational crutches and some are mis-
decision-makers have limited attention for
leading and even mal-adaptive.
problem solving, they must address problems
serially, one-at-a-time, which means they are
The Principle of Uncertainty forever juggling inputs, prioritizing them via
the allocation of attention and the sense of
Individuals operate in an environment of urgency that inputs generate. The salience of a
almost constant risk and uncertainty. Because particular problem is almost always generated
of human cognitive architecture, uncertainty is by non-rational elements in politics—by scan-
far more fundamental to choice than expected dal, by crisis, by the mobilization of critics—
utility theory admits. Not only are individuals rather than calm decision to allocate the scarce
unaware of the outcomes that will result from resource of attentiveness.
strategic choices, but they are uncertain of In governments, as well as in all organiza-
the procedures of choice themselves and are tions, attention is allocated in a process politi-
even uncertain about their own preferences. cal scientists call agenda setting. Organizations
Uncertain outcomes may produce dependence also suffer from limited attention spans, and
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 56


must process at least major problems serially. to actual human strategies in real world
Routine problems may be delegated and han- games?
dled according to rules, but fresh problems can-
not be handled this way. It is common for many
problems to press forward on the governmental FRAMING EFFECTS AND PUBLIC POLICY
agenda, and whether the most severe problems
receive the most attention is an empirical ques-
Studies linking framing effects present an espe-
tion. In individuals, as well as in organizations,
cially intriguing challenge to our understand-
the allocation of decision-making attention is
ing of rational political behavior. Framing
crucial for understanding the immediate and
studies have demonstrated that an individual’s
future behavior of an organization.
understanding of a policy idea depends heavily
Because it is explicitly oriented toward under-
on the context and heuristic cues surrounding
standing the processes of decision-making,
a policy problem (Garrett 2003; Jones 1994;
bounded rationality has been widely applied to
Kahneman and Tversky 2000). Rather than
the study of public policy and public adminis-
holding stable opinions about public policies,
tration. Bounded rationalists have explored a
individual opinions are volatile, and shift
range of policy problems, ranging from agenda
according to the framing of the political issue
setting (Kingdon 1995; Baumgartner and Jones
(Zaller 1992). The relationship between a voter
1993), congressional decision-making (Kingdon
and a policy idea is thus at least partially
1973), federal budgets (Padgett 1980), incre-
contingent on the context in which the voter
mentalism (Wildavsky 1964; Davis, Murray,
is presented the policy choice. By ‘directing
Dempster and Widalsky 1966) and risk aversion
attention to one attribute in a complex
(Kahneman and Tversky 2000). As a model of
problem space,’ policy entrepreneurs can pro-
choice, bounded rationality is at least as broadly
duce desired changes to public responses to
applicable as comprehensive rationality.
policy problems (Jones 1994, 104). Public
receptivity to public policies seems highly
responsive to shifts in the framing of a policy


Concerns over competing models of political
choice are by no means limited to researchers
in political science. Some of the most hostile
critiques of the rational choice model have Daniel Kahneman and Amos Tversky’s
come from pioneering work by behavioral research in prospect theory provides one inter-
decision theorists, who have struggled to rec- esting framework for understanding how
oncile the assumptions of the rational calculat- framing effects limit rational behavior. As a
ing economic man with research in cognitive response to errors they saw in Bernoulli’s
psychology that insists that human decision- expected utility assessment of choice under
making is frequently neither optimizing nor risk (Kahneman 2002) prospect theory
rational. The most intriguing contributions advances a boundedly rational model of risky
from behavioral decision theory have come choice, by focusing on the reference point and
from efforts to answer two basic questions framing of decisions rather than the strict util-
about decision-making. First—how do indi- ity of ‘final asset positions’ put forward by
vidual preferences respond to changes in the rational choice theorists (Kahneman 2002,
framing or emotional stimuli of the problem? 460). The approach grew from a series of stud-
Second—how do game theoretic strategies ies in decision-making that demonstrated
based on rational choice assumptions compare how individual preferences were shaped by
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 57


environmental context. People would hold the manner in which people receive and then
different preferences for virtually identical psychologically allocate the use of money has
decisions based on how the problem was pre- an important bearing on how the money is
sented. Kahneman explains: perceived and spent. ‘The effect on current
Preferences appeared to be determined by attitudes to consumption … of winning the $300 football
gains and losses, defined relative to a reference point, pool should be the same as having a stock in
but Bernoulli’s theory and its successors did not incor- which I own 100 shares increase by 3 dollars
porate a reference point. We therefore proposed an a share, or having the value of my pension
alternative theory of risk, in which the carriers of utility
increase by $300. The marginal propensity to
are gains and losses—changes of wealth rather than states
of wealth. Prospect theory (Kahneman and Tversky, 1979) consume all types of wealth is supposed to be
embraces the idea that preferences are reference depen- equal’ (Thaler 1992, 108–109).
dent. (461–462). Yet people do not approach the gain of $300
equally. The $300 won in lottery or betting
Approaching the study of risky choice from
is seen as a sudden windfall, and might be
the perspective of changes in wealth rather
spent with clear conscience on celebration. A
than overall states of wealth, Kahneman and
sudden $300 increase in the pension plan will
Tversky arrive at the intuitively appealing
be treated with miserly caution, an essential
notion that humans cope with gains and losses
part of a savings account linked to concerns for
differently—that people ‘are risk adverse in the
future financial stability. Although the amount
domain of gains, and risk seeking in the domain
is the same, people perceive it differently
of losses’ (Quattrone and Tversky 1988, 723).
because it was accounted for differently. Money
Prospect theory further demonstrates that
can thus be better thought of as placed in
people make different choices depending on
various mental accounts. Money appropriated
how decisions are framed—the weights they
for one use is spent or saved differently than
give in decision-making shift in response to
money appropriated for another (Thaler
how their attention is directed. ‘An individual
would prefer a sure gain of $80 over an 85%
Findings from prospect theory and mental
chance to win $100. The same individual
accounts are compelling because they provide
would prefer a gamble offering an 85%
empirical evidence that individual decision-
chance of losing $100 to a certain loss of $80’
making is strongly constrained by framing
(Berejikian 2002, 762.) Although the chance to
cues which illicit emotional responses. Unlike
win $100 is the same as the chance of losing the
the calculating maximizer posited by rational
$100, individuals generally make different
choice theory, ‘in real life, generating maxi-
choices. In the domain of gains people value
mum utility is neither simple nor smooth, and
certain gains over possible gains. In the
is affected by the cognitive and emotional con-
domain of losses, people will make riskier deci-
stitution of the decision maker’ (Jones 2001,
sions in the hopes of avoiding any loss. These
26). The strategic choices pursued by individu-
findings clearly contradict the tenets of ratio-
als may not reflect calculations of optimality,
nal choice theory, which would assume that a
but rather their disposition towards loss aver-
rational chooser would prefer certain gains or
sion or risk acceptance, or even their emo-
certain losses equally.
tional response to a sudden windfall.
In a similar vein, Richard Thaler’s research
on mental accounting further suggests that
individual preferences and expectations are
shaped by emotional cues. Basic economic
theory suggests that money is fungible, that
money ‘has no labels’ and is treated the same Unlike studies of framing effects, which examine
way regardless of how it is earned (Thaler how individual preferences shift in response
1992, 108). However, Richard Thaler draws a to external emotional cues, experimental
series of hypothetical scenarios to show that game theory began by asking if people behave
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 58


in a manner consistent with game theoretic maximizing behavior in individuals in the lab,
expectations of rational self-interest. To test but they also provide strong evidence that
accurately assumptions of economic choice, human decision-making cannot be fit neatly
individuals must be given clear guidelines and into economic assumptions of comprehensive
all relevant information with which to form rationality. Individual trading behavior in ulti-
strategic preferences. In spite of this conven- matum games has been shown to be highly
tion, the most popular experimental games context-specific. Repeated play games suggest
have provided mixed evidence at best that that learning, social interaction, and norms of
individuals behave in a manner consistent with reciprocity strongly shape ultimatum trading
game theoretic assumptions of strategic pref- (Nowak, Page and Sigmund 2000; Goeree and
erences. Players with full information about Holt 2000). A sense of entitlement seems to
the rules and objectives of the game frequently shape how a proposer or responder plays the
prove themselves to be inefficient in forming game. Players who earn the responsibility of
game theoretic strategies (Fudenberg and proposer engage in distinctly different trading
Levine 1997). behavior than those who are simply designated
By far the most intriguing findings have the role of proposer (Hoffman and Spitzer
come from research in the now famous ‘divide- 1985). Researchers have demonstrated the
the-dollar’ games, which captured the imagi- influences of culture (Roth et al. 1991; Henrich
nation of decision theorists exactly because 2000), and proposer and respondent charac-
they have failed to substantiate the most basic teristics such as sex (Solnick 2001) and age
hypothesis drawn from expected utility theory.1 (Murnigham and Saxon 1994) in ultimatum
In the simplest form of the game, the ultima- trading strategies. Post-game surveys of players
tum game, a proposer and a responder are suggest that norms of fairness, reciprocity,
asked to share a sum of money. The proposer cooperation, and generosity express them-
must divide the sum however he sees fit with selves strongly, even in single shot anonymous
the responder. Because neither player receives trading.
any money if the responder rejects the offer, The brunt of experimental research in
rational choice theory would lead us to predict divide-the-dollar games has remained nar-
that the smallest permissible offers would be rowly focused, first on revealing the patholo-
made and accepted. The proposer would ratio- gies of comprehensive rationality, and then in
nally seek to maximize the monetary gains revealing altruistic, fair or other ‘anomalous’
available in game play, while the responder behavior in research subjects. The discussion
would prefer even an incremental gain over no and conclusion sections of most ultimatum
gain at all. game research almost invariably speculate that
Yet mounting evidence suggests that indi- evolutionary norms of fairness, cooperation,
viduals rarely if ever play as pure income max- or rational risk aversion constrain responders
imizers. Proposers frequently offer up to half from making decidedly one-sided offers. This
of the sum being divided, while responders points to a common complaint of experimen-
routinely reject offers of under one-third of tal research—it is a powerful critique of one
the total—even when neither player knows the model of rationality, but so far the tradition
other, and often simply play against a com- has done little to generate and test new and sci-
puter (Nowak, Page and Sigmund 2000). Even entifically realistic alternatives of choice.
in the dictator form of the divide-the-dollar Experiments are fantastic devices for testing
game, where the responder is required to particular behavioral assumptions, but they
accept any offer made by the proposer, an are less useful in generating new ones.
impressive 76% of responders chose to divide Despite these limitations, efforts to replicate
the money equally (Kahneman, Knetsch and assumptions of comprehensive rationality have
Thaler 1986). succeeded in fueling a movement to integrate
Ultimatum game experiments not only rou- findings of actual behavior into a more realis-
tinely fail to reproduce purely rational utility tic model of human choice. Findings from
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 59


cognitive psychology, cultural anthropology, reality albeit at the sacrifice of a neater and more
political science, and behavioral economics, parsimonious model. Nonetheless, as the maximand
more closely approximates reality, the more reality it
demonstrate how considerations outside of should capture’ (Levi 1997, 24–25).
strategic rationality influence the procedures of
individual decision-making. Experts across dis- Individual preferences may be modified to
ciplines increasingly agree that a successful capture not only an individual’s desire to max-
model of human choice must integrate empiri- imize gains, but an equally strong ‘ethical’
cal findings about the effects of culture, attention belief that the individual’s behavior harms no
streams and issue salience, emotional cues, fram- other individual’s. These rational choice models
ing effects, and identity and social perspective integrate assumptions of fairness, altruism,
into a parsimonious model of decision-making. reciprocity, justice, etc., into the ordered pref-
This research agenda has led to renewed interest erences of their rational actor. The individual
in established theories of decision-making, such continues to hold ordered preferences, and
as bounded rationality and prospect theory, as continue to make decisions based on whatever
well as a wealth of new and modified theories will yield their most preferred outcome. The
from the rational choice tradition. problem, of course, is that trade-offs among
these various goals must be explicitly built into
the model, and one of the primary findings of
RATIONAL RESPONSE: THIN behavioral decision theory centers on the diffi-
RATIONALITY AND NEW culties people have with such trade-offs.
INSTITUTIONALISM Another hallmark of recent research in
rational choice theory has been to give greater
attention to the institutions in which political
Although a few dogmatic economists and
actors make choices. Institutional Rational
political scientists have continued to conduct
Choice Theory examines how institutions pro-
research through assumptions of comprehen-
vide specific information about the gains and
sive rationality, most recent rational choice
losses available to actors in a given sphere.
studies in political science have attempted to
Institutional rational choice theory is espe-
integrate empirical findings of individual pref-
cially concerned with understanding how
erences into renewed models of individual
organizational cultures permit cooperation in
rationality. It is clear that the traditional ratio-
places where selfish maximization might prove
nal choice assumption of self-interest has
problematic (Miller 1992; Ostrom 1998). In
failed—that the self-interest axiom cannot
many situations, rational maximization at the
explain instances of altruism, ethical restraint,
individual level leads to collective disaster—as
or fairness—has lead an increasing number of
in Garret Hardin’s (1968) tragedy of the com-
theorists to allow maximization for any goal.
mons, where unlimited common grazing
The analyst refuses to make assumptions about
rights lead to destruction of the common pas-
motives for maximization. These steps may
ture illustrates. Through learning or experi-
complicate the model or limit broad theoreti-
mentation, organizational rules may develop
cal generalizability; however, the growing evi-
that encourage collective outcomes that overcome
dence documenting inconsistencies between
the “tragedy” of individual self-maximization
rational choice assumptions and actual human
(Ostrom 1999a). Once in place, these rules are
behavior warrants a revision of the theory to
enforced through a system of formal and
include the potential rewards outside of mate-
informal incentives that control selfish behav-
rial or status gain. Margaret Levi points out:
ior. This framework provides ‘a general lan-
The addition of non-egoistic considerations or motiva- guage about how rules, physical and material
tional norms [as community standards or fairness prin-
conditions, and attributes of the community
ciples] does increase the complexity and difficulty of
analysis… . The advantage of an assumption in which affect the structure of action arenas, the incen-
actors consider net payoffs that include both material tives individuals face, and the resulting outcomes’
and ethical factors is that it may better approximate (Ostrom 1999b, 59).
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 60


These institutional and cultural extensions management; however we fear the privileged
are the most important additions to rational place such research now holds in the study of
choice theory to be put forward in recent years. organizations provides students with a simplis-
Yet, by incorporating so many “soft” variables tic and misleading view of the dynamics of
they move quite briskly away from the tradi- bureaucratic policy-making.
tional approach. How can one predict whether A more comprehensive understanding of
an organizational culture will emerge to over- bureaucratic decision-making can be gained
come rational self-interest? Clearly not always, through the lens of bounded rationality, which
and hence the institutional structure is depen- offers a scientifically sound alternative model
dent on some other context that is left unspec- of behavior while sacrificing little parsimony.
ified in the model. Rather than focusing exclusively on control,
bounded rationality focuses on how changes in
the external environment shape information
processing both within and across organiza-
tions. This approach to the study of organiza-
tions will invariably yield a more sophisticated
understanding of bureaucracy, as it is better
In spite of improvements, rational choice suited to link the procedures of human choice
models simply are insufficient to capture criti- with broader policy processes.
cal aspects of organizational behavior. Because
this research remains focused on outcomes
rather than processes, rational choice theories
are overwhelmingly focused on problems of
collective action and individual utility maxi-
mization, which provide only limited insights
into organizational behavior. Bounded ratio- Rational choice decision-making as applied to
nality can examine the roots of collective the implementation stage of the policy process
action problems, but it also can shift research generally takes the form of principal-agent
attention toward the more interesting ques- models of the bureaucracy’s interactions with
tions of attention allocation and organiza- the legislative, executive, and judicial branches
tional information processing. of government. In this section we argue that
Recent studies in public administration have the underlying assumptions of the principal-
focused a good deal of attention on problems of agent model of politico-bureaucratic relations
delegation and control in organizational decision- render it inaccurate, not only in terms of ade-
making. Indeed, virtually all theoretical discus- quately explaining the relationship, but also
sions of policy implementation have centered as a description of the relationship and policy
on these questions to the exclusion of such processes more generally at the implementa-
important topics as adaptive problem-solving tion stage.
and information-processing. While principal- Principal-agent models of behavior were
agent dilemmas illuminate some important originally developed in the economic analysis
aspects of bureaucratic behavior, its promi- of firms (Alchian and Demsetz 1972);
nence in studies of public policy is partially an although an emphasis on efficient production
artifact of the rational choice model of behavior. and incentive structures in organizations dates
The fundamental problems of principal-agent much earlier (Taylor 1911). The principal- agent
dilemmas—those of information asymmetries, relationship is essentially a contract. In order
moral hazards, and incentive structures—are to increase production, the principal enters
those that map neatly onto the most basic into an agreement with the agent for the pro-
assumptions of individual utility maximization. duction of the good. The principal’s goal is to
Pioneering work in principal-agent models ensure the efficient production of the good;
have spurred a rich discussion in public meanwhile, the agent’s goal is to avoid time,
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 61


effort, and resources necessary for producing about the relationship of the actors to their
the good. goals. The principal-agent model assumes that
For example, if we were examining a busi- a given actor’s goals are known. Formal principal-
ness firm, we might consider the employer the agent modeling has rightly recognized that
principal and the employees as agents. In the there exists uncertainty about the goals of the
model, employees have incentive to “shirk”, that other actors in the relationship (Bendor and
is, loaf. This means that managers must closely Moe 1985; Bendor, Taylor and Van Gaalen
monitor them to insure efficient production of 1985; Bendor, Taylor and Van Gaalen 1987a;
a given good. No system of monitoring can Bendor, Taylor and Van Gaalen 1987b; Moe
completely eliminate all free-riding, because 1984), but this does not take account of the
the resources for such monitoring face the law uncertainty that shrouds an actor’s knowledge
of diminishing marginal returns. This is the of her own goals. Actors may not be able to
essence of the principal-agent problem. define their own goals, especially in contexts
Why do principals face the problem of free in which trade-offs must be made between
riding? The answer lies in information asymme- competing goals.
tries found in the relationship between the prin- The model also assumes that strategies for
cipal and the agent. In the principal-agent attaining given goals are easily attached to the
model, the agent holds two key informational goals. The principal’s goal of efficient produc-
advantages over the principal. The first infor- tion is attained by developing the appropriate
mational advantage lies in the selection process. incentive structure such that it is to the agent’s
Principals cannot know a priori the exact quali- material disadvantage to shirk rather than work.
fications of a given agent, while the agent has a Likewise, the agent’s goal is to maximize the
better idea of his or her competence. The agent benefit to be gained from the contractual rela-
may exaggerate her ability in order to attain the tionship with the principal through mani-
position, and even if most exaggerations are pulation or information. The final assumption
caught by the employer’s personnel policies, concerning goal oriented behavior is complete
some incompetents will slip through. This first information. The model assumes that if the
informational asymmetry is usually termed principal is able to observe agent outputs, the
adverse selection. The second informational principal is able to judge whether or not these
advantage held by the agent involves the moni- outputs conform to the goal of efficient produc-
toring of the agent after the contract has been tion (Worsham, Eisner and Ringquist 1997).
established. Moral hazard results when the prin- Finally, the principal-agent approach makes
cipal lacks the resources or ability to engage in certain assumptions about the nature of
constant monitoring of the agent. The agent is the principal-agent relationship over time. The
then able to shirk, reducing the efficiency of the principal-agent approach is based in compara-
firm. The task of the principal is to induce com- tive statics (Bendor and Moe 1985). Rela-
pliance on the part of the agent through a tionships change only when exogenous ‘shocks’
system of incentives. to a system disturb a system otherwise in equi-
In order to invoke the principal-agent librium. It should be noted that this is not
framework, the analyst must assume goal entirely a problem of principal-agent theory
incongruence. The goal of the principal is the per se. Comparative statics and equilibrium
production of some good, while the goal of the analysis also has roots in systems theories in
agent is to reap the rewards of the contract both political science and sociology (Eisner,
with the principal without expending the Worsham and Ringquist 1996; Worsham, Eisner
effort to produce the good. To put it simply, the and Ringquist 1997; Worsham, Rinquist and
goal of the agent is to shirk. If goal incongru- Eisner 1998).
ence does not exist, then principal-agent Given these assumptions, it is imperative to
models are inappropriate for examining the the success of the contractual relationship that
relationship. Underlying the assumption of the principal control the behavior of the agent.
goal incongruence is a further assumption There is no room for shared goals or learning
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 62


and adaptation of behavior over time, or the Moe (1984) argues that for principal-agent
potential counterproductive results on pro- theory to be useful in the study of implemen-
ductivity of an environment of continual dis- tation, we must take account of the fact that
trust and suspicion fostered by monitoring. politicians are not necessarily motivated by the
The theme of control has been central in sub- efficient production of public service; they may
sequent scholarship on the implementation be more concerned with political efficiency
stage of the policy process. rather than production efficiency. Moreover,
Rational choice approaches were used in the the major problem of control might not be
study of public policy long before the advent of shirking, but could involve several different
the now-dominant principal-agent approach, possibilities, including material benefit of
most notably in Niskanen’s (1968; 1971; 1975) some sort, ranging from budgetary slack to
model of the budget-maximizing bureaucrat and promotion, but might also be policy related. At
Anthony Downs’ (1967) study of information a minimum, scholars utilizing the principal-
limitations and bureaucracies. The principal- agent frameworks would need to grapple with
agent model really burst onto the scene in politi- the issues of political efficiency and diversity of
cal science with works by Mitnick (1980; see also goals—not trivial alterations.
Mitnick 1973) and Miller and Moe (1983a;
1983b). This culminated in an article by Moe
(1984), The New Economics of Organization,
which formally introduced principal-agent
approaches to the study of policy implementa-
tion to political scientists. Moe insightfully uni-
fied principal-agency from the economic analyses With the formal introduction of principal-agent
of business firms to the issue of how democrati- theory, work on policy implementation col-
cally elected officials can control unelected public lapsed to analyses of how a political principal is
bureaucrats; hence principal-agency could be to control a shirking bureaucracy. One group of
used to illuminate a critical problem in democra- scholars argues that political control of the
tic theory. Unfortunately, it altered the perspec- bureaucracy is achieved ex ante (McCubbins
tive on bureaucracy from one of balancing 1985; McCubbins, Noll and Weingast 1989).
problem-solving, information-acquisition, and Democratically elected officials gain control of
the advantages and disadvantages of delegation, the bureaucracy through setting the ‘structure
to one stressing a single- minded emphasis on and process’ of organizations before their actual
formal procedures of monitoring and control. creation through detailed legislation specifying
It also reversed the old normative emphasis administrative procedure, personnel, and orga-
in public administration on keeping politics nizational structure (McCubbins, Noll and
out of administration. At least since Woodrow Weingast 1989; Banks and Weingast 1992). In this
Wilson (1887), political scientists have been view, Congress ‘stacks the deck’ against the agency
concerned with the distinction between poli- during its creation in order to ensure subservience
tics and administration, mostly from the per- to Congress or organized interests at later periods.
spective of trying to keep politicians from The deck-stacking thesis focuses on political
interfering with professional implementation efficiency—the political coalition in control
by demanding special favors from supposedly wishes to continue its advantage through later
neutral bureaucrats. Principal-agency suggests implementation. Other scholars argued that the
that all independent behavior of bureaucrats is political branches of government could also
not motivated by professional problem- achieve ex post control of the bureaucracy
solving, but is based in the desire to shirk. through budgeting, personnel, staffing, political
Principal-agency assumes that all attempts by appointments, ongoing interactions with the
politicians at control are motivated by “efficient bureau, and congressional oversight hearings
production” rather than crass political gains (Bendor and Moe 1985; Bendor, Taylor and Van
that add inefficiencies to public production. Gaalen 1987a; Miller and Moe 1983b).
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 63


But is congress the principal in the American is no room for politicians to prefer neutral
system of divided powers? Both congress and competence in administration—the major aim
the president attempt to influence the bureau- of the Progressive Movement so important
cracy (not to mention the judiciary). Some in American political life during the first quar-
scholars argue that congress is the institution ter of the 20th Century. To the extent that
wielding the most influence or control on the Progressives were successful (and in many
bureaucracy (McCubbins 1985; McCubbins, respects they were very successful), they left
Noll, and Weingast 1989; Weingast and Moran a legacy of removing political control and the
1983). Other scholars see the president, with his politicized administration that the deck-
ability to influence the bureaucracy through stacking thesis suggests.
budgets, OMB, and political appointments, as
the abler of the two branches to achieve control
and focus their theoretical and empirical atten-
tion on him (Golden 2000; Wood 1988; Wood
and Waterman 1991).
McCubbins, Noll, and Weingast (1989) put Beginning in the 1990s, scholars began to argue
forth three hypotheses concerning agency that descriptive flaws led agency theorists to
structure and process and its effects on the explanations of bureaucratic behavior that
subsequent regulatory environment. First, the were less than adequate (Eisner, Worsham, and
agency’s design should, ‘create a political envi- Ringquist 1996; Hindera and Young 1998;
ronment that mirrors the politics at the time of Krause 1996; Meier, Pennington, and Eller
enactment’ (McCubbins, Noll, and Weingast 2001; Potoski and Nemacheck 2001; Worsham,
1988, 444). That is to say, winners and losers at Eisner, and Ringquist 1997). Principal-agent
the formulation stage of the policy process analysis generally has difficulty accounting for
should benefit or lose in the same relative pro- the fragmented nature of policy processes in
portions at the implementation stage due to the American system of government. That is,
institutional design. Second, agency structure there are multiple, competing principals in the
and process will favor those constituencies of American system of separated powers (Choi
the winning coalition at the formulation stage. 2001). Moreover bureaucracies themselves
Finally, agencies will exhibit an ‘autopilot’ have constituencies and engage in advocacy
characteristic (McCubbins, Noll, and Weingast policy-wise for these constituencies (Hindera
1988, 444). As the preferences of the favored and Young 1998; Meier, Pennington, and Eller
constituency change, so will the preferences 2001; Potoski and Nemacheck 2001). These
and thus behavior of the agency. bureaucratic constituencies could also be con-
More recent scholarship is divided on the sidered the principals.
efficacy of structure and process as a means for All of these potential principals compete to
controlling the bureaucracy. Balla (1998) finds influence or control the bureaucracy. These
that the open comment procedures for the new competing principals often have multiple, con-
pay schedule and rules regarding Medicare did flicting goals, especially under divided govern-
not favor the groups presumably favored by ment. Once the political context of multiple,
the dominant coalition in congress. Balla and competing principals is taken into account,
Wright (2001) did find that advisory commis- bureaucratic intransigence becomes much
sions correctly represented the political fac- harder to pin down conceptually. To whom
tions involved in the legislation of drinking should the agency respond? Why should the
water, but these authors did not examine the agency respond to one principal to the exclusion
actions of these commissions. of others? We argue that the more appropriate
Note that in the development of the principal- question to ask is, how do bureaucracies make
agent literature, the descriptive fit of the trade-offs among the competing and contradic-
principal-agent model is somewhat strained. tory demands of their political environment?
For example, in the deck-stacking thesis there Answering this question requires that we take
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 64


an information processing approach to the How do principals know which strategies will
study of bureaucratic behavior. lead to the desired outcomes?
A second descriptive criticism of agency Furthermore, it was argued long before
theory involves the goals of the agent. The goals principal-agent theory came to dominate imple-
of the agency are assumed to be adverse to those mentation studies that success in terms of goals is
of the principal. Principal-agent modeling has difficult to measure (Lipsky 1980; Worsham,
difficulty addressing agents whose goals are not Eisner, and Ringquist 1997). How do principals
consistently adverse to those of the principal. know desirable outputs when they see them?
Recent research at the street level calls into ques- Monitoring outputs can be counterproductive as
tion the assumption that bureaucrats are pri- agents shift away from productive problem-
marily motivated by the goal of shirking solving (as well as shirking) toward the goal that
(Brehm and Gates 1997). Krause (1996) argues is monitored. This is a classic critique of stan-
that the development of goals over time is a dardized testing in schools as teachers “teach to
two-way process. Agencies influence and even the test”. The linkage between strategies and goals
help to refine the goals of political principals. cannot be understood outside a dynamic rela-
Second, using principal-agent models requires tionship. This suggests that bureaucratic out-
the researcher to acknowledge that both the comes, rather than outputs, are the more
principals’ and agents’ goals are known and important indicators of goal success, which, in
easily prioritized. Bender and Moe (1985) note turn, suggests the efficacy of a problem solving
that the relationship between the bureaucracy approach to implementation.
and its political principals is likely dynamic, and If bureaucrats’ interests are not necessarily
the goals of the actors are likely to evolve and are adverse to those of the principal, an emphasis
clarified over time, both by the interactions of on control as a description of the process of
the bureaucracy with the political principals implementation may be misleading. Golden
as well as through organizational factors. (2000) finds that bureaucrats in various regula-
Worsham, Eisner, and Ringquist (1997) have tory agencies in the Reagan administration for
argued that we cannot assume that bureaucratic the most part were responsive to the policy
agents are motivated to maximize material directions of the administration, even in the
return to themselves; they may have other absence of formal control procedures. Brehm
goals—some of which may correspond to those and Gates (1997) find that bureaucrats are not
of the policymaking branches. This raises fundamentally interested in shirking. They
several questions concerning the type of behav- argue that functional motivations and peers are
ior we should expect from the bureaucracy. the prime influences on intra-organizational
When confronted with a choice between behavior to the exclusion of influence by supe-
attaining one of many possible goals, how riors within the organization. Bureaucrats, for
does an actor choose? We are back to a serious the most part, work. Their research adds a
problem in all rational theories of choice: to strong empirical foundation to arguments advo-
make any progress, some very strong assump- cating taking organizational context seriously
tions must be made about what goals are used (Lipsky 1980; March and Simon 1958; Simon
in the process of maximization. A bounded 1947; Wilson 1989).
rational view of individual decision-making
takes account of the difficulties of making trade-
offs among multiple and competing goals.
Principal-agent theories of policy processes
also assume that the strategies, or means for
attaining goals, are easily attached to the goals
themselves. Although the studies outlined How are we to come to grips with the fact
above all note that goals are not easily that bureaucrats are often predisposed to be
achieved, the uncertainty that surrounds the responsive, but that policy outcomes are not
means to achieve the goals is rarely modeled. always exactly as political principals would
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 65


have them? If organizational decision-making in environments characterized by ambiguity

is our focus, then we should approach the and uncertainty (Jones 2001; Jones 2003;
problem by asking how organizations process Krause 2003). An emphasis on discovering the
information from both their political and task ways in which agencies balance competing,
environments (see Wilson 1989), and how this contradictory demands in their environments
information is used in making trade-offs mandates that scholarly attention be turned
among competing demands and functional toward specifying agency decision-making
requirements (Jones 2001; Jones 2003). processes. It also mandates that scholarly
An information-processing approach would attention be turned toward explaining bureau-
analyze the ways in which administrative cratic behavior more broadly, rather than one
structure and processes are developed in ways facet of the relationship between bureaucracy
that aid in focusing attention on particular and the elected branches of government.
problems and types of information. It would
need to rely on bounded rationality as a deci-
sional underpinning, because the cognitive
limits of individuals and organizations color
how they process information. This approach
would view hierarchies or organizational
structures as mechanisms for focusing attention If we adopt a problem-solving, information-
(Jones 2001). Normatively, an information- processing perspective based in bounded ratio-
processing approach sees structure and process nality, we are equipped to examine both
aimed at enhancing the problem-solving capa- scientific and normative questions in a more
city of both agencies and Congress more gen- productive light than if we begin with a rigid
erally, not just enhancing one partner at the principal-agent model based in rational choice.
expense of the other. Constraining agency Because bounded rationality begins with the
decision-making renders the agency less able assumption that both “principals” and “agents”
to adjust both to its task environment and its are goal-seeking entities with multiple potential
political environment (Jones 2001; Wilson objectives, and assumes that both are fallible in
1989). Keeping politicians completely out of the pursuit of these goals, it leads to normative
implementation can cause agencies to cut off analyses that neither treat the “professional
an important source of feedback. bureaucrats” as invariably correct, as did the
It is reasonable to argue that members of Progressives, nor the elected politicians, as do the
congress might well prefer a flexible agency proponents of principal-agency. Scientifically, it
that is adaptable both to its political and task is much less likely to “get stuck” on one partial
environment; delay may be desirable in order and incomplete understanding of legislative-
to maintain policy stability (Carpenter 1996). bureaucratic relationships, because it has a
Agencies are rightly cautious in responding too much stronger inductive component than does
quickly to demands from their environments. principal-agency or any other models derived
Agency delay is a result of the agency reducing from rational choice.
the uncertainty in its environment. Once the Because bounded rationality begins by look-
agency is sure that a given signal is meant to ing to the procedures of individual choices,
influence policy in a given direction, the it is well equipped to explore endogenous
agency adjusts accordingly. dynamics of interpersonal and hierarchical
This approach calls into question a theory of relationships at the heart of principal-agent
bureaucratic behavior that fails to consider the dilemmas. But the approach isn’t limited to the
uncertainties inherent in the environments of internal dynamics of public administration.
agencies and questions the amount of Because bounded rationality looks to how
resources that are devoted to questions of con- decision-makers interact with exogenous envi-
trol without actually specifying how agencies ronmental changes in time, information flows,
go about making decisions—decisions made and attention—it is also able to understand
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 66


dynamic change in the policy outputs of relationships are not a caldron of conflict. The
organizations. In this regard, we find it curious environment is dynamic and conflict varies
that researchers in public administration are according to both the level of information that
calling to expand principal-agent models of principals and agents posses and the level of
bureaucratic behavior (Waterman and Meier goal conflict’ (197).
1998; Bendor, Glazer and Hammond 2001). Perhaps the most troubling set of problems
A scientifically sound and parsimonious identified by critics of principal-agent models
alternative approach connecting individual have been that they are founded in artificial
and organizational behavior already exists. To and even incorrect assumptions of human
demonstrate the value of bounded rationality behavior. Under empirical scrutiny, the assump-
in the study of organizations, we describe first tion of goal incongruence generally, and agent
how the model might be applied to principal- shirking specifically, is less pronounced than
agent dilemmas and to broader and more principal-agent theory would have us believe.
interesting questions about bureaucratic infor- In many organizations, agent behavior is deter-
mation processing. We contend that bounded mined more by the organizational culture and
rationality actually provides a more robust feelings of agent solidarity than through incen-
understanding of principal agency. More tive structures and principal monitoring
importantly, we argue that those researchers (Brehm and Gates 1993). The importance of
who study organizational behavior through institutional norms and cultures inspired Gary
the lens of bounded rationality look to a Miller’s Managerial Dilemmas, one of the most
greater and more appealing range of research exciting innovations in modern organization
questions than the problems of delegation and theory (Miller 1992). Miller’s work challenged
control. the Skinnerian notion that workers in agency
were most effectively induced to work through
incentives and punishment, arguing instead
that an organizational culture is far more effec-
tive in establishing worker compliance or non-
compliance with principal goals (Miller 1992).
Nothing in bounded rationality rejects that
Recent empirical challenges to the central incentives and monitoring within organiza-
assumptions of comprehensive rationality have tions can shape individual behavior—and
spurred calls for reform in either the method or Miller’s interest in organizational culture is
the behavioral model employed in studies of entirely consistent with procedural rational-
interpersonal relationships. In their elegant ity’s interest in the influence of external factors
review Theories of Delegation, Bendor, Glazer, in individual choice.
and Hamond (2001) conclude with a challenge Yet bounded rationality is not merely as good
to improve the study of principal-agent dilem- as new institutionalism in explaining principal-
mas, complaining in part that too little attention agent relationships—it performs better. This
is given to a) the ‘real world’ institutional con- superiority stems from the considerable atten-
text of principal-agent dilemmas (266–267); tion bounded rationality places in the cogni-
b) pre-existing information and policy pro- tive constraints on decision-making. New
grams that might constrain an actor’s behavior; institutionalism is an enormous improve-
and c) the complexity of the task environment ment over traditional principal-agent studies
facing actors in a given environment (267). because it built such exogenous factors as
Waterman and Meier (1998) attack the limita- culture into the decision-making model.
tions of standard principal-agent dilemmas Unfortunately it adds, in an ad hoc way, what-
from another direction, deriving an interesting ever the analyst thinks actors are “maximizing”,
set of hypotheses by relaxing assumptions and generally ignores the trade-offs implied
of information asymmetries and goal incon- between the standard assumption of self-interest
gruence. They argue ‘all political-bureaucratic and the other imputed goals that might get
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 67


maximized. Find a deficit in the theory, and a preferences; however, bounded rationality
new goal is added. distinguishes itself by capturing psychological
Bounded rationality takes but one addi- as well as environmental constraints on choice.
tional step by showing how biological hard- Although inductive research is more time
wiring and psychological traits shape how consuming, such an approach will ultimately
individuals act within organizations, including yield a more compelling portrayal of principal-
the complex relationship between incentives agent dilemmas. In bounded rationality, indi-
and response. Having moved so rapidly away vidual responses to monitoring, incentive
from simple models of self-interest maximiza- structures, routines, and organizational mis-
tion, and having recognized that “informal sion are no longer abstracted from artificial
norms” such as a sense of fairness can be models, they are grounded in the science of
maximized, why do rational choice theorists observation.
not simply take the next small step and intro-
duce the individual and organizational cogni-
tive architectures that bounded rationality
Bruno Frey’s research on ‘the crowding out
effect’ (Frey 1993; Osterloh and Frey 2000) is
an excellent example of how bounded ratio- While bounded rationality may well improve
nality might improve models of principal- research in principal-agent dilemmas, the thrust
agent relationships, and it shows the practical of this argument is not to rescue principal-agent
danger of believing that agents simply respond theory. Instead, we hold that starting with the
to incentives provided by the principal. assumption of bounded rationality will stimu-
Drawing from psychological studies that show late a wide range of interesting research ques-
individuals respond negatively to control, Frey tions in organizational information processing,
argues that ‘under readily identifiable condi- which will ultimately provide us with a far more
tions, increased monitoring reduces an agents robust understanding of bureaucratic behavior.
overall work effort’ (1993, 663). Increased This final section looks to the unique emphasis
monitoring breaks the ‘psychological contract’ of bounded rationality on how individuals and
with the agent, lowering his or her incentive to organizations receive, process, and act upon a
respond to incentives or punishment. Not only variety of information cues.
is increased monitoring not terribly effective, The promise of bounded rationality is that it
but it might prove harmful to organizational allows researchers to connect individual
productivity—a blunder based on theory behavior to organizational output without the
rather than analysis. Frey then compares rigid and misleading constraints of rational
behavior across organizations to identify choice. For the study of public policy processes,
which types of organizations will produce such a model is enormously appealing. It
a positive (increased productivity) or a nega- allows researchers to explore political behavior
tive (decreased productivity) by increasing in the aggregate by making assumptions about
monitoring—a much more scientific and the behavior of the individual.
inductive approach than the standard deduc- The earliest advances in organizational pro-
tive stance taken by principal-agent theorists. cessing models emerged from the study of
By looking to the psychological and envi- public budgeting (Simon 1947; Wildavsky
ronmental constraints on individual choice, 1964; Davis, Murray, Dempster and Wilasky
bounded rationality allows researchers to cap- 1966); see Thompson and Green 2001 for a
ture realistic and counter-intuitive behavior critical review). Bounded rationality was used
that escapes rational choice analysis. New insti- to explain incremental changes in public bud-
tutionalism and bounded rationality have both gets. The high costs of generating alternatives
discovered the importance of organizational and making fully optimal decisions, coupled
rules and norms in delineating individual with the uncertainty of outcomes, led to what
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 68


might be best characterized as risk averse Low salience issues will generally follow an
policy incrementalism. At the core of many incremental pattern—little new information
explanations of budget incrementalism was or alternatives will be integrated into the solu-
Simon’s notion of satisficing. To limit the costs tion searches. When attention is focused on a
of making decisions, both individuals and high salience policy problem, we anticipate a
organizations looked to alternatives that were broader solution search, greater generation of
good enough. Budget incrementalism emerged alternatives, and increased legislative activity.
from the risk averse convention of referring to However, even when attention is squarely
the prior-year’s budget when creating a new focused on a problem, decision-making will
one. ‘Because the reference point for decision- be bounded by cognitive and emotional
making is always some point decision made in constraints.
the past, the outcomes of local search, com-
bined with a propensity to limit bargaining
and enforcement costs, are small or incremen-
tal changes in policy’ (Thompson and Green
2001: 3).
While early models of organizational infor-
mation processing focused on the politics of In The Politics of Attention, Bryan Jones and
incrementalism, more recent studies of Frank Baumgartner (2005) outline the compo-
bounded rationality have focused on the nents of a comprehensive model connecting
dynamics of policy change. Behavioral models individual and organizational behavior. This
here have been driven by an intriguing empir- model is drawn from bounded rationality, and
ical reality—neither budgets specifically (Padgett attempts to capture both the dynamics of indi-
1980; Carpenter 1996) nor policy agendas more vidual and collective choice at each stage of the
generally (Baumgartner and Jones 1993) policy process.
adhere to the strict pattern of incrementalism. Figure 3.1 depicts their information-
Instead, budgets and agendas follow the trajec- processing model. Notice at each stage the
tory of punctuated equilibriums—periods authors have characterized the decision-making
of incrementalism followed by a sudden flurry process at the individual and the systems level,
of policy activity (Baumgartner and Jones because organizations fall prey to the same
1993). These recent studies have improved our kinds of cognitive limits as do individuals
understanding of the policy processes by inte- (Jones 2001).
grating the role of shifting attention into At the core of the model is the problem of
models of organizational behavior (Jones issue attention. Organizational agendas reflect
2003). Students of how governments set policy what actors believe to be the most salient or
agendas emphasize the role of the allocation of pressing concerns. Because both organiza-
attention in determining the behavior of tional and individual attention is limited,
policy makers (Cohen, March, and Olsen 1972; agenda setting necessitates that organizations
Kingdon 1995; Baumgartner and Jones 1993). prioritize political problems, focusing first and
Shifts in the external environment produce foremost on the most urgent concerns. These
shifts in the preferences and goals of policy- concerns may be dictated less by informed rea-
makers. When an institution or organization’s soning and more by emotional responses to
attention is squarely focused on a specific political problems, ‘as emotions are critical in
policy domain, we might anticipate increased determining priorities’ (Jones 2001: 73–74). In
activity, legislation, and spending. However, organizations, the high priority issues will
because the allocation of attention is limited to receive the brunt of attention, while less salient
a few select issues, many policy programs con- concerns will fall to the wayside. If these less
tinue incrementally—and follow the conserva- salient issues demand political action, they
tive, risk averse path outlined by prior research may follow the path of incrementalism and
in public budgeting. emphasis on ‘pre-packaged’ solutions.
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 69


Individual Level System Level

Which issue
Attention allocation to attend?
Agenda setting

Issue A Issue B Issue C . . . Issue K

Which attributes
Problem characterization Problem definition
to incorporate?

Attribute I Attribute II ... Attribute G

Which solutions
Alternative generation Proposals and debate
to examine?

Alternative 1 Alternative 2 Alternative 3

Choice Collective (policy) choice

Which alternative
to select?


Figure 3.1 An information-processing model of choice

The next step in the information-processing The final two stages of the model describe
model takes the form of problem characteriza- how individuals and institutions reduce a
tion and organizational problem definition. problem to a manageable set of alternatives, and
Received information is rarely neutral— then select a policy solution. For example—
individuals and organizations must decide how policy proscriptions addressing failing schools
to characterize and approach a particular might take any number of forms—smaller class
problem. Understanding how individuals char- sizes, increased funding, increased oversight,
acterize and organizations define the nature of a school and student accountability, curriculum
political task is essential in the policy process. overhaul, etc. Because decision-makers struggle
For example, failing students in public educa- at evaluating and making trade-offs (Sniderman
tion might be seen primarily as a problem of et al., 1991; Tetlock 2000; Jones 2001), choices
waning parental involvement, or it might be must be reduced to a few actionable alternatives.
seen as a systemic failure of the school system. In organizations, this is often done through
The received information may be characterized debate. Failing schools can be solved through
in any number of ways, depending on the pre- increasing funding, or through increasing
conceived attitudes of the decision-maker, or accountability and oversight. The choice is then
the larger organizational culture. made according to institutional procedures.
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 70


It is important to note that political respond to the uncertain and ambiguous

decisions may not represent individual res- information pressing down on them.
ponses to politically neutral information. A growing body of evidence criticizing the
People often become attached to preferred empirical and theoretic inadequacies of compre-
solutions beyond their direct utility—a process hensive rationality has led an increasing number
Herbert Simon referred to as ‘identification of researchers to call for a renewed model of
with the means’ (Jones 2001). The emotional choice in public decision-making. In political
orientation toward a solution or particular set science, rational theorists have admitted to a vari-
of solutions bounds the alternatives a decision- ety of organizational and cultural constraints on
maker is willing to consider. In politics, politi- the maximization of self-interest, to the extent
cal ideology affects how people prioritize that even altruism can be “rational”. This destroys
problems, construct problem spaces, and orga- parsimony and the deductive method, for one
nize solutions. Because ideologies help define never knows what is being maximized without
“the self ” the solution implied by the ideology empirical examination. Yet rational choice theo-
will be very resistant to information. rists still refuse to take the final step, the admission
The entire process of attention allocation, that cognitive limits of individuals also affects the
problem definition, and attaching solutions to decision-making process.
problems must of necessity be a disjoint and A full return to bounded rationality is
episodic process. Governmental response will necessary to carry forward the new demands for
not be proportionate to the severity of the realistic yet theoretically tractable models of
problem—the “exogenous shocks” of rational individual and organizational choice. Only by
choice’s favored comparative statics. One of dropping the increasingly weighty baggage of
the major empirical predictions of bounded rational choice and its misleading theories repre-
rationality is that, taken in dynamic terms, sented by principal-agency can we proceed pro-
policy responses will be far more punctuated ductively, with an increased appreciation for
than would be predicted from a fully rational observation as well as theory. After all, the
response alone (Jones and Baumgartner 2005). approach was born in the disciplines of political
science, organization theory, and public admin-
istration. Instead of borrowing from economists,
poorly equipped to study politics, we are simply
re-acquainting ourselves with our past.
This essay has distinguished between rational
choice and bounded rationality in studies of
public policy. We began with a comparison NOTE
between the two approaches, then detailed the
objections to rational choice stemming from 1. For a good review, see Colin Camerer and Richard
the laboratory experiments in behavioral deci- Thaler’s ‘Anomalies: Ultimatums, Dictators and Manners,’
Journal of Economic Perspectives (1995) 9: 209–219
sion theory. Taking principal-agent theory as
the major perspective used today to analyze
policy implementation, we showed how it has REFERENCES
mis-directed scholarly attention to one issue in
bureaucratic behavior to the exclusion of many
others, and produced a questionable approach Alchian, Armen A., and Harold Demsetz. 1972.
Production, Information Costs, and Economic
to understanding how to improve bureau-
Organization. American Economic Review
cratic performance. We briefly examined
62 (December): 777–795.
the information-processing approach, based in Arnold, R. Douglas. 1990. The Logic of Congressional
bounded rationality, which insists that we Action. Yale University Press.
focus both on the complex and uncertain Balla, Steven J. 1998. Administrative Procedure and
environment of policy-makers and on their Political Control of the Bureaucracy. The American
cognitive capacities that channel how they Political Science Review 92 (3): 663–673.
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 71


Balla, Steven J., and John R. Wright. 2001. Interest of SEC. Paper read at The Annual Meeting of the
Groups, Advisory Committees, and Congressional American Political Science Association, August
Control of the Bureaucracy. American Journal of 30–September 2, 2001, at San Francisco, CA.
Political Science 45 (4): 799–812. Cohen, Michael, James G. March, and Johan P.
Banks, Jeffrey S., and Barry R. Weingast. 1992. Olsen. 1972. A Garbage Can Theory of Organi-
The Political Control of Bureaucracies Under zational Choice. Administrative Science Quarterly
Assymetric Information. American Journal of 17: 1–25.
Political Science 36 (2): 509–524. Davis, Otto A., Murray A. H. Dempster, and Aaron
Baumgartner, Frank R. and Bryan D. Jones. 1993. Wildavsky. 1966. A Theory of Budget Process.
Agendas and Instability in American Politics. American Political Science Review 60: 529–547.
Chicago: University of Chicago Press. Downs, Anthony. 1957. An Economic Theory of
Bendor, Jonathan, and Terry M. Moe. 1985. An Democracy. New York: Harper.
Adaptive Model of Bureaucratic Politics. The Downs, Anthony. 1967. Inside Bureaucracy. Edited
American Political Science Review 79 (3): 755–774. by A. R. C. R. Study. New York: Harper-Collins.
Bendor, Jonathan, Serge Taylor and Roland Van Eisner, Marc, Jeff Worsham, and Evan Ringquist. 1996.
Gaalen. 1985. Bureaucratic Expertise Versus Crossing the Organizational Void: The Limits of
Legislative Authority: A Model of Deception and Agency Theory in the Analysis of Regulatory
Monitoring in Budgeting. The American Political Control. Governance (October).
Science Review 79 (4): 1041–1060. Frey, Bruno. 1993. ‘Does Monitoring Increase Work
Bendor, Jonathan, Serge Taylor, and Roland Van Effort? The Rivalry with Trust and Loyalty.’
Gaalen. 1987a. Politicians, Bureaucrats, and Economic Inquiry 31: 663–670.
Asymmetric Information. American Journal of Friedman, Milton. 1953. ‘The Methodology of
Political Science 31 (4): 796–828. Positive Economics.’ In Milton Friedman, Essays
Bendor, Jonathan, Serge Taylor, and Roland Van in Positive Economics. Chicago: The University of
Gaalen. 1987b. Stacking the Deck: Bureaucratic Chicago Press.
Missions and Policy Design. The American Fudenberg, Drew and David Levine. 1997. ‘Measur-
Political Science Review 81 (3): 873–896. ing players’ losses in experimental games.’
Bendor, Jonathan, Glazer, A, and T. Hammond. Quarterly Journal of Economics 112 (2): 507–536.
2001. ‘Theories of Delegation. In Nelson Polsby, Garrett, Elizabeth. ‘The Law and Economics of
ed., Annual Review of Political Science. Palo Alto, Elections: The Law and Economics of ‘Informed
CA: Annual Reviews. Voter’ Ballot Notations’ (Symposium). Virginia
Berejekian, Jeffrey D. 2002. ‘Model Building with Law Review 85 (November 2003): 1553.
Prospect Theory: A Cognitive Approach to Inter- Gigerenzer, Gerd, Peter Todd, and the ABC Research
national Relations.’ Political Psychology 23: Group. 1999. Simple HeuristicsThat Make Us
759–786. Smart. Oxford University Press.
Brehm, John, and Scott Gates. 1993. ‘Donut Shops and Goeree, Jacob K. and Charles Holt. 2000. ‘Asymetric
Speed Traps: Evaluating Models of Supervision on inequality aversion and noisy behavior in
Police Behavior.’ American Journal of Political Science alternating-offer bargaining games,’ European
37: 555–81. Economic Review 44: 1079–1089.
Brehm, John, and Scott Gates. 1997. Working, Golden, Marissa Martino. 2000. What Motivates
Shirking, and Sabotage. Ann Arbor: University of Bureaucrats? Politics and Administration During
Michigan Press. the Reagan Years. New York: Columbia University
Brehm, John, and Scott Gates. 1997. Working, Press.
Shirking, and Sabotage: Bureaucratic Response to a Hardin, Garrett. 1968. ‘The Tragedy of the
Democratic Public. Ann Arbor: University of Commons.’ Science 162: 1243–1248.
Michigan Press. Henrich, Joseph. 2000. ‘Does Culture Matter in
Camerer, Colin and Richard Thaler. 1995. ‘Anomalies: Economic Behavior? Ultimatum Game Bargaining
Ultimatums, Dictators and Manners.’ Journal of among the Machiguenga of the Peruvian Amazon.’
Economic Perspectives 9: 209–219. The American Economic Review 90: 973–980.
Carpenter, Daniel P. 1996. Adaptive Signal Hindera, John J., and Cheryl D. Young. 1998.
Processing, Hierarchy, and Budgetary Control in Representative Bureaucracy: The Theoretical
Federal Regulation. The American Political Science Implications of Statistical Interaction. Political
Review 90 (2): 283–302. Science Quarterly 51 (3): 655–671.
Choi, Jin-Wook. 2001. Rethinking Political Control Hoffman, Elizabeth, and Mathew L. Spitzer. 1985.
Over Independent Regulatory Agencies: The Case ‘Entitlements, Rights and Fairness: An Experimental
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 72


Examination of Subjects’ Concepts of Distributive Decisions, 1949–97. American Political Science

Justice.’ Journal of Legal Studies 14: 259–97. Review 97 (1): 171–188.
John, Peter. 2003. ‘Is There Life After Policy Streams, Levi, Margaret. 1997. ‘A Model, A Method, and a
Advocacy Coalitions, and Punctuations? Using Map: Rational Choice in Comparative and
Evolutionary Theory to Explain Policy Change.’ Historical Analysis’. In Comparative Politics:
Policy Studies Journal 31 (4): 481–498. Rationality, Culture, and Structure, Mark Lichbach
Jones, Bryan D. 1994. Reconceiving Decision Making and Alan Zuckerman, eds, pp. 19–41. New York:
in Democratic Politics: Attention, Choice, and Cambridge University Press.
Public Policy. Chicago: The University of Chicago Lupia, Arthur, Mathew McCubbins, and Samuel
Press. L. Popkin, eds., Elements of Reason. Cambridge:
Jones, Bryan D. 2001. Politics and the Architecture of Cambridge University Press.
Choice: Bounded Rationality and Governance. Lupia, Arthur, Mathew D. McCubbins, and Samuel
Chicago: The University of Chicago Press. L. Popkin. 2000. ‘Beyond Rationality: Reason and
Jones, Bryan D. 2003. Bounded Rationality and the Study of Politics’. In Lupia, Arthur, Mathew D.
Political Science: Lessons from Public Admi- McCubbins, and Samuel L. Popkin, eds., Elements
nistration and Public Policy. Journal of Public of Reason. Cambridge: Cambridge University
Administration Research and Theory 13 (4): 395–412. Press.
Jones, Bryan D. and Frank R. Baumgartner 2005. Lipsky, Michael. 1980. Street-Level Bureaucracy:
The Politics of Attention: How the Government Dilemmas of the Individual in Public Services.
Prioritizes Problems. Chicago: University of New York: Russell Sage Foundation.
Chicago Press. March, James G., and Herbert A. Simon. 1958.
Jones, Bryan D. and Frank R. Baumgartner. 2005. The Organizations. New York: Free Press.
Politics of Attention: How the Government Prioritizes Mayhew, David. 1974. Congress: The Electoral
Problems. Chicago: University of Chicago Press. Connection. New Haven: Yale University Press.
Jones, Bryan D., Tracy Sulkin, and Heather A. McCubbins, Matthew D. 1985. The Legislative
Larsen. 2003. Policy Punctuations in American Design of Regulatory Structure. American Journal
Political Institutions. American Political Science of Political Science 29 (4): 721–748.
Review 97 (1): 151–169. McCubbins, Matthew D., Roger G. Noll, and Barry
Kahneman, Daniel and Amos Tversky. 1979. R. Weingast. 1989. Structure and Process: Politics
‘Prospect theory: An analysis of decisions under and Policy: Administrative Arrangements and the
risk.’ Econometrica 47: 313–327. Political Control of Agencies. Virginia Law
Kahneman, Daniel and Amos Tversky. 1986. ‘Rational Review 75: 431.
Choice and the Framing of Decisions.’ In Daniel Meier, Kenneth J., Michael S. Pennington, and
Kahneman and Amos Tversky, eds. 2000. Choices, Warren S. Eller. 2001. The Effect of New Goals
Values and Frames. Cambridge University Press. and Priorities on Agency Responsiveness to
Kahneman, Daniel and Amos Tversky, eds. 2000. Traditional Clientele: An Examination of the
Choices, Values and Frames. Cambridge University EEOC. Paper read at The Annual Meeting of the
Press. American Political Science Association, August
Kahneman, Daniel. 2002. ‘Maps of Bounded 29–September 2, 2001, at San Francisco, CA.
Rationality: A Perspective on Intuitive Judgement Miller, Gary J., and Terry M. Moe. 1983a. Bureau-
and Choice.’ Nobel Prize Lecture, December 8th, crats, Legislators, and the Size of Government.
2002. American Political Science Review 77 (June):
Kahneman, Daniel, Jack Knetsch, and Richard H. 297–322.
Thaler. 1986. ‘Fairness and the Assumptions of Miller, Gary J., and Terry M. Moe. 1983b. The
Economics’ Journal of Business 59: S283–S300. Positive Theory of Hierarchies. Paper read at The
Kingdon, John. 1995. Agendas, Alternatives, and Annual Meeting of the American Political Science
Public Policies, 2nd. Ed. Boston: Little, Brown. Association, at Chicago.
Kingdon, John. 1973. Congressmen’s Voting Decisions. Miller, Gary. 1992. Managerial Dilemmas: The
New York: Harper & Row. Political Economy of Hierarchy. Cambridge:
Krause, George A. 1996. The Institutional Dynamics Cambridge University Press.
of Policy Administration: Bureaucratic Influence Mitnick, Barry M. 1973. Fiduciary Rationality and
Over Securities Regulation. American Journal of Public Policy: The Theory of Agency and Some
Political Science 40 (4): 1083–1121. Consequences. Paper read at The Annual Meeting
Krause, George A. 2003. Coping with Uncertainty: of the American Political Science Association, at
Analyzing Risk Propensities of SEC Budgetary New Orleans, LA.
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 73


Mitnick, Barry M. 1980. The Political Economy of Awards. Paper read at The Annual Meeting of the
Regulation. New York: Columbia University Press. American Political Science Association, August
Moe, Terry M. 1984. The New Economics of 30–September 2, 2001, at San Francisco, CA.
Organization. American Journal of Political Quattrone, George and Amos Tversky. 1988.
Science 28 (4): 739–777. ‘Contrasting Rational and Psychological Analyses
Monroe, Kristen R. 1996. The Heart of Altruism: of Political Choice,’ American Political Science
Perceptions of a Common Humanity. Princeton Review 82: 719–736.
University Press. Rockman, Bert. 2000. ‘Theory and Influence in the
Muringhan, Keith J. and Michael S. Saxon. 1994. Study of Bureaucracy: Micro- and Neoinstitutional
‘Ultimatum Bargaining by Children and Adults,’ Foundations of Choice.’ Journal of Public
unpublished working paper, University of British Administration Research and Theory 11 (1): 3–27.
Columbia Faculty of Commerce. (Cited in Thaler Roth, Alvn, Vesna Prasknikar, Masahiro Okuno-
1992). Fujiwara and Shmuel Zamir. 1991. ‘Bargaining
Muringhan, Keith J. and Madan Pillutla. 1995. and Market Behavior in Jerusalem Ljubljana,
‘Being Fair or Appearing Fair: Strategic Behavior Pittsburgh, and Tokyo: An experimental study.’
in Ultimatum Bargaining.’ Academy of Manage- American Economic Review 81: 1068–95.
ment Journal 38: 1408–1427. Simon, Herbert A. 1947. Administrative Behavior: A
Newell, Allen. 1990. Unified Theories of Cognition. Study of Decision-Making Processes in Adminis-
Cambridge, MA: Harvard University Press. trative Organizations. 4th ed. New York: Simon
Niskanen, William A. 1968. The Peculiar Economics and Schuster, Inc.
of Bureaucracy. Papers and Proceedings of the Simon, Herbert A. 1985. ‘Human Nature in Politics:
Eightieth Annuel Meeting of the American Economic The Dialogue of Psychology with Political
Association 58 (2): 293–305. Science.’ The American Political Science Review
Niskanen, William A. 1971. Bureaucracy and Repre- 79: 293–304.
sentative Government. Chicago: Aldine-Atherton, Simon, Herbert A. 1991. Models of My Life.
Inc. New York: Basic Books.
Niskanen, William A. 1975. Bureaucrats and Simon H.A. 1999. The potlatch between political
Politicians. Journal of Law and Economics 18: science and economics. In Competition and
617–644. Cooperation: Conversations with Nobelistsabout
Nowak, Martin A., Karen Page and Karl Sigmund. Economics and Political Science, J Alt, M Levi,
2000. ‘Fairness versus Reason in the Ultimatum E Ostrom, eds. Cambridge, UK: Cambridge
Game’ Science 289: 1773. University Press.
Olson, Mancur. 1982. The Rise and Decline of Slovak, Paul. 1990. Choice. In Thinking: An invita-
Nations. Yale University Press. tion to cognitive science, vol. 3. Daniel Osherson
Osterloh, Margeret and Bruno Frey. 2000. ‘Motivation, and Edward E. Smith, eds. Cambridge, Mass: MIT
Knowledge Transfer, and Organizational Forms.’ Press.
Organization Science 11: 538–550. Sniderman, Paul, et al. 1991. Reasoning and Choice.
Ostrom, Elinor. 1998. ‘A Behavioral Approach to the Cambridge: Cambridge University Press.
Rational Choice Theory of Collective Action’. Solnick, Sara. 2001. ‘Gender Differences in the
American Political Science Review 92: 1–22. Ultimatum Game.’ Economic Inquiry 39: 189.
Ostrom, Elinor. 1999a. ‘Coping with Tragedies of the Taylor, Frederick W. 1911. The Principles of Scientific
Commons’. In Nelson Polsby, ed., Annual Review of Management. New York: W. W. Norton.
Political Science 2. Palo Alto, CA: Annual Reviews. Tetlock, Philip. 2000. ‘Coping with Trade-offs:
Ostrom, Elinor. 1999b. ‘Institutional Rational Choice: Psychological Constraints and Political Impli-
An Assessment of the Institutional Analysis and cations’. In Lupia, Arthur, Mathew D. McCubbins,
Development Framework.’ In Theories of the and Samuel L. Popkin, eds., Elements of Reason.
Policy Process, Paul A. Sabatier, ed. Boulder, CO: Cambridge: Cambridge University Press.
Westview. Thaler, Richard H. 1992. The Winner’s Curse;
Padgett, John F. 1980. Bounded Rationality in Paradoxes and Anomalies of Economic Life.
Budgetary Research. American Political Science Princeton University Press.
Review 74: 354–372. Thompson, Fred and Mark Green. 2001.
Potoski, Matthew, and Christine L. Nemacheck. Organizational Process Models of Budgeting,
2001. Bureaucratic Influence in the Legislative Research in Public Administration, 55–81.
Arena: 1984 Drug Price Competition-Patent True James L., 2000. ‘Avalanches and Incre-
Restoration Act and FDA Distributive Program mentalism: Making Policy and Budgets in the
Peter-3383-Chapter-03.qxd 6/8/2006 7:32 PM Page 74


United States.’ American Review of Public Wood, B. Dan. 1988. Principals, Bureaucrats, and
Administration 30: 3–18. Responsiveness in Clean Air Enforcements.
Waterman, Richard and Kenneth Meier. 1998. American Political Science Review 82: 213–234.
Principal-agent Models: An Expansion? Journal Wood, B. Dan, and Richard W. Waterman. 1991. The
of Public Administration Research and Theory Dynamics of Political Control of the Bureaucracy.
2: 173–202. The American Political Science Review 85 (3):
Weingast, Barry R., and Mark J. Moran. 1983. 801–828.
Bureaucratic Discretion or Congressional Control?: Worsham, Jeff, Marc Allen Eisner, and Evan J.
Regulatory Policymaking by the Federal Trade Com- Ringquist. 1997. Assessing the Assumptions: A
mission. Journal of Political Economy 91: 765–800. Critical Analysis of Agency Theory. Administration
Wildavsky, Aaron. 1964. The Politics of the Budgetary and Society 28 (4): 419–440.
Process. Boston: Little, Brown. Worsham, Jeff, Evan Rinquist, and Marc Eisner, eds.
Wilson, James Q. 1989. Bureaucracy: What Govern- 1998. A Theory of Political Influence of the
ment Agencies Do Why They Do It. New York: Bureaucracy. Edited by J. D. White, Research in
Basic Books, Inc. Public Administration. Stamford, CT: JAI Press.
Wilson, Woodrow. 1887. The Study of Adminis- Zaller, John R. 1992. The Nature and Origins of Mass
tration. Political Science Quarterly 2: 197–222. Opinion. Cambridge University Press.
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 75

Policy Design: Ubiquitous,
Necessary and Difficult

D AV I S B . B O B R O W

INTRODUCTION professional association and journal; standards

for certification; broader social attribution of
The familiar function of a handbook for a special expertise; consensually shared views of
professional community is to provide a survey core foundation knowledge; or a widely
of the state of the art in some pertinent aspect accepted program for improving capabilities.
of professional practice. A handbook typically Policy design merits attention, nevertheless,
identifies the major instruments of a profes- because it is ubiquitous, necessary, and difficult.
sional repertoire, and assesses their strengths For better or ill, it has and does go on and is
and weaknesses to suggest best practices. In explicitly so labeled for public policies in areas
doing so, it cites known regularities or pat- as diverse as: foreign affairs (e.g., Hoag, 1976);
terns, established theories, and well-tested trade policy (e.g., Brainard and Martimort,
techniques. That is usually done with the 1996); international exchange rate coordination
explicit or tacit claim that they are either the (e.g., Cohen and Wyplosz, 1995); industrial
specialized property of the field or used espe- policy (e.g., Cody Kitchen and Weiss, 1990);
cially well by its members. environmental quality (e.g., Pellikan and van
It is hard to argue that policy design has der Veen, 2002); food stamps (e.g., Ohls and
much in the way of such possessions other Beebout, 1993); national information systems
than borrowing from other specializations and (e.g., Laudon, 1986); government spurred inno-
often relaxing their standards for advanced vation (e.g., Roessner, 1988); International
professional practice. Why then does policy Monetary Fund programs (e.g., Killick, 1995);
design warrant inclusion in a handbook on managed health care (e.g., Hurley, Freund, and
public policy? Paul, 1993); child health care (e.g., Goggin,
One possible answer is a sociology of the 1987); health insurance (e.g., Oliver, 1999);
profession’s presence of a more or less orga- internal migration (e.g., Castro et al., 1978);
nized group of specialist experts. Yet, unlike macroeconomic policy (e.g., Taylor, 1993);
policy analysis, policy design shows few of homeland security (e.g., Demchak, 2002); and
the trappings of a professional community: democratic governance (e.g., Schneider and
self-identification as a policy designer; a Ingram, 1997).
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 76


Further, the instances of work explicitly both reasons is the notion that engineering,
labeled policy design are far outnumbered by unlike architecture, denies the possibilities for
those in which designs are prescribed or evalu- unwanted responses by those who must go along
ated without bearing that label in a prominent for the design to materialize and work as envi-
and explicit way. With or without a policy sioned by its engineers. Such an assumption of
design title, policy assessments by lay people a lack of voice, exit options, resistance strate-
and “policy wonks” applaud or bemoan the gies, and bargaining power is of course more
consequences attributed to such efforts. That is often than not extremely unrealistic (Stewart
no less the case for international concerns than and Ayres, 2001).
for the dominant focus of the literature, quint- Under either metaphor, designs are represen-
essential domestic policy. We are, of course, tations of sets of choices and instructions to select
well into a world in which designs which and apply materials sufficient to produce an
assume that consequences, causes, and policy intended result. ‘Designers put things together
options can be neatly segregated by national and bring new things into being, dealing in the
borders are doomed to disappointment. process with many variables and constraints’
Subsequent sections of this chapter address: (Schon, 1987: 42). Architecture of course draws
policy design as a messy imperative; root con- on available scientific knowledge, but does so in
ceptions of policy design; subsequent elabora- the pursuit of normative standards and makes
tions and differentiations of its conception; and use of a lot of intuitive craft and creativity. Policy
core strategies for its conduct – a suggested ten designers do so as well, and the quality of their
commandments. endeavors does not depend only on the technical
Before turning to those matters, it seems robustness of the “policy sciences” (Lerner and
helpful to recognize the core metaphors for Lasswell, 1951; Dror, 1971).
policy design, and the core activities of policy A grossly over-simplified skeleton which a
designers they imply. As an initial orientation stereotypic designer of policies qua engineer
for subsequent modification, central metaphors might flesh out in a representation appears in
are those of architectural drawings or engineer- Figure 4.1. Note it’s linear, mechanical, and
ing blueprints, i.e., the products of the design non-reflective appearance.
sciences (Perlmutter, 1965; Alexander, 1964; A policy design is stipulated in the expecta-
Alexander, 1982). Policy designs are representa- tion that adopting it will produce particular
tions of what might be turned into realities. The patterns of human or organizational activity
analogy with architecture and architects should and chains of consequences. Individually, or in
be taken seriously, if only because it appears combination, they will suffice to achieve cer-
again and again (e.g., Schon, 1983; Weimer, tain predicted desirable changes in targets
1992: 136; Meier and Smith, 1994: 440). (intermediate products) which as a set will
Architectural metaphors have in many quar- provide the necessary and sufficient conditions
ters gained favor over the engineering ones for for a valued outcome.
several reasons. One is the view of ‘social engi- This skeleton of course raises more questions
neering’ as inappropriately and inherently a than it answers. Why bother? Can we flesh it out
matter of arrogant command by those in adequately if at all? What ingredients go into a
control of positions of power, be they in policy? How do we specify a valued outcome?
Washington, Moscow, or Brussels, i.e., anti- Who are relevant persons, organizations, and
democratic. A second is the not necessarily collectivities? What are the pertinent human
warranted view of it as a failure in practice, and organizational actions and socially relevant
e.g., some constructions of what has happened chains of consequences? What determines
with efforts to engineer a welfare state in the them? What policies will have what effects on
U.S. and elsewhere. Social engineers are not as those determinants? What are the targets to be
wise or smart as they and those who employ achieved? What human actions and chains will
them might like to think about picking do that? What determines if those in a position
winners and rejecting losers. Related to to decide to pick a particular representation will
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 77


Policy design

Changes in human and Changes in chains of

organizational activity consequences

Changes in targets leading

to intermediate products

Changes in valued outcomes

Figure 4.1 An initial visualization

Source: Modified from Dorfman, 1986: 108.

choose a particular one? Even if they do, will the prospective occurrence of unwelcome eventu-
sketch or blueprint be followed as building or alities, and the failure to materialize of attrac-
implementation proceed? What body of theory, tive ones we expected. We look to ruling elites
strategies, techniques, procedures and processes to design or at least adopt designs for a desir-
will enable us to answer these and other ques- able future and invent or at least implement
tions in ways which are consistent, ethically ways of bringing it about. We may well reward
appropriate, and pragmatically feasible? There and punish them accordingly within our
are then prima-facie grounds for doubt about understanding of their responsibilities and
the normative and pragmatic status of the policy possibilities. On their part, ruling elites
approach in the strawman of Figure 4.1. Those usually try to anticipate what will trigger par-
are far from sufficient grounds for turning one’s ticular punishments and rewards. The market
back on policy design. for ways to pursue preferred or at least tolera-
ble futures never closes with the attendant
incentives for purveyors to enter and stay in
that market. Applied social scientists have long
been among those purveyors, or at least staffed
them and tried to influence them and their
We often are far from indifferent about our per- customers.
sonal and collective futures as individuals, When those ways involve more or less
members of social groupings, and citizens of explicit ensembles of material and non-
political entities. We act, and delegate to others material ingredients, context conditions, and
the responsibility to act and induce us to act, processes, we can think of them as designs.
in ways which hold normative and pragmatic They can appropriately be thought of as public
promise in our eyes. That promise lies in policy designs when they involve government
forestalling, or limiting the damage potentially units or officials acting directly, or as a target
posed by, distressing outcomes and achieving or of or motivator for actions by others. In a
at least improving the odds of appealing ones. common-language sense, designs are recom-
We tend to hold those in positions of collec- mended plans, but ones not at all necessarily
tive responsibility accountable for the actual or involving dirigiste central planning.
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 78


Public policy design is ubiquitous since “way”, and never been labeled policy design. In
organized social and political action always the contemporary world, the difference is
involves ingredients, always has a context, always massive broadening in the range of others and
contains processes, and in modern society phenomena whose impacts affect us, accelera-
always involves some aspect of governance. That tion in the speed with which they do so, and near
is no less true for those who believe in a benign instant information about and media reporting
or unavoidable, “invisible hand” economic or of ostensibly outcome relevant developments.
theological. For them, policy designing is still Those secular trends increase the incentives
warranted if only to let the invisible hand come to engage in policy design while making the
closer to functioning at its ostensible best rather endeavor more difficult. The difficulties
than its imaginable worst. Policy design involve: 1) greater complexity in terms of the
attempts are not the monopoly of those who number and variety of actors and factors;
buy into any one particular notion of the role of 2) rapid changes raising the risks of designs
the state or embrace any particular view of being rendered obsolete or seriously flawed
desirable and just distributions of power, before they are finished; and 3) extreme vul-
wealth, status or rectitude. nerability to a loss of support due to declines
Policy design then claims our attention if two in issue priority and “from the get-go” expo-
conditions are met. The first is that we are not sure to organized opposition.
indifferent to ‘who gets what, when, and how’ The feasibility of relatively attractive paths
(Lasswell, 1961), a range of concern which into the future seems then to depend in increas-
includes processes, and aggregate and distribu- ingly important ways on interdependencies with
tional outcomes. The second is that we acknowl- and sensitivities to others and on contextual
edge the existence of more than one possible factors. From this perspective, policy design is
positively or negatively valued state of affairs in necessary for timely anticipation of and accom-
some slice of the future, rather than granting modation to what we cannot control, as well as
inevitability to a particular prospect (Dror, 1986). for making the most of what we can control. A
Defeating prophets of inevitability, e.g., about ship’s captain must steer in the face of expected
inequality, may be facilitated by showing that a or unexpected bad weather just as he or she must
present state of affairs was not itself inevitable but in a benign sea. The wise captain, naval architect
rather results from a set of previous actions, be and engineer tries to have built in resilience
they ones of commission and omission (e.g., to shocking circumstances imaginable and
Fischer et al., 1996). Those actions could follow unknowable (Demchak, 2002). Policy design
from an integrated, purposeful plan (deliberate tries to help “captains” and citizens by providing
intent) or merely follow from unintended second courses of action and material aids which can
or third order effects from foreseeable or unfore- be used and retain their efficacy in ‘turbulent
seeable relevant events. environments’ (Ackoff, 1979).
In sum, when both conditions seem to be met, Policy design also is necessary for anticipa-
the responsible course becomes to try to ‘tilt the tory policy analysis. Policy analysis broadly
odds’ toward more preferable outcomes. conceived is fundamentally the “scoring” of
Preferences can be for a previously unexperi- possible behaviors in terms of value criteria.
enced future, maintenance of the status quo, or Policy analysts are in search of at least one set
even resurrection of an ostensibly attractive past. of behaviors which seems likely to be at least
They, alternatively, can be to forestall some envi- satisfactory or ideally optimal. That quest is
sioned future, erode or shatter the status quo, or hostage to: the possible behaviors we envision;
forestall reversion to some past. Policy elites and the foresight with which we specify sequences
publics have then little choice but to frame and and quantities and qualities of behaviors; and
choose among alternative policy designs. That is the extent to which we encompass the range of
hardly new, even if it may in pre-industrial soci- considerations and priorities of outcomes
eties have emphasized an accommodation to and their associated valuations. Anticipatory
spirits and nature, as with the Navajo Indian policy analysis can be no better than the policy
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 79


designs it considers. Unfortunately, the necessity and intuition, determines what behaviors on
of policy design may often only be the mother the field are desirable, and tries to prepare and
of invention and not of success. Like even the persuade players (as agents) to make them.
best batters in baseball or strikers in soccer, the Like even the wisest coach, a policy designer
harsh statistical reality is that the vast majority has to live with the possibility of the actual
of the time our best efforts will not achieve a players putting advice into action well, poorly,
hit or a goal. Even when they do, that is no or not at all. The designer can, however, be
guarantee of future successful performance by more or less competent at fitting strategies
repeating the previously successful moves and techniques to the available players, the
(counter to Chung, 1993). nature of the opposition, field conditions, and
Many of the inherent difficulties are well the time remaining in a game, series, or season.
known – the lack of stable and fully revealed And, of course, like changing coaches, firing
goal preferences; limited understanding of and replacing designers may be a more com-
causality; at best partial knowledge of the mon response to disappointment than chang-
future; counter-measures deliberate and unin- ing those who adopt or implement policy
tentional of unanticipated vigor, skill and type; designs.
the presence of residual designs, institutional Second, policy design is often a competitive
interests, legacies, and memories rather than a sport. Experts compete for the chance to
clean slate; and hard to fathom complexity. design policies. If they get it, their designs are
Policy design is a “game of failure” and disap- pitted against alternatives. When designs go
pointment, and thus not for the easily discour- into the white water of the policy process they
aged or those who do not find the act of are placed in the hands of individuals who may
participation inherently gratifying compared well be competing for personal success, and of
to only being a spectator. teams (e.g., political parties and bureaus) who
That is not to say that technique and strat- compete to improve or at least do as well for
egy do not matter for achievement, that all fail- themselves as they have before. Ultimately,
ures must be repeated, or that an unsuccessful designs in practice and the political entities
specific effort cannot set the stage for subse- who have put them more or less into practice
quent relative success. At least in the long-run, are competing against benchmarks used by
the odds for both positive achievements and relevant audiences.
avoiding failures can be improved, if only at The policy designer has reason to take into
the margin. Since policy design is a game that account the benchmarking criteria at work and
will be played, it makes sense to seek ways to their interpretation, while recognizing his or
play more effectively. What those are and the her limited (but perhaps not non-existent)
capacity to put them into use will of course capacity to stipulate them. Flying in the face of
vary. Some strategies and techniques may have prevalent benchmarks and interpretations
merit almost generically and others only for increases the chances that designs will be
particular situational factors and available ignored, rejected, or distorted in implementa-
resources. An implication is that policy design- tion. Even if fulfilled, a design may be deprived
ers need to be adept at generically useful skills of the status of a model for emulation in other
and situational diagnosis, command a broad jurisdictions or time periods. In effect a policy
repertoire to draw on for particular situations, design provides a promise, albeit a hedged one.
and be realistic about available resources. The promise competes with held images of the
Some final prefatory implications of a team past, present, and future to elicit “gambles” or
sport metaphor are worth noting as they bear bets from elites and general populations.
on policy design. First, while the designer is a Further, there is no guarantee that: a) the
player at the stage of suggesting what to do, he design teams and their employers or sponsors
or she is most of the time not in a position to are all playing the same game, or b) if they
adopt or implement policy designs. The role is are, accept the same rules, or c) if they do,
more that of a coach who, through analysis accept the same measures and standards of
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 80


performance. This lack of homogeneity can conditions … an economy must possess certain
lead to “design wars” between experts rooted in basic characteristics … and some government
different disciplines, ideologies, and intellec- activity.’ A government must have at hand
tual traditions (see Bobrow and Dryzek, 1987; ‘a minimum of instruments of economic
Linder and Peters, 1992). It can also lead in the policy … and these must be properly used.’
world outside the design studio, think-tank, or Because the goals or values sought are never in
policy planning bureau to profound disap- reality singular but always multiple (a multi-
pointments, surprises, and conflicts. attribute utility function), the instruments
More encouragingly, a policy design and the should be multiple as well.
activities associated with designing need not Beyond those basics, design is informed by an
move fully from paper or hard drive into prac- ‘awareness of … [goal related] … potentialities
tice, or fully achieve their promise, to provide and advantages’ derived from descriptions of
benefits. Efforts at designing or applying a the past and present, and projections. It also
policy can raise performance norms for policy should be informed by awareness of ‘varying
content and process. Those experiences can circumstances. Depending on circumstances
enlighten us about what does not work as some elements of … policy will require more
desired, and focus us on vulnerabilities for emphasis and attention or will appear more or
future amelioration. More arguably, they can less promising than others.’ Further, the action
leave satisfaction with the benefits of the elements of designs (‘programs’ and their com-
endeavor (“a game well-played”) even if one’s ponent ‘projects’) ‘by their very nature have to
favorite loses and thus maintain or increase be guesses and must be revised periodically.’
enthusiasm for trying again. If policy design is (Tinbergen, 1958: 4–7). In sum, Tinbergen
ubiquitous and necessary, it makes sense to called attention to the starting points of uni-
think of it as being played not only in a long versal requirements or necessary conditions,
season, but for many seasons. government policy instruments as means,
recognition of alternative possible outcomes,
situational diagnosis, uncertainty, and sufficient
flexibility for revision in light of experience.
While Tinbergen did emphasize economic
matters and governmental instruments, he
More than four decades ago, Jan Tinbergen made it quite clear that a far broader set of
(1958) addressed policy design. His specific actors and factors should be considered in
concern was with development, which can be policy design. These include the private sector,
thought of more broadly as progress toward a general population attitudes, the administra-
preferred but yet to be realized collective future. tive organs of government with their varied
His perspective continues to echo in subsequent proclivities and capabilities, current economic
conceptions of the motivation for and funda- structure, natural resources, geographic situa-
mental characteristics of policy design. He tion, particularly strong personalities, the size
advocated a turn to design as a replacement for of the entity involved, and its sensitivity to
‘decisions … taken on the basis of vague ideas of behavior external to it. Policy design then is the
general progress, and often somewhat haphaz- ‘solution of a jig-saw puzzle of considerable
ardly … a process of trial and error … [marked complexity.’ (Tinbergen, 1958: 35)
by] … setbacks and crises, and probably a good In light of these considerations, the challenges
deal of misplaced energy and effort …’ to the policy designer are to devise a set of steps
(Tinbergen, 1958: 3). Policy design effort fol- which together make up a ‘coherent and coordi-
lows from dissatisfaction with a record of nated whole.’ That, in turn, calls for avoiding the
behavior including its efficiency. pitfalls of sole reliance on quantitative data, wish-
Motivated by that dissatisfaction, the policy ful thinking about available resources, omission
designer builds on awareness that ultimate, of interactive effects, and inattentiveness to time-
broad goals require achieving ‘certain general lags and sequences of action. Policy designs are
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 81


working drafts to be scrutinized and modified, The Tinbergian perspective surely can
based on eclectic information, internal consis- encompass many more recently offered concep-
tency, completeness, and the direct and primary tions. Herbert Simon’s view of design as the
as well as indirect and secondary consequences process of ‘changing existing situations into pre-
of their constituent programs and projects ferred ones’ (Simon, 1972: 55) becomes com-
(Tinbergen, 1958: 9–33). Given the importance patible when one invokes his conception of
of what will not be known with certainty, the satisficing. If a current situation, and current
responsible designer will consider offering prospects for the future, are satisfactory there
‘alternative programs,’ each with a somewhat need be only limited interest in design. If
different base of assumptions. Failures will still they are not, designers need quest after only
result because of factors such as the ignorance of what would make them satisfactory, and not
politicians, pressure from vested interests, minis- after what would make them optimum.
terial rivalries, incentives offered by legislatures, Reconciliation for inclusion is not even required
and external interference. for conceptions such as those of Linder and
Policy designers should be modest about the Peters (1988) ‘the purposive or goal-directed
‘role to be played by scientific knowledge and rearrangement of a problem’s manipulable fea-
insight … The relevant facts of life are too many tures’; May’s (1981) ‘identification and mani-
and too varied to make it possible to reach pulation of key design variables to create viable
decisions without a strong intuitive feeling for alternatives’; or Wildavsky’s art of finding solu-
human relations’ (Tinbergen, 1958: 68–69). tions to policy problems that specify desirable
Policy designers engage more in a craft than a relationships between ‘manipulable means and
science. Policy design a la Tinbergen stands in attainable objectives’ (Wildavsky, 1979: 15–16 in
contrast to ‘an interventionist perspective, Weimer, 1993: 111). Tinbergen’s recognition of
requiring a precise forecast of events so an situational variety is a harbinger of the empha-
external hand can intervene to assure meeting sis on it in ‘case-wise’ prescriptions for policy
of the exogenously determined targets’ (Saeed, analysis and design (Bruner, 1986, 1983).
1994: 11). Nevertheless, Tinbergen’s perspective, wise as
it was and is, leaves open a wide and difficult
range of issues about the practice of policy
design. Some of these are about steps in the
creation of a meritorious design. Others are
about what its contents should be. Still others are
Since Tinbergen’s volume appeared, much has about the intellectual sources of or approaches
been written in several intellectual communi- to providing those contents. Taken to exclusive
ties about what policy design should and can extremes, the different positions in these debates
be. In large measure, these contributions amount to different conceptions of policy
amount to elaborating and differentiating ele- design (for comparisons of some of the major
ments explicit or implicit in his perspective. contenders, see Bobrow and Dryzek, 1987;
Elaboration and differentiation do not Linder and Peters, 1992). Yet we need not, and
necessarily carry with them implications of should not, treat them as either/or choices.
universal superiority, or exclusive, most quin- Continued contention about what aspects of
tessential status. They can instead suggest the the policy design elephant to emphasize is then
opposite – a broadening of the scope of policy about claims to shares in a mix, rather than
design rather than a narrowing. After Schon’s something to be resolved by giving the mix a
‘reflective practitioner’ who engages in a con- more homogenous, less multi-faceted character.
versation with a design situation (Schon, 1983, Those claims or suggestions may contribute to
1987), the elaborations and differentiations an ‘armamentarium’ of ‘frames with which to
may as a whole suggest an enriched and more envisage’ coherent designs and ‘tools with
varied, rather than a more circumscribed and which to impose … [designer’s] … images on
narrowly focused conversation. situations’ (Schon, 1987: 218).
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 82


This position, applied to political entities and and negative outcomes (issues of representation)
not just economic ones and to designs for out- and the quantity and distribution of a varity of
comes not limited to economic ones, adopts benefits (goods) received and costs (bads)
and generalizes the counsel of Adelman and borne (e.g., for indigenous knowledge as intel-
Taylor (1986: 84) ‘Ultimately, appropriate lectual property in Norchi, 2000, or for wealth
forms … need to evolve which can be adapted creation and public health and welfare in
flexibly to economies and societies at vastly dif- Daniere and Takahashi, 1999).
ferent levels of development, having different From another angle, problem definition
objectives and constraints, operating with differ- involves a grasp of causality. What brings
ent information and power structures, and hav- about particular human or organizational
ing differing social norms and values. The actions and chains of consequences? Why do
evolution of such planning technologies requires they produce particular intermediate products
melding a historical and socio-political perspec- or performance on target phenomena? Why do
tive with the tools of operations research in a particular sets of products and performances
much more flexible and synergistic manner …’ produce one or another outcome? With weak
Major elaborations and differentiations and partial answers to these questions, we
modify, almost always by complicating, the simply cannot know with confidence answers
visualization introduced in Figure 4.1. I will to a host of very practical questions. What
first consider two major sorts of elaborations needs to be changed? By how much and how
which are by no means incompatible with each quickly? What will induce those changes with-
other. One proliferates elements of a policy out triggering responses which countervail
design, and thus tasks to be dealt with in the them? Consideration of causality in policy
design process. A second provides maxims, design also raises issues about factors external
wise guidance, for the undertaking of those to a particular policy system which can affect
tasks. Neither sort is the particular property of its behavior, the triggers of the bad weather or
a single technical discipline. Each introduces turbulence to which we referred earlier.
“cascades” of questions to be dealt with. Context altering events and developments
may be unanticipated in their entirety or mag-
nitude. We may be unable to specify their fre-
quency, time of onset, and duration. Consider
the impacts of the collapse of the Soviet Union
on the foreign policy designs of most other
Some of the additional tasks involve problem nations, of a tide of refugees on welfare deliv-
definition (DeLeon, 1994; Schneider and Ingram, ery systems, of assassination or atrocity on
1989). What are we proposing the design to rem- inter-group relations, or of an agricultural
edy? A wrong or incomplete answer will lead to blight on food provision. The “exogeneous”
designs which are social failures, either because cannot be relied on to remain helpfully walled
they deal with a remedy for a different malady off. Yet some possibilities are surely imagin-
altogether or yield an outcome which the osten- able, and thus can be accommodated in a
sible beneficiaries view more as a threat or harm robust design. Others are not, but such possi-
than as a remedy or benefit. bilities can be accommodated at least in part
From this angle, problem definition involves by building in slack and redundancy, even at a
value clarification and identifying dissatisfac- sacrifice of efficiency. Still, other changes can
tions or as yet unmet or endangered desires as only be handled by a quick and thorough
goals (Lasswell, 1971). These are certainly reversion to a reconsideration of problem
matters of moral philosophy – the warranted redefinition (after Demchak, 2002).
claims of the poor, the rich, nationals, foreigners, From a third angle, problem definition
the living, future generations. They call for involves a reasonably accurate description of a
judgments about whose fate should have how current situation, including its probably irre-
much standing in defining what are positive versible trends. That may seem rather mundane,
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 83


but descriptive reality is often not well known or Yet neglect of it flies in the face of the reality of
a matter of consensus. For major policy the attention to presentation and public and
matters, domestic and international, it more media relations which take a large part of the
often is contested ground. And it is fought over time of policy-makers and takers, and of those
or obscured in large measure because our who would influence them.
understandings of it affect what even the best I suggest that adoption, implementation, and
processes of value clarification and causality interpretation usually should be integral parts
determination will find or have found. This of policy design. An architect adapts his design
aspect of problem definition can take a lot of to the proclivities of those who must agree to
hard work. Reflect for a moment on the diffi- buy it, use it, and build it – while trying to influ-
culties of: listing the relevant shapers and shak- ence their notions of what they want and
ers of foreign trade conduct in particular can have. This caveated position accepts that
industries and flows as more than general cate- on occasion it may be useful to formulate a
gories; determining the percentage of welfare design simply as a utopian, aspirations raising
recipients who genuinely would prefer to have endeavor. Far more frequently, adoption or
jobs under varying conditions; estimating the implementation or interpretation failure are
current national costs of mental illness let alone fates to be avoided. What is in, and interpreted
its incidence; or assessing the closeness in time to be in, a Figure 4.1 type design will affect the
(imminence) of any particular country having chances of adoption, and the fidelity of imple-
ready-to-use weapons of mass destruction (as mentation in spirit and letter. That does not
with Iraq, Iran and North Korea). If we don’t argue for taking the design course with the most
know where we are, how can we determine a popular “face” and established odds of adoption
course to move from there to someplace else or and implementation. Doing so could vitiate the
even to maintain the status quo? reason for undertaking the design effort in the
To compound matters further, the problem first place – to improve what would otherwise
may not be an absence of good designs in the be outcomes. It does call for steps in design
compass of Figure 4.1 but lie with some of the which will serve several purposes.
factors and actors which pose themselves The first is to give a design thought to be com-
between the design as representation and the mendable in terms of valued outcomes the best
design in practice. Such obstacles involve possible chance of being realized. The second is
adoption and implementation (Bardach, 1977; to help focus design efforts on critical obstacles
Pressman and Wildavsky, 1973) and interpre- to preferred outcomes. The ‘bottleneck’ obstacles
tation (Edelman, 1977, 1971). Will the design may lie with the processes of and incentives at
be adopted in full or in part by authorized work for adoption and implementation and pre-
authorities and processes and sustained in that dominant interpretations. In that case, adoption
status for its intended life? Even if it is, will and implementation institutions and symbolic
those persons, organizations, and social group- costumes are important parts of policy designs.
ings, who must act to make the adoption more If those in being are problematic, design efforts
than a piece of paper, do so in ways the design should re-invent them. The third is to filter alter-
assumes? How will the policy be portrayed native proto-designs thought to hold equivalent
along the way from design to adoption to promise for preferred outcomes. Adoption and
implementation to a verdict on its results? At implementation feasibility and interpretative
least by now few policy experts would deny the ease can provide a reasoned basis for selecting
importance of adoption and implementation. the design to be developed in detail, to be trans-
The pertinent issues for a conception of policy formed from a sketch into a blueprint. The
design are whether those matters are part of it, fourth is that what goes on in adoption and
and if so what role they should play in it. implementation will affect interpretation of the
Interpretation is less discussed with relation to differences made in beneficial results and in
policy design, and even abhorrent to some valued outcomes. We often care about processes
given its connotations of “spin” and deception. and not just outcomes.
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 84


Problem definition: value

clarification, causalities,
current realities

Policy design


Policy adoption

Yes No


Implementation through Implementation through

changes in human and changes in chains of
organizational activity consequences


Changes in targets, intermediate products


Changes in valued outcomes


Key: direct effects

indirect effects

Figure 4.2 An elaborated visualization

Acceptance of this inclusionary position on actions are influenced by changes in chains of

adoption, implementation, and interpretation, consequences and vice-versa. Achievements or
of course increases the burden of problem shortfalls in targets affect subsequent actions
definition. It adds more about which we need and chains. Perceptions that valued outcomes
to ascertain values, understand causality, and are being realized, exceeded, or placed at
know current reality. increased risk do matter for subsequent
An appealing feature of Figure 4.1 is its human and organizational actions and chains
sequential quality. Policy designers are like many of consequences. We can and should add adop-
of us in liking to cross tasks off of their to-do tion and implementation experiences into
lists and move on. Unfortunately, reality is that world of feedbacks. The feedbacks are sel-
rarely so permissive. Human and organizational dom without intermediating interpretations.
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 85


We then arrive at a visualization more like with reason to be rather sanguine about what
Figure 4.2. will result while we wait for the research to pan
The spaghetti in this picture contrasts with out. Another possibility is to “do” policy design
the neatness of Figure 4.1. Rather than linear in ways which take deficits in problem defini-
movement, we have chronic recursiveness to tion; and setting into account, a “yellow-light”
one or more previous tasks. Rather than unfil- position. That position does not gainsay the
tered impacts on subsequent stages we have need for better foundations for problem defini-
intervening interpretative processes, often tion, indeed it may amount to advocacy for
highly contested about their main content. efforts to “hold the fort” until those improve-
This template is a far more realistic skeleton of ments are available and provide openings to take
what goes on in and around public policy sys- advantage of them as they become available.
tems than that of Figure 4.1. Even so, making Tinbergen of course took the “proceed with
the visualization directly useful for specific caution” position. The wisdom to which we
policies, located specifically as to place and now turn elaborates his position and has
time, involves unpacking the elements and the implications for designing as an activity. It
flows and filters between them to an extent emphasizes exploration, making provision for
that defies representation in a single picture. “post-launch” adjustments or even “mission”
To return to the architectural analogy, consider cancellation, and hedging.
how a sketch of a building on a single page Exploratory activities include interactions
proliferates into increasingly detailed, volumi- and pre-negotiations with the parties who will
nous representations of parts of it. “live” in the design (e.g., users, clients, stakehold-
ers, adopters, implementors, target populations)
and best judgment impact statements about
policy consequences. Explorations include
attempts to understand the dynamics of the cur-
rent system and its participants. An example is
We can readily agree on the design situation most the ‘reference mode’ in systems dynamics mod-
conducive to sound problem definition and thus els capturing the ‘important feedback loops
to high confidence policy design. The fortunate existing between the … elements in the system
policy designer would have a well-filled physical that create the particular time variant patterns’
or electronic shelf. It would contain reason- already present (Saeed, 1994: 22). Others feature
ably clear specifications of values, based on probes to illuminate how the parties will behave
widely accepted patterns of actor representation. if placed into one or another designed future
It would provide strong causal theories and (“what if” games, exercises, simulations and
empirically established sensitivities to legislation, even “trial balloons”) under a range of possible
administrative guidance, judicial decisions, conditions, and with what results (e.g., labor
resource inputs, organizational reforms, and market measures in response to changes in tech-
public relations campaigns – together with their nology and external competition, as in Ho, 2000,
temporal leads and lags. It would make available or welfare reform financial incentives as in
detailed, timely and accurate profiles of current Robins, Michalopoulos, and Pan, 2001).
states of affairs. Dream on. Still others, as circumstances permit, include
Part of the policy design problematique deals “trying out” all or parts of the policy design
then with what to do in the substantial absence (prototypes in pilot studies) to determine its
of such attractive resources. One possibility performance (field experiments) or examining
defers policy design altogether, since it amounts situations where it in effect has been tried out
to laboring on “bound to fail” (or only to suc- (natural experiments). (See the classic work of
ceed by chance) illusions. This “further research Campbell, 1969; and Cook and Campbell, 1979,
is needed” position stands aside for an indeter- and recent examples with respect to rural
minate period of time from the pursuit of development programs in Fox and Gershman,
valued collective outcomes. It is tolerable only 2000, welfare to work programs, as in Loeb
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 86


and Corcoran, 2001 and Bloom et al., 2001, poses. Policy designers are usually invested at
occupational safety, as in Murphy-Greene and least psychologically in their products. So too are
Leip, 2002, and home financing to check urban those who adopted them and have gained rents
sprawl, as in Blackman and Krupnick, 2001.) by a prominent role in implementing them. So
Even if systematic try-out data are not available, too are the beneficiaries of the intermediate
a more informal scrutiny by experts knowledge- products. Getting any or all of them to respond
able about previous design efforts may be constructively to adjustment signals takes fore-
undertaken (e.g., for international economic thought about how to hold up to them persua-
sanctions, as in Miljkovic 2002, or the “murder sive prospects of reward and punishment.
boards” to which intelligence estimates and Adjustment is more likely when designs are
weapons designs are sometimes subjected). pursued on the basis of a widely shared
Provisions for adjustment bear on avoiding premise that adjustment is their normal and
locked-in commitment to all or part of a appropriate fate, given imperfect forecasting
design, and the supply of and responsiveness to capacities (as argued in Aaron, 2000). It is less
“back-talk from the situation” as the design is likely when doing so invokes an attribution of
followed. The former amounts to recommend- incompetence or evil. In the first situation,
ing flexibility in the design, keeping options adjustments amount to ‘upgrades’ or ‘product
open for choice as the design unfolds in prac- improvements’ which usefully take into
tice. That can be done by delegation in place of account experiences and changes in opportu-
specification, for example, granting adminis- nities. In the second, they are confessions of
trative discretion. It can be done by building defects which call into question the basic merit
conditionalities into the design to be triggered of the design itself. Such confessions simply
by specified states of affairs. It can be done by provide an enemy with weapons.
structuring the design in terms of diversified Of course, the ultimate in avoiding excessive
“stages of choices,” each one of which is selec- commitments is not to make commitments at
tive while maintaining more than one alterna- all, or only the most modest and limited ones.
tive (competitors) to be selected from at a later Extreme incrementalism is the ultimate hedge.
point when uncertainty reducing additional It argues for narrow and thin designs as they
information will be available. involve changes from what will otherwise be the
Back-talk (feedback) provision at its sim- case. The chances of disappointment are less if
plest involves designing in transparency one tries to change too little, than if one tries to
through such “sunshine” devices as public change too much. One is less likely to propound
access to information, whistle-blower protec- designs which will worsen matters. The chances
tions, and reporting and review requirements. of designs being realized in practice benefit
At another and more demanding level, it ben- from “going with the flow” rather than challeng-
efits from the design containing explicit ‘flags’ ing most prevailing norms and standard operat-
relevant to a need for adjustment. These may ing procedures – or the incentives which
be milestones (what should have happened by support them. The counsel of incrementalism
a certain point as the design was put into prac- also implies placing policy designs on a “short
tice) or indicators confirming or disconfirm- leash” (e.g., through such mechanisms as sunset
ing that the design is performing as intended. provisions, or one-year funding, or institutional
It often is desirable and necessary to design in “checks and balances”). Go through the yellow
organizational cells empowered to collect and light very slowly and, after that, consider what to
process such information.1 do at the next intersection.
Yet even if designs have ample provision for An incremental approach surely can be part
the supply of feedback, other considerations of a “learning strategy” but in that case it too
merit attention to increase the chances of requires some of the same information provid-
responsiveness to it. These are more matters of ing and responsiveness encouraging features as
the design management situation than of the the adjustment emphasis discussed previously
design itself and the incentives that situation and blurs into it.
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 87


Another form of hedging is ‘compromise’ difficult to reverse or require more time and
(Arrow, 1986). The design can still be broad effort to do so. Opponents can mobilize and
and deep, but does not try to optimize in terms engage in blocking actions. Stances of adjust-
of intermediate target products or valued ment and hedging limit the incentives to the
outcomes. Designers ‘respect the uncertainty parties to bet strongly on a particular design,
of their basic premises’ by trying to ‘strike a or to accept it more than superficially. While
balance between meeting all contingencies, such stances do avoid the discrediting risks of
with great waste of resources for the contin- unfulfilled grand promises, they also work
gencies that do not arise, and meeting none, against the sorts of appeals that can mobilize
with the penalties that follow. How the balance enthusiastic, intense support.
between safety and daring is struck depends on
the probabilities of the different possibilities,
the costs of providing against them, and the
penalties for failure to meet them’ (Arrow,
1986: 162, 165–66).
These maxims all have strong arguments in
their favor. Unfortunately, they also tend to have Other than further research, there are three non-
several drawbacks. First, they increase the tasks fatalistic responses to the challenges of policy
involved with policy design and thus the design a la Tinbergen introduced previously.
resources and time it takes. That is reasonably One is of course to proceed with due humility,
self-evident for the exploratory and adjustment convinced that there is no other socially respon-
providing features and for hedging through sible course and with modest hopes about
compromise. Less self-evidently, it is also improvement in the tools available to us. A
true for incrementalism. That road leads to end- second, which does not buy into inevitable
less, short, and tight deadline policy design policy failure, is to drop a “pretentious” rhetoric
cycles. Each one involves a full set of tasks, as in of design in favor of the “old-fashioned way” to
Figure 4.2. The parties to policies in practice deal with policy problems. Get a bunch of smart,
have ample reason to act in their own sweet motivated, and knowledgeable people together
time as their bargaining power increases as and hope that they will be creative, politically
deadlines approach. The proliferation of design agile, and lucky. The problem is that even the
tasks, and increases in the resources and time “best and the brightest” can turn out to be wrong
they consume, raise the chances of policy design or lack the mandate for effective agency. A third,
fatigue and lessen the energies and speed of design with a fundamental difference, is that to
actually moving out to realize a design. which we now turn.
Second, for reasons mentioned earlier, it is Design with a difference does not directly
not obvious that many of the envisioned tasks address policies as choices for classic content
are any more capable of successful completion areas (e.g., social welfare, public health, eco-
than those discussed with reference to pro- nomic growth, defense). What it designs
blem definition. Consider the determinations instead are processes and procedures for
involved in Arrow’s compromising. Nor is it any generic purposes of problem definition, value
clearer than it was for problem definition as to clarification, and policy adoption, implemen-
when we will have done these tasks well enough tation, and interpretation. Normative posi-
to give us warranted confidence in proceeding. tions and pragmatic judgments come together
Third, delay in making major policy design to make process and procedure design the road
commitments has its own risks (as with U.S. to better policy performance. While oversim-
health insurance reforms in Oliver, 1999, or the plified, it is helpful to distinguish two strands –
“Oslo process” intended to resolve the Israeli– institutionalists and, after Linder and Peters
Palestinian conflict). Windows of opportu- (1992), ‘deliberationists.’
nity can close. Situations can deteriorate and Pragmatically, both are impressed by the dif-
unwanted ones can become either more ficulties noted in meeting the demands of
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 88


Figure 4.2, the requisite “matching” of policy (Weimer, 1992). Those reforms ostensibly will
design elements to current and altered charac- shift distributions of power and information
teristics of the world in which the design is to so that the relevant parties behave better. That
be put into practice. They are pessimistic about will, however, only happen if the incentives
our ever having the control or foresight to do cross a hard to discern strength threshold and
so well enough. Normatively, institutionalists include clear signals about their future applica-
and deliberationists see the need for protec- tion, as with pro-environment technology
tions against exploitation of the roles of efforts by firms in Norberg-Bohm, 1999.
problem definer, policy designer, adopter, Deliberationists focus more on the quality
implementer, and interpreter by special inter- of the discourse from which policy emerges
ests. Where they differ is in how to deal with than on what in particular emerges. To “get the
those pragmatic and normative problems. discourse right”, some recommend ‘explicit
Institutionalists, after Douglas North standards of discourse’ (MacRae, 1988;
(1990: 3), attempt to prescribe the ‘rules of the Anderson, 1987). Others concentrate on par-
game’ designs include and the mechanisms to ticular ‘models of deliberation’ (Fischer, 1981;
make those rules operational, that is, institu- Paris and Reynolds, 1983; Linder and Peters,
tional designs. One example is the formation 1992: 226–7). The former focus on what ought
of partnerships to contain and resolve what to be discussed; the latter on who will partici-
would otherwise be policy, program, and pro- pate and their freedom to raise matters they
ject paralyzing conflicts, as with the World care about. The latter places in the hands of the
Commission on Dams (Brinkerhoff, 2002). parties who will be involved and affected the
A more general example is that of the turn in tasks of value and option formation, context
development thinking toward governance to appraisal and choice, rather than stipulating
emphasize actions to bring about ‘accountability, them in the design. That empowerment osten-
a conducive legal framework, and transparency’ sibly will bring to bear creativity, information,
(World Bank, 1991: ii) drawing on general prin- diagnostic skills, value clarifications, and a
ciples about conducive conditions and necessary sense of ownership which a temporally or
properties of each. Accountability, it is claimed, situationally distant professional elite of policy
has a better chance when procedures and designers cannot muster. Interactions between
processes make for continuing competition, e.g., the parties along process and procedure lines
through full or partial privatization. Participatory provided by a designer will itself make for
opportunities for affected publics to articulate ‘matching’ and is likely to do so better.
their interests are helped by designing in NGO These recipes seem very different from those
roles and decentralizing responsibility to more of Tinbergen and his descendants. Yet three
proximate bureaucracies and elected office- questions remain to be answered. Can institu-
holders. A legal framework is as it should be tionalist and deliberationist approaches be
when ‘rules are known in advance … [and are] married with his in ways which compliment
… actually in force, mechanisms … ensure appli- each others’ strengths and compensate for each
cation of the rules, conflicts … [are] … resolved others’ weaknesses? Would institutionalist and
through binding decisions of an independent deliberationist recommendations necessarily
judicial body or through arbitration, there are work as envisioned or are some additional
known procedures for amending the rules when properties required? Is making those proper-
they no longer serve their purpose.’ A design ties available and getting the procedures to be
which provides for these conditions will be more realized into practice similar to some of the
likely to ‘prevent predatory government actions challenges posed in Figure 4.2?
and agent rent-seeking’ (World Bank, 1991: iii). The answer to the first question is largely
How to get institutions to behave as desired positive. For institutional design, that is implied
fundamentally amounts to taking advantage of by our previous discussion of adoption, imple-
the incentives they already have and ones pro- mentation and interpretation. Tinbergen after
vided by “reforms” in procedures and processes all recognized the importance of the targets of
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 89


institutional design. Yet he also recognized that discourse, the investment they are willing to
policy satisfaction also includes policy results make in it, and how binding the understandings
in terms of particular tangible and psychic reached will be. One can imagine cases in which
benefits and costs and not just processes and the participants would be diverse and represen-
procedures. Most assessments of policy perfor- tative, have substantial if not identical commu-
mance find it to be the joint product of both nicative skills, give high priority to the dialogues,
policies and institutions involved with them and have substantial influence with those they
(e.g., Keiser and Meier, 1996). Satisfaction can represent (as illustrated in Walters, Aydelotte,
be denied by getting institutions wrong (Meier and Miller, 2000). One can also imagine cases of
and Smith, 1994), but achieving satisfaction the opposite (as illustrated in Pelletier et al.,
takes more than getting them right. Even if, for 1999). Relevant parties are advertently or inad-
example, one can make implementing bureau- vertently left out. Some participants, for all sorts
cracies quite compliant in a Weberian fashion, of reasons, are highly effective communicators
we may still find abhorrent or stupid the policy and others are “bumps on the log.” Attendance is
they are complying with. a pro-forma or watchdog activity rather than a
Deliberationists’ communicative suggestions constructive engagement. Those who go to the
could certainly contribute to an improved grasp meeting are the people their home group neither
of the values at work, defining policy problems, trusts or respects.
knowing current realities, and understanding The differentiations from Tinbergen’s posi-
the reasons for human and organizational tion of institutionalists and deliberationists call
actions (e.g., Berger, 1977). What would be con- for different rules of the policy design and policy
ducive interpretations and conditions for adop- process game than often prevail. How to get that
tion and effective implementation might be to happen and meet relevant conditionalities
made clearer. At the same time, communicative involves a system of constructions, actions and
discourse could be improved by the sorts of consequences, rather like those in Figure 4.2.
analyses and information at the heart of Institutionalists would be wise to consider
Tinbergian policy design, and be substantially the variety in what has happened in light of
disappointing in their absence. their rather similar prescriptions for post-
Whether the institutionalists’ or the delibera- Communist nations. They also might well con-
tionists’ recommendations would have the bene- sider the implications of historical experience,
fits they envision seems contingent. The former which shows that nations with ‘basic success
pose recipes which experience suggests are less stories … [with respect to meeting basic human
than guaranteed in their consequences across needs] … have had very different economies
situations. That is why we see alternating cycles from an institutional and political point of
of centralization and decentralization (Montias, view’ (Streeten, 1986: 29).
1986). Institutionalists’ recipes may fail to Deliberationists would do well to note the
acknowledge how learning from one pass at particular circumstances apparently needed to
them may lead to behavior which deprives them get a process of evolving policy through a more
of subsequent efficacy. For example, rationaliz- cooperative dialogue in place for environmen-
ing closing of U.S. military bases was helped in talists with ranchers and agriculturalists in the
one pass by establishing closing procedures American West under Department of Interior
“behind a veil of ignorance” where individual sponsorship. They may find sobering how
legislators did not know if and how their district some steps in line with a shared information,
would be affected (Weimer, 1992). While the transparency standard can, in particular, con-
procedures have remained in place for subse- texts countervail other parts of their prescrip-
quent rounds, legislators became fully aware of tions. Consider uses of courts and legislatures
the potential for adverse district consequences to obstruct or enhance pursuit of data sought
and moved to block further closings. by those already holding negative views about
The deliberationist approach depends on who the effects of smoking or particular industrial
gets to participate, the skills they bring to the products and processes. In those instances,
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 90


greater achievement of transparency could and Linder and Peters, 1988, 1984). From this
has had the consequence of deterring “emanci- perspective, the rules for inclusion in the policy
patory” policy dialogue (Hilts, 1999). designer’s basic repertoire emphasize relevance
Institutionalist and deliberationist approaches to the challenges posed by Figure 4.2 and to a
are potentially important and useful. They are wide range of policy domains in terms of both
not sufficient unto themselves, automatically issue areas and levels and forms of governance.
successful, or devoid of the challenges facing Any selection on those bases is of course open
other approaches to policy design. to argument, but I suggest ten core strategies.
They belong if only because going entirely with-
out any one of them exposes the designer and
his or her designs to substantial, hard to avoid
grief. The core ten are then in effect a necessary
part of the repertoire, not to be confused with a
complete one. For convenience they can be
When we consider the intellectual instruments labeled as: discipline breakout; minimum neces-
available to policy designers, it is possible to sary conditions; safeguards; placement; opposi-
find abundance or a rather thin portfolio. tional analysis; borrowing; tinkering; backward
If one includes the bodies of instruments mapping; forward mapping; and judgment
used in the relevant sciences and professions – shaping. They may seem banal, until one reflects
by no means limited to the social sciences – on how often they are slighted with, at least in
abundance seems an accurate description. retrospect, disappointing results.
Consider Dorfman’s (1986: 108) list of relevant Discipline breakout rejects as more than
specialties for environmental policy: engineer- stimulating grist for the designer’s mill the
ing; agronomy; meteorology; hydrology; chem- framing provided by any single technical and
istry; soil sciences; epidemiology; toxicology; professional disciplines. The designer must
oncology; zoology; biology; forestry; geology; break out from the constrained definitions of
economics; social sciences. Of course, the social problems, lists of actors, patterns of causation
sciences themselves contain many disciplines and correlation, and inventories of policy moves
and we can and should also recognize the perti- set by any single such discipline. Reality always is
nence of law, the information sciences, and more complicated than that and policy designs
political and social philosophy. If we consider are instruments to deal with reality rather than
the wealth of different policy issues for which scholastic gymnastics. The most efficacious
there is some demand for policy design the list designs will then be full of what are tangential
would get much longer still. Each has its instru- matters from any orthodox single frame per-
ments which are potentially applicable and spective, as argued in Ascher, 2002 and exempli-
examples can be found of their use in policy fied in Brewer and Kakalik, 1979.
design (e.g., for economics Cody, Kitchen and A minimum necessary conditions strategy
Weiss, 1990; Henderson, 1986; Cohen and has the policy designer identify and address
Wyplosz, 1995; Brainard and Martimort, 1996). what has to be the case, or be made the case, for
There are far too many strategies and tech- a preferred outcome or range of outcomes to
niques to review here, particularly if considered have much of a chance of realization. If those
with responsible cautions about the appropri- conditions are not met, the design should be a
ateness of their theoretical assumptions and non-starter. If those necessary conditions seem
feasibility of their information requirements. in reach via the design, it should receive seri-
The long-standing abundance just discussed ous consideration and warrants elaboration to
for practical purposes contrasts with a much enrich it with sufficient conditions. This strat-
smaller repertoire of options when we limit egy involves thinking through how to avoid
our consideration to strategies formulated, well-known, chronically possible failure by key
with a special focus on policy design and elements in any policy design, e.g., recurrent
generically applicable to it (Weimer, 1992: 135; types of government and market failure. It also
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 91


involves dealing with what are for the outcome including by legacy policy networks, e.g., for
the design pursues crucial parties, and crucial agricultural environmental regulations, as in
desiderata for them. Montpettit, 2002.
For example, in addressing the necessary Oppositional analysis assumes that attempts
conditions for Israeli–Palestinian peace, Yorke to pursue a policy design in practice will be the
(1990) focuses on the need to meet minimum target of deliberate foiling moves by adversaries
Israeli requirements (‘full peace and security’) at one or more points in Figure 4.2 (Wohlstetter,
and Palestinian ones (‘an independent state’). 1964, 1968). Thus, it becomes important to
She reasons that those minima can only be pro- envision the stimulus the policy as designed will
vided with normalization between Israel and itself pose to those who would object to it (mea-
some other states in the region. Given the estab- sure/counter-measure sequences). The active
lished suspicions and grievances, sustained nature of the environment also calls for antici-
movement toward peace requires simultaneous pation of secular trends and changes, for exam-
judgments by the parties that the advantages of ple, in technology, which may enhance the
peace outweigh the advantages of a continua- opportunities available to opponents. Still fur-
tion of the status quo … all have a stake in keep- ther, attention should go to the possibility of sit-
ing the peace and … that peace holds fewer risks uational changes (shocks), unplanned by any of
than the status quo’ (Yorke, 1990: 117). the affected parties, which may provide attrac-
That last phrase illustrates the third strategy, tive opportunities for opposition. Imagining
safeguards. The risks in following through such changes in the ‘terms of competition’ usu-
with a policy design do not fall primarily on its ally requires very detailed analysis of the end-to-
designers but on the affected parties. end chains of developments encompassed in a
Realization of the design in practice requires Figure 4.2 design and thus of the complete life-
their tolerance and often active steps on their cycle of a policy design in practice.
part. They are more likely to entrust their fate Design work usually rests on faith that there
to the design when it provides and is seen by are unachieved but, in principle, possible
them to provide means to limit damage should opportunities to do better than the status quo
it go awry or break down. For Yorke, those pro- and the looming future. It remains to go from
visions involve guarantees and commitments faith to credible demonstration of what those
by third parties external to the Middle East. are and the realistic availability of means to
Also, given the very mixed record of policy achieve a more preferred future. Borrowing is
design results, the responsible designer recog- a strategy for doing so (Rose, 1993; Schneider
nizes a similar need whatever the affected par- and Ingram, 1988; Weimer, 1993). It embraces
ties may think. The strategy of safeguards, it what seems to have worked well elsewhere in
should be noted, is not equivalent to incre- circumstances sufficiently analogous to those
mentalism, which carries its own risks. facing the policy designer. The premise is that
Placement begins from awareness that such “solutions” or “best practices” worked on
policy designs apply to systems with histories. similar problems in similar contexts, so they
That is true for the actors in them (individuals, should work for me.2 The wheel need not be
organizations, social groupings, politico-legal reinvented but only emulated. Warranted bor-
entities) and policy issues and options. The rowing clearly involves a host of judgments
designer starts not from a clean slate but from about performance elsewhere, and similarities
a clutter. Placement (Neustadt and May, 1986) to the situation facing the designer.
develops chronologies for pertinent actors and The greater the scope of what is borrowed,
issues of major experiences in their past. Those the harder it will be to get it into practice in the
provide no precise causal guidance as to prece- relevant situation. Even if what is borrowed is
dents which will be used in the present and quite limited and narrow, no two situations
future, but they are suggestive of how policy will be precisely alike in all pertinent respects.
designs alternatives will be treated differently These realities, plus the persistent appeal
from actor to actor and issue area to issue area, of incrementalism, argue for a strategy of
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 92


tinkering (Weimer, 1993). The idea is to tune uncertainties which would damp optimism;
or tweak what exists, or what is borrowed and attraction to sure gains and underweighting of
the context to receive it, without challenging only probable ones; and favoring small steps,
most of what is in place. Designers go for rela- and aggregates of small steps, away from status
tively ‘feasible manipulations’ of what already quo behavior over large steps singly and in
exists (May, 1981). The policy world, even its aggregate. A particularly misleading tendency is
reformist elements, usually is full of those who to extrapolate the ease or difficulty of the initial
prefer tinkering. It then will be hard to follow steps in a policy to that of the critical for out-
a less incremental design before considering comes later steps, e.g., as in much of the Bush
such options and discrediting them. Administration responses to 9/11, including the
Backward-mapping and forward-mapping Afghan and Iraq ventures (Bobrow, 2003).
each start from the same premise that “end-to- Moves to shape responses to designs include
end”, full life cycle treatment is needed for possibilities to take advantage of the ways in
robust design, a premise shared with opposi- which preferences are context-dependent.
tional analysis. The backward and forward Support for a particular policy design becomes
strategies differ in where they would have a contingent on the presence of alternative
designer start (Elmore, 1985). In backward designs. By presenting an increased number of
mapping the idea is to start just before the end alternatives, the designer calls into action the
in the production of outcomes and work dynamics of “tradeoff contrast” and “extreme-
toward the beginning, identifying who must do ness aversion”. For the former, imagine two
what and what must be the case all along the policy designs one of which offers both greater
way (Elmore, 1979–80). Forward-mapping benefits and costs than the other. Introducing a
starts at the initial point of realizing a policy third design with the greater benefit but at a
design and works toward its conclusion. It lays much higher cost will increase preferences for
out the transition from what exists, or is cur- the first design at the expense of the second. For
rently “programmed” by man and nature to extremeness aversion, imagine two policy
exist, to a more preferable outcome. Think of designs, each of which has a great positive puta-
backward or forward mapping as yielding a set tive consequence and a great negative one. If a
of stripmaps which together span the journey third design with more limited positive and
from origin to destination, but which can be negative consequences is introduced, it will pull
produced and read, starting with the last strip support away from one or both of the first two.
or the first. These strategies call for moving These tendencies operate for designers and
briskly from generalized principles to the not just for the parties affected by policy
detailed particulars of a policy context, taking design.3 In sum, understanding judgment
into account possible changes in it, such as shaping has substantial implications for how
those highlighted by oppositional analysis. we develop policy designs, for their quality,
The final strategy, judgment shaping, draws and for their chances of acceptance.
on research in behavioral decision theory At their extremes, any one of the ten com-
(Tversky and Kahneman, 1974; Kahneman mandment strategies is in some tension with
and Tversky, 1979; Kahneman and Lovallo, one or more of the others. Balanced and judi-
1993; Tversky and Simonson, 1993). That cious attention to all of them can, however,
research yields human tendencies in making improve capacity to deal with the challenges of
judgments. They imply correctives policy policy design.
designs should supply, and design moves to
shape the judgments of those involved in their
adoption, implementation and interpretation.
The tendencies differ from traditional models NOTES
of rational judgment.
Tendencies to correct for include: treating 1. Of course, pledges for continuing assessment may
problems as isolated unique instances; denial of not actually have been implemented, as with the waivers
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 93


granted to American states for departures from previous Bobrow, Davis B. and John S. Dryzek (1987). Policy
federal health care program guidelines. Analysis by Design. Pittsburgh: University of
2. Related to this strategy is a filtering one of “counter- Pittsburgh Press.
borrowing”. In it, the search is for failures in similar Brainard, S. Lael and David Martimort (1996).
contexts and what is found should then be rejected as a
‘Strategic Trade Policy Design with Asymmetric
design alternative.
3. The implications for the self-awareness and discipline
Information and Public Contracts.’ Review of
of policy designers are significant, with implications anal- Economic Studies, 63: 81–105.
ogous to those incorporated into the professional prepara- Brewer, Gary D. and James S. Kakalik (1979).
tion of psychiatrists. Handicapped Children: Strategies for Improving
Services. New York: McGraw-Hill.
Brinkerhoff, J.M. (2002). ‘Global Public Policy,
REFERENCES Partnership, and the Case of the World Commission
on Dams.’ Public Administration Review, 62: 324–36.
Bruner, Ronald D. (1986). ‘Case-wise Policy
Aaron, Henry J. (2000). ‘Seeing Through the Fog: Information Systems: Redefining Poverty.’ Policy
Policymaking with Uncertain Forecasts.’ Journal Sciences, 19: 201–23.
of Policy Analysis and Management, 19: 193–206. —— (1983). ‘Case-Wise Policy Analysis: Another
Ackoff, Russell (1979). ‘The Future of Operational Look at the Burden of High Energy Costs.’ Policy
Research is Past.’ Journal of the Operational Sciences, 16: 97–125.
Research Society, 30: 93–104. Campbell, Donald T. (1969). ‘Reforms as
Adelman, Irma and J. Edward Taylor, eds. (1986). Experiments.’ American Psychologist, 24: 409–29.
The Design of Alternative Development Strategies. Castro, Mary and International Labor Office.
Rohtak, India: Jan Tinbergen Institute of Devel- (1978). Migration in Brazil: Approaches to Analysis
opment Planning. and Policy Design. Liege: Ordina.
Alexander, Christopher (1964). Notes on the Synthesis Chung, Kyong Won (1993). ‘The Miracle of Han
of Form. Cambridge: Harvard University Press. River: Government Policy and Design Manage-
Alexander, Ernst A. (1982). ‘Design in the Decision- ment in the Motor Industry.’ Design Management
Making Process.’ Policy Sciences, 14: 279–92. Journal, 4: 41–47.
Anderson, Charles (1987). ‘Political Philosophy, Cody, John, Richard Kitchen and John Weiss (1990).
Practical Reason, and Policy Analysis,’ In Frank Policy Design and Price Reform in Developing
Fischer and John Forester, eds. Confronting Values Countries: Guidelines with Special Reference to
in Policy Analysis. Beverly Hills: Sage, 22–44. Industry. New York: St. Martin’s.
Arrow, Kenneth J. (1986). ‘Planning and Uncertainty,’ Cohen, Daniel and Charles Wyplosz (1995). ‘Price
In Adelman and Taylor, op.cit., 161–70. and Trade Effects of Exchange Rate Fluctuations
Ascher, William (2002). ‘The Brewer-Kakalik and the Design of Policy Coordination.’ Journal of
Research on Services for Handicapped Children: International Money and Finance, 14: 331–47.
A Paradigm of Applied Policy Sciences.’ Policy Cook, Thomas D. and Donald T. Campbell (1979).
Sciences, 35: 305–10. Quasi-Experimentation: Design and Analysis for
Bardach, Eugene (1977). The Implementation Game. Field Settings. Boston: Houghton Mifflin.
Cambridge: MIT Press. Daniere, Amrita G. and Lois M. Takahashi (1999).
Berger, Thomas R. (1977). Northern Frontier, ‘Public Policy and Human Dignity in Thailand:
Northern Homeland: Report of the Mackenzie Environmental Policies and Human Values in
River Inquiry. Toronto: James Lorimer. Bangkok.’ Policy Sciences, 32: 247–68.
Blackman, Allen and Alan Krupnick (2001). DeLeon, Peter (1994). ‘Reinventing the Policy
‘Location-Efficient Mortgages: Is the Rationale Sciences: Three Steps Back to the Future.’ Policy
Sound?’ Journal of Policy Analysis And Manage- Sciences, 27: 77–95.
ment, 20: 633–49. Demchak, Chris C. (2002). ‘Un-Muddling Homeland
Bloom, Howard S., et al. (2001). ‘Testing a Financial Security: Design Principles for National Security
Incentive to Promote Re-employment among in a Complex World.’ The Forum, 1. www.bepress.
Displaced Workers: The Canadian Earnings com/forum.
Supplement Project (ESP).’ Journal of Policy Dorfman, Robert (1986). ‘Benefit/Cost Analysis of
Analysis and Management, 20: 505–23. Environmental Problems,’ In Adelman and
Bobrow, Davis B. (2003). ‘Losing to Terrorism: Taylor, op.cit., 103–23.
An American Work in Progress.’ Metaphilosophy, Dror, Yehezkel (1986). Policy Making Under
34: 387–406. Adversity. New Brunswick, NJ: Transaction.
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 94


—— (1971). Design for Policy Sciences. New York: —— and Amos Tversky (1979). ‘Intuitive
American Elsevier. Prediction: Biases and Corrective Procedures.’
Edelman, Murray (1977). Political Language: Words Management Science, 12: 313–27.
that Succeed and Policies that Fail. New York: Keiser, Lael R. and Kenneth J. Meier (1996). ‘Policy
Academic. Design, Bureaucratic Incentives, and Public
—— (1971). Politics as Symbolic Action. New York: Management: The Case of Child Support
Academic. Enforcement.’ Journal of Public Administration
Elmore, Richard F. (1985). ‘Forward and Backward Research and Theory, 6: 337–64.
Mapping: Reversible Logic,’ In Kenneth Hanf and Killick, Tony (1995). IMF Programmes in Developing
Theo A.J. Toonen, eds. Policy Implementation in Countries: Design and Impact. New York: Routledge.
Federal and Unitary Systems. Boston: Martinus Lasswell, Harold D. (1971). A Pre-View of the Policy
Nijhoff, 33–70. Sciences. New York: Elsevier.
—— (1979–80). ‘Backward Mapping: Imple- —— (1961). Politics: Who Gets What, When, and
mentation Research and Policy Design.’ Political How. Cleveland: World.
Science Quarterly, 94: 601–16. Laudon, Kenneth C. (1986). Dossier Society: Value
Fischer, Claude S., Michael Hout, Martine Sanchez Choices in of the Design of National Information
Jankowski, Samuel R. Lucas, Ann Swidler, and Systems. New York: Columbia University Press.
Kim Voss (1996). Inequality by Design: Cracking Lerner, Daniel and Harold D. Lasswell, eds. (1951).
the Bell Curve Myth. Princeton: Princeton The Policy Sciences: Recent Developments in
University Press. Scope and Methods. Stanford: Stanford University
Fischer, Frank (1992). ‘Participatory Expertise: Press.
Toward the Democratization of Policy Science,’ In Linder, Stephen H. and B. Guy Peters (1992).
William N. Dunn and Rita Mae Kelly, eds. Advances ‘A Metatheoretic Analysis of Policy Design.’ In
in Policy Studies Since 1950, Policy Studies Review Dunn and Kelly, op.cit., 201–37.
Annual Volume 10. New Brunswick: Transaction, —— (1988). ‘The Analysis of Design or the Design
351–76. of Analysis.’ Policy Studies Review, 7: 738–50.
—— (1981). Politics, Values and Public Policy. —— (1984). ‘From Social Theory to Policy Design.’
Boulder: Westview. Journal of Public Policy, 9: 237–59.
Fox, Jonathan and John Gershman (2000). ‘The Loeb, Susanna and Mary Corcoran (2001). ‘Welfare,
World Bank and Social Capital: Lessons from Ten Work Experience, and Economic Self-Sufficiency.’
Rural Development Projects in the Philippines Journal of Policy Analysis and Management,
and Mexico.’ Policy Sciences, 33: 399–419. 20: 1–20.
Goggin, Malcolm L. (1987). Policy Design and the MacRae, Duncan (1988). ‘Professional Knowledge for
Politics of Implementation: The Case of Child Policy Discourse.’ Knowledge in Society, 1: 6–24.
Health Care in the American States. Knoxville: May, Peter J. (1981). ‘Hints for Crafting Alternative
University of Tennessee Press. Policies.’ Policy Analysis, 7: 227–44.
Henderson, David (1986). Innocence and Design: Meier, Kenneth J. and Kevin B. Smith (1994). ‘Say It
The Influence of Economic Ideas on Policy. Oxford: Ain’t So, Moe: Institutional Design, Policy
Blackwell. Effectiveness, and Drug Policy.’ Journal of Public
Hilts, Philip J. (1999). ‘Law on Access to Research Administration Research and Theory, 4: 429–42.
Data Pleases Business, Alarms Science.’ New York Miljkovic, Dragan (2002). ‘Economic Sanctions as
Times, July 31, A1ff. the Propositional Sustainability Problem.’ Policy
Ho, Lok Sang (2000). ‘Wage Subsidies as a Labour Sciences, 35: 1–15.
Market Policy Tool.’ Policy Sciences, 33: 89–100. Montias, J. Michael (1986). ‘On the Centralization
Hoag, Malcolm W. (1976). United States Foreign and Decentralization of Economic Activities,’ In
Policy: Why Not Project Interdependence by Design?; Adelman and Taylor, op.cit., 171–90.
Los Angeles: International Institute for Economic Montpetit, Eric (2002). ‘Policy Networks, Federal
Research. Arrangements, and the Development of Environ-
Hurley, Robert E., Deborah A. Freund, and John E. mental Regulations: A Comparison of the
Paul. (1993). Managed Care in Medicaid: Lessons Canadian and American Agricultural Sectors.’
for Policy and Program Design. Ann Arbor: Health Governance, 15: 1–20.
Administration Press. Murphy-Greene, C. and L.A. Leip (2002). ‘Assessing
Kahneman, Daniel and Dan Lovallo (1993). ‘Timid the Effectiveness of Executive Order 12898:
Choices and Bold Forecasts: A Cognitive Perspective Environmental Justice for All?’ Public
on Risk Taking.’ Management Science, 39: 17–31. Administration Review, 62: 679–87.
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 95


Neustadt, Richard E. and Ernest R. May (1986). —— (1988). ‘Systematically Pinching Ideas: A
Thinking in Time: The Uses of History for Decision Comparative Approach to Policy Design.’ Journal
Makers. New York: Free Press. of Public Policy, 8: 61–80.
Norberg-Bohm, Vicki (1999). ‘Stimulating “Green” Schon, Donald A. (1987). Educating the Reflective
Technological Innovation: An Analysis of Practitioner: Toward a New Design for Teaching
Alternative Policy Mechanisms.’ Policy Sciences, and Learning in the Professions. San Francisco:
32: 13–38. Jossey-Bass.
Norchi, Charles H. (2000). ‘Indigenous Knowledge —— (1983). The Reflective Practitioner: How
as Intellectual Property.’ Policy Sciences, Vol. 33 Professionals Think in Action. New York: Basic
no. 3–4, pp. 387–398. Books.
North, Douglas (1990). Institutions, Institutional Simon, Herbert A. (1972). The Sciences of the
Change, and Economic Performance. New York: Artificial. Cambridge: MIT Press.
Cambridge University Press. Stewart, Jenny and Russell Ayres (2001). ‘Systems
Ohls, James C. and Harold Beebout. (1993). The Theory and Policy Practice: An Exploration.’
Food Stamp Program: Design Tradeoffs, Policy, and Policy Sciences, 34: 79–94.
Impact: A Mathematica Policy Research Study. Streeten, Paul (1986). ‘Basic Needs: The Lessons,’ In
Washington: Urban Institute Press. Adelman and Taylor, op.cit., 27–37.
Oliver, Thomas R. (1999) ‘The Dilemmas of Taylor, John B. (1993). Macroeconomic Policy in a
Incrementalism: Logical and Political Constraints World Economy: From Econometric Design to
in the Design of Health Insurance Reforms.’ Journal Practical Operation. New York: W.W. Norton.
of Policy Analysis and Management, 18: 652–83. Tinbergen, Jan (1958). The Design of Development.
Paris, David C. and James F. Reynolds (1983). The Baltimore: Johns Hopkins University Press.
Logic of Policy Enquiry. New York: Longman. Tversky, Amos and Itamar Simonson (1993).
Pelletier, David et al. (1999).‘The Shaping of Collective ‘Context-Dependent Preferences.’ Management
Values through Deliberative Democracy: An Science, 39: 1179–89.
Empirical Study from New York’s North —— and Daniel Kahneman (1974). ‘Judgment
Country.’ Policy Sciences, 32: 103–31. Under Uncertainty: Heuristics and Biases,’ In
Pellikan, Huib and Robert J. van der Veen. (2002). Richard Zeckhauser, et al., eds., Benefit-Cost and
Environmental Dilemmas and Policy Design. New Policy Analysis. Chicago: Aldine, 295–307.
York: Cambridge University Press. Walters, Lawrence C., James Aydelotte, and Messica
Perlmutter, Howard V. (1965). Toward a Theory of Miller (2000). ‘Putting More Public in Policy
Social Architecture, Tavistock Pamphlet 12. Analysis.’ Public Administration Review, 60: 349–59.
London: Tavistock. Weimer, David L. (1993).‘The Current State of Design
Pressman, Jeffrey and Aaron Wildavsky (1973). Craft: Borrowing, Tinkering, and Problem Solving.’
Implementation. Berkeley: University of California Public Administration Review, 53: 110–20.
Press. —— (1992). ‘Claiming Races, Broiler Contracts,
Roessner, J. David, ed. (1988) Government Inno- Heresthetics, and Habits: Ten Concepts for Policy
vation Policy: Design, Implementation, Evaluation. Design.’ Policy Sciences, 25: 135–59.
New York: St. Martin’s. Wildavsky, Aaron (1979). Speaking Truth to Power.
Robins, Philip K., Charles Michalopoulos, Elsie Pan. Boston: Little, Brown.
‘Financial Incentives and Welfare Reform in the Wohlstetter, Albert (1968). ‘Theory and Opposed
United States.’ Journal of Policy Analysis and Systems Design,’ In Morton A. Kaplan, ed., New
Management, 20: 129–50. Approaches to International Relations. New York:
Rose, Richard (1993). Lesson-Drawing in Public St. Martin’s, 19–53.
Policy: A Guide to Learning Across Time and Space. —— (1964). ‘Analysis and Design of Conflict
Chatham, NJ: Chatham House. Systems,’ In Edward S. Quade, ed., Analysis for
Saeed, Khalid (1994). Development Planning and Military Decisions. Santa Monica: RAND, P-
Policy Design: A System Dynamics Approach. 387-PR, 103–48.
Aldershot: Avebury. World Bank (1991). Managing Development: the
Schneider, Anne and Helen Ingram. (1997). Policy Governance Dimension, Discussion Paper.
Design for Democracy. Lawrence: University Washington: The World Bank.
Press of Kansas. Yorke, Valerie (1990). ‘Imagining a Palestinian State:
—— (1989). ‘The Powers of Problem Definition: An International Security Plan.’ International
The Case of Government Paperwork.’ Policy Affairs, 66: 115–36.
Sciences, 22: 97–121.
Peter-3383-Chapter-04.qxd 6/8/2006 7:32 PM Page 96
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 97

Networks and Bargaining in
Policy Analysis


INTRODUCTION of interaction, a pattern that in the early years

of the third millennium A.D. is conceptualized
This chapter1 tells how scientific calls for as a network by most theorists. It is a story of
rationalism and for understanding non- how the rational model has conceded to inter-
rational behavior have fought visible, but at active theories of political and administrative
times less conspicuous, wars over the use of processes, and how the conceptions of policy
various models of interaction in policy processes have been broadened from being
processes. The concept of network (depicting based on the polity and politics in a narrow
various types of linkages between actors) has sense to being a societal affair involving many
been quite victorious, but that does not mean types of actors.
that the features it covers are new. However, the general ideas of networks have
Robert Hoppe has expressed the transfor- been present in the literature on mutual
mation of policy analysis over time well: It has adjustment for many years. Earlier on, how-
gone from “Speaking Truth to Power” to ever, there was less agreement about the right
“Making Sense Together” (Hoppe 1999, 201). term. This article establishes common themes
In this chapter, the difference between the two on the variation. The approach is systematized
statements is illustrated by the models of the historically. In the view of this author, social
rational actor and of mutual adjustment. We theories do not exist in any abstract sense.
analyze some core features of these models, They are constructed by scholars who interact
and from there we shall discuss a number of with one another and inspire one another in
developments within the literature in the complex, international research networks,
second half of the 20th century, in order to more or less in a Kuhnian (Kuhn 1962) way.
gain a better understanding of how theorists However, real paradigm shifts are rare in the
have dealt with human interaction in the social sciences (Lakatos 1974), while marginal
policy process. Subsequently, we shall go shifts in theoretical approaches are frequent.
through the way in which various schools of Scholars are subject to fads and fashions, they
thought have dealt with the resulting pattern apply explicit and implicit comparisons, they
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 98


compete for attention within their scholarly policy analysis. In particular, the fourth
community. Policy analysts also react to and generation of evaluation (Guba and Lincoln
analyze the same empirical phenomena in 1989) is closely linked to governance and
society. Thus, they create competition, innova- deliberative policy analysis. Fischer (1995)
tion and the diffusion of ideas, which often brings some of the pieces together nicely.
bear considerable resemblance to one another,
and which are discussed in groups of scholars
who share some fundamental views on social ORGANIZED ORDER VERSUS
theory. They then apply the theories with some MUDDLED PROCESSES
variation, according to the circumstances of
their empirical research.
The classical, rational and the mutual adjust-
In short, theory is contingent on time and
ment models form the backbone of this
space, and thus the present network under-
chapter. In this chapter, we shall focus on how
standing of policy has come about as a result
these models treat the decision-making
of scholars interacting and discussing the
processes and the interaction between actors.
possible interpretations of social phenomena –
in this case policy processes. This chapter
explores some trends that have been present Rational Policy-making
for the last 30 years or so without pretending
any full coverage, since the theme of policy “Speaking Truth to Power” (Wildawsky 1979)
networks is vast. And since the author was pre- indicates a troubled relationship between
sent most of the time, participating in several science and politics, between those finding the
networks, the critical reader may find some true state of the world and those wanting to
autobiographical biases. The reader will find rule it. Indeed, much of the policy literature is
other recent accounts of the development of concerned with authority, expertise and order
the theme in Hoppe (1999), Hajer and (Colebatch 1998). First, the policy literature
Wagenaar (2003), and Fischer (2003), each tai- deals with core activities of governments, set-
lored to a specific context (and all critical of ting up authority relations to back up the ideas
traditional policy analysis). In addition, Hill of the policy principles so that they may be
and Hupe (2002) provide a general discussion carried through authoritatively. Second, it dis-
from the angle of implementation, particularly cusses that policy principles do not come from
the tensions between top-downers and bottom- an empty space, they are based on in-depth
uppers (explained below). knowledge of the affairs the policy aims at reg-
The discussion will be selective, it is not ulating. This knowledge may come from gov-
possible to digest all types of network policy ernmental or external sources, but it is brought
analysis within one short chapter. We have together in the contents of the policy. And
omitted the trends towards a transnationaliza- third, the literature expects the policy to aim at
tion of domestic policies, which has been due solving a number of important problems
to international regimes, like the EU, the UN, within the target area, thus creating some sort
the WTO etc., discussing how policies are of order in that segment of society. In the end,
negotiated in complex settings involving many the policy may not be successful, but still,
actors, including various NGOs; see for exam- problem-solving is an important aspect of the
ple Linkage Politics (Rosenau 1969). Following general understanding of policy.
patterns towards institutionalization within Mostly authority, expertise and order has
the EU, there has been a merge of literature on been dealt with in the orderly fashion brought
intra- and inter-state relations, to some degree about by a top-down perspective, using a
captured by the concept of multi-level gover- sequential model of policy-making. Policy is
nance (Hooghe and Marks 2001). We also created, decided upon and implemented step
ignored the evaluation literature which, of by step by collecting information, weighing the
course, is relevant for methodology within pros and cons of various possible ways of
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 99


acting, and then deciding on the course of execution of the policy, executing it, and
action that – in the vein of Pareto-equilibrium – evaluating the results.
will provide most people with most happiness The feedback elements of the model give it a
for the lowest costs. Public (sub)agencies then dynamic feature, and Dror stresses the demands
execute the policy without much further ado. for iterative processes. He also leaves room for
This model, often named rational,2 consti- “extra-rational” behavior based on limited
tutes a core in the sequential model of policy- resources, uncertainty, and lack of knowledge as
making (see chapter by Charles O. Jones), well as creativity and intuition (Dror 1968,
a model with good heuristic qualities, and a 157–158), but the aim of the model is to limit
model that fits the picture which has domi- the importance of such elements in order to
nated constitutions separating politics and enhance optimal policy-making – understood
administration, as well as the minds of man- as “one that is not distorted by the noise that is
agers, and their supporting management con- in fact inherent in all, and especially complex,
sultants and also much of the literature on structures” (Dror 1968, 200). The task, then, is
management. It is a model of leaders being in to organize processes so that at least one unit
control at the apex of the organization, from contributes to each phase, and so that the con-
where they can design the processes desired to tributions of various units add up to an overall
obtain the goals of the organization. A good optimal operation at low costs and with little
example of how these lines of thought have distortion. However, there is no one single
been used in the literature is provided by model for organizing – one may use hierarchy
Yehezkel Dror, who in 1968 published his or polycentric structures in various forms,
Public Policymaking Reexamined (Dror 1968), depending on the demands of the situation. The
followed in 1971 by two companion books judgment of success or failure rests on the con-
(Dror 1971; Dror 1971) to substantiate some tribution of the participants to the process, not
of the contentions of the first book. His aim to a particular organizational form.
was twofold: to advance the study of policy- In other words, Dror does not subscribe to a
making and to contribute to the improvement monolithic hierarchy. Nevertheless, he empha-
of public policymaking – which lacks the sizes the need for overall systems management,
proper use of knowledge. metapolicy-making and comprehensive public
Dror’s optimal model has three major stages policy-making in order to promote adjustment
(Dror 1968, 163–196): Metapolicy-making, and take advantage of new knowledge, and to
policy-making and post-policy-making, and prevent sub-optimization by single units. The
within those there are eighteen sub-stages, one key to such a demand is better personnel: pro-
of which is continuous communication and fessional staffs, units to survey and retrieve
feedback channels interconnecting all phases. knowledge, and units for policy-oriented
Metapolicy-making involves seven stages of research. This form of manpower is to be sup-
processing values, processing reality, process- ported by computerized systems and it must
ing problems, developing resources, designing be managed in new ways (this is 1968), “in
the policy-making system, allocating problems, order to stimulate interprofessional teamwork
values and resources, and finally determining and creativity” (Dror 1968, 274). In addition,
the policy-making strategy. Policy-making there must be some systematic evaluation and
involves another seven stages of suballocating learning feedback from experience.
resources, making and prioritizing operational The model, then, relies on our capabilities to
goals, ditto for other significant values, prepar- produce knowledge based on science, and to
ing a set of major alternative policies (includ- feed it into the policy-making process in order
ing some “good” ones), predicting benefits and to enhance enlightened choices within a com-
costs of those policies, identifying the best prehensive system, and in order to avoid incre-
policies in that light, and then deciding mental policy-making (see next section) which
whether the best alternatives are “good” poli- in Dror’s opinion amounts to nothing but con-
cies. Post-policy-making involves motivating the servatism in disguise. In a later edition of the
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 100


book (Dror 1983), Dror has added an making a comprehensive analysis of all possible
introduction in which he laments the lack of means to obtain an end, the administrator
advance of the policy sciences in the direction resorts to comparing only a few which often do
he has recommended. He also acknowledges not deviate much from past uses, and the one
that active participation in governmental roles selected is the one that creates agreement
has, in the meantime, taught him some impor- among the participants in the policy-making
tant lessons about policymaking: “Social process, no matter what their ideological
science studies from the outside do not pene- standpoints might have told them to do.
trate into the realities of central high-level For the purposes of this article, Lindblom’s
decision making”, and “… ominous policy- discussion of how agreement comes about is
making weaknesses are built-in into core com- crucial. It is not a long or even deep analysis. It
ponents of governance, with present policy is a short, nearly an ideal type description of
predicaments overtaxing maximum policy- how almost every interest in the USA has its
making capacities.” (Dror 1983, x–xi). The watchdog, and that in the formation of a policy
problem he faces is that of research and advice a process of mutual adjustment takes place
versus politics of all sorts. The original book is among various interest groups and public agen-
based on the ideal of science as an integral part cies; and even though all these actors may not
of the desired model optimal policy-making, have an explicit focus on a particular policy
and the political dimensions were not treated goal, the result of the processes will be a viable
in-depth – a problem Dror did not solve, policy. Thus there is no comprehensive income
no matter how many times he paid heed to policy in the USA, but “a process of mutual
other sources of information, including extra- adjustment among … (various actors) …
rational forms3. accomplishes a distribution of income, in which
particular income problems neglected at one
point in the decision process becomes central at
Mutual adjustment in policy-making another point.” (Lindblom 1959). Furthermore,
policies are not made once and for all, but
This line of argumentation in the policy litera- changed and adapted in a never-ending and
ture is concerned with the empirical character- continuous process in which those who lost at
istics of the policy process in a political setting. one point may gain at another. Moreover, since
Analysis of policy cannot be understood in changes are incremental, losses (and gains) for
isolation from the ways politicians, administra- each policy process are endurable.
tors and representatives of interest in society at The underlying understanding of this
large interact about themes of common inter- process is one of a large number of actors, con-
est. One core argument, promulgated by tinuously interacting about a host of themes,
Charles E. Lindblom, is that the information rarely coordinated by any central agency, but
rendered in and by such processes has as much rather performing according to some analogy
value as information produced by researchers of the hidden hand of the economic market.
and other experts. So, where proponents of the Lindblom indicates this without really concep-
rational model recommend problem-solving tualizing it in footnote number 7 in the article:
based on the authority of expertise, followers of “The link between the practice of successive
mutual adjustment advise problem-solving limited comparisons and mutual adjustment
based on the authority of agreements reached of interests in a highly fragmented decision-
among interested parties. making process adds a new facet to pluralist
Lindblom’s most famous text is, undoubt- theories of government and administration.”
edly, “The Science of Muddling Through” Lindblom expanded this line of thinking in his
(Lindblom 1959), originally published in The intelligence of Democracy (Lindblom 1965)
Public Administration Review, but reprinted in with the subtitle Decision making through
numerous Readers. The message is relatively mutual adjustment. The book sets the tone on
simple, but also highly contested; instead of page 3: “… people can coordinate with each
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 101


other without anyone’s coordinating them, they differ sharply in their interpretation: the
without a dominant common purpose, and rational model subscribes to comprehensive
without rules that fully prescribe their rela- uses of scientific knowledge, whenever possible,
tions to each other.” the model of mutual adjustment puts science on
The principles of this argument are found in a par with any other type of knowledge. This
a much earlier paper from 1955, “Bargaining. does not mean that Dror’s model ignores other
The hidden hand in Government” (Lindblom means of acquiring knowledge, but any infor-
1988), and it is a largely un-referenced, but mation should be put into a context of priorities
insightful, discussion of how bargaining coordi- set beforehand. The model of mutual adjust-
nates policy, how it takes place in and among ment does not rely on pre-set goals, but on
public agencies, and how actors are motivated agreement acquired during the process.
for that particular behavior. The key is that no Both models are created for Western demo-
one trusts hierarchy to bring forward “every fact cratic and pluralistic societies. Therefore, they
and value favorable to him. We want a social both contain elements of communication and
mechanism in which every man can speak for interaction which are useful for our subse-
himself or find someone to speak for him.” So quent discussion of networks. However, their
bargaining involves actors and brings forward understanding of how to play a role in a demo-
more aspects to a matter. In other words, the cracy is quite different. One is based on tech-
policy-making process is a matter of politics in nocratic knowledge, depending on how
the broadest sense, and in politics there is not politicians allow it to be expressed. The other
only one truth available. Researchers mostly fol- one is based on knowledge in the demos,
low the political master designated by the hier- depending on how it may express itself.
archy, but other parties interested in the matter Rational models are often seen as command-
may contribute with other views. and-control systems, featuring the (democratic
In the quote above, we find one clue to and elected) top. Dror does not subscribe to
Lindblom’s subsequent career of advocating such a view, but recommends interaction
for pluralism in policy analysis. There is between stages and between actors in the
more, of course; the arguments are unfolded in process – within the frames of goal-setting. The
The Intelligence of Democracy and used in model presupposes that the politicians ulti-
Lindblom’s and Cohens’ Usable Knowledge mately are in control of the bureaucracy and
(Lindblom and Cohen 1979). One basic hence, in Dror’s terms, they control meta-
message is that there is no privileged knowl- policy-making. The bureaucrats provide politi-
edge in the policy process,4 and another is that cians with documentation for any verifiable
the process can only be successful if agreement statement and they substantiate that all relevant
(not only compromise) is reached: then the information has been scrutinized. In turn, the
process has acquired a rationality which serves politicians are controlled by the voters at the
a democratic solution. general elections and by the watchdog function
provided by a free press. So, the rational model
is also to be applied in a pluralistic setting.
The two models compared The model of mutual adjustment is basically
one of interaction, but the number of actors is
The two models are, indeed, adversaries. an open question, dependent on the democra-
Dror explicitly renounced incrementalism, tic procedures of society. It requires a pluralis-
and Lindblom, of course, wrote to warn tic society and a political system that allows
against any belief in the rational model. Dror is various societal interests to enter the policy-
not a rationalist in the classic sense, but his making processes and participate with a
model should be seen as an approximation to prospect to win attention and influence now
rational decision-making. and then. Who exactly will win and when is
The models share an interest for the role of then an open (empirical) question. These
knowledge in the policy-making process. But conditions should be fulfilled in a polity in a
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 102


pluralistic society like the USA. But 18 years used the systems approach and wanted to
after his seminal article (Lindblom 1959) was explain the outputs and outcomes of the polit-
published, Lindblom conceded in a much ical system without really analyzing the con-
acclaimed book (Lindblom 1977) that there tents of the policy processes (Sharkansky and
might be a bias in the policy system which Hofferbert 1969). But empirical analyses com-
provided certain actors with more clout than ing closer to the dynamics of the policy process
others – in the American case, big business. In led to the conclusion that focus on the out-
a later book (Lindblom 1990), Lindblom stated comes of single organizations like the legisla-
that, although imperfect, he saw no alternative ture was not really helpful: ideas and principles
to pluralism; instead the challenge was to cope in parliamentary law were often changed
openly with the problems to reduce adverse during processes of adaptation in the executive
consequences as much as possible. branches and in implementing organizations
at the regional and local levels.
Pressman and Wildawsky’s examination of the
fate of a federal program in a local setting is a
classic example (Pressman and Wildawsky 1973).
One main explanation of the changes was the
Policy analysis has its main roots in American long chain of decision-makers from Washington,
political science – with a little help from DC, to Oakland, CA, which they called the deci-
friends in economics and sociology – and in sion path, and they viewed each decision-maker
public administration, which, however, for as a relatively autonomous actor who could in
most policy purposes itself is rooted in politi- effect block progress. In a second edition in 1979
cal science. Political science developed a strong (Pressman and Wildawsky 1979), Wildawsky
platform in the 1950s and 1960s based on wrote a new chapter on “Implementation in con-
various versions of systems analysis – David text”, and referred to Hugh Heclo’s use of issue
Easton (e.g. Easton 1965) and Karl Deutsch network as a heuristic device to understand how
(Deutsch 1963) are examples of mainstream policies were coordinated.
thinking in the field. And, regardless of the Heclo had coined the term network much
potentials for other ways of doing analysis in, earlier. In a review article on policy analysis he
for instance, Deutsch’s cybernetic ideas, politi- wrote that one should be careful “not to reify
cal scientists focused their interest on organi- collectivities into individual deciders but to
zations within the political systems, often understand the networks of interaction by
conceptualized as institutions, meaning inter- which policies result” (Heclo 1972, 106), and
est organizations, political parties, parlia- he recommended analyzing within programs
ments, the executive, local government and (instead of analyzing organizations). This he
other organizational forms of political life. did himself in Britain, in Sweden and in the
Their aim was to theorize about these compo- USA, research which led him to core concepts
nents of the political system – an example of within policy analysis: policy communities and
such a partial analysis is Sjöblom (1968) on issue networks (Heclo and Wildawsky 1975).
political parties in a multiparty system, Policy communities were more stable interac-
strongly influenced by David Easton and tion patterns among policy interests, issue net-
Anthony Downs (1957). works were mostly ad hoc mode.
However, most policy analysts were not so Heclo was not alone in such research. In a
interested in theorizing about components number of research settings, scholars were
within the political system. The systematic searching for theoretical and conceptual solu-
policy movement started in the second half of tions to their observations of multiple actors
the 1960s (e.g. Ranney 1968) and became a interacting in policy formulation and imple-
thriving field in the 1970s, first of all as policy mentation. Many of them share empirical
implementation research. Many policy analysts observations, but their point of departure in
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 103


various disciplines means that their analytical micro-theoretical foundations in economics

interests and concerns differ. (Williamson 1975) and sociology (Selznick
Within research on interest organizations 1957). We shall return to institutionalism below.
and their relations to the state, the term neo- Planning researchers found a need for
corporatism (Schmitter 1974) was created to conceptualizing coordinators in town planning
indicate a particular and generalizable pattern of based on multiple agencies in local govern-
interaction in society, giving industrial interests ment. One conceptual solution to these find-
in a crucial role in politics, but without much ings was the invention of the reticulist (Friend,
formal representation in decision-making Power and Yewlett 1974) as an actor that links
bodies, and mostly without formally delegated other actors together in networks. These
powers. This was in contrast to corporatism authors drew on organization theory, whose
proper (as was the case in Fascist Italy), where practitioners observed inter-organizational
organized interests would have formal state phenomena in many settings. Some were seen
powers, Schmitter’s ideas were followed up on by to reduce the importance of market relations
various projects which lead to theorizing about and hence a break with some elements of eco-
the segmented state or state sectors, indicating nomic theory of the market. Examples were
much of what Heclo had termed policy commu- interlocking directorates, where corporations
nities. But there was an important difference in shared a number of individuals on boards of
their view on the degree of integration within directors and hence were able to coordinate
the networks. While Heclo, Wildawsky and policies (Pennings 1980); an early and socially
others supported a pluralist view of politics and broader oriented example of this line of think-
hence looked for alternation in the importance ing was Wright Mills’ book on the Power Elite
of actors within the network, researchers analyz- (Mills 1956).
ing policy sectors worked within a tradition Other relations were seen as variations in
looking for closed interrelations among actors. features of the organizational society where
Schmitter’s ideas became very influential in private organizations communicated with one
research in North European countries, primar- another about common purposes and engaged
ily regarding relations between interest organi- in new relations with the state in order to
zations and the state. In Europe he influenced influence public policies. And, likewise, since
several research agendas regarding collective the state engaged in more and more policies
action and interest organizations (Czada and and programs that would affect various orga-
Windhoff-Héritier 1991) as well as the border- nizational interests, it had concerns and needs
line between public and private (Streeck and for coordination which could be satisfied by
Schmitter 1985). In Norway, a research program better communication with organized inter-
on power led to theorizing about new seg- ests. As a consequence, the state and private
mented forms of state power within policy organizations became interdependent, and
sectors, with voters and the parliament in less there was a need to conceptualize the relations.
prominent positions than the constitutional One line of such inter-organizational research
design would lead you to think (Olsen 1978), was based on resource exchange as the medium
and government, administration and interest for sustaining interorganizational relations, but
organizations in strong positions. Within the focus was on the macro-aspects of exchange;
broader social theory, Norwegian researchers one influential source was Benson (1975) who
coined the phrases of the negotiated economy used the (Marxist) logic of substructure and
(Hernes 1978), a concept indicating that market superstructure from political economy to tease
forces were replaced by negotiations between out basic forces like money and authority, which
social organized interests and the state (Pedersen were then brought into play in a superstructure
and Nielsen 1988). These results led to an of organizational interactions. He developed his
increased interest in analyzing institutional first model into an analytical model of a two-
aspects of society, based on a mix of macro- and leveled policy sector – understood as a subset of
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 104


a large number of interorganizational networks one another for updating and development of
in society (Benson 1982). In Europe, Rod their understanding of the environment.
Rhodes used these and other sources in organi-
zation theory as inspiration for developing his
models of state-local government relationships
(Rhodes 1979; Rhodes 1986) which he contin-
ued to use for an extensive discussion of policy
networks and policy communities as organizing
factors in British politics and administration Above we inspected some of the roots of policy
(Rhodes 1997). network analysis and its development in the
Another line of inter-organizational research 1970s and early 1980s. We shall now discuss the
had a micro focus and could be said to have subsequent development of analytical pers-
some inspiration from literature dealing with pectives, which include traditionalists, institu-
increased division of labor in society. How tionalism, governance, and trends towards
can such sectors be understood? Most ratio- deliberative discourse analysis. They have devel-
nales in such analysis are based on theories oped historically, of course, so they overlap, and
of public choice, which lead to the conception to some degree they both react to and build
of a service industry (Ostrom and Ostrom upon one another in the sequence of institu-
1977; Ostrom, Parks and Whitaker 1978) – tionalism in the 1980s, governance in the 1990s
which had many common features with a and deliberative analysis in the late 1990s and
policy sector. The basic idea was to counter now under further development in the 2000s.
theories of (large) bureaucratic organization Traditionalists were found all the time – but
by theories of (small) organizational coopera- some of them changed with the currents.
tion, making a case for small-scale govern- What were those trends about? Grossly over-
ment and governmental agencies which would simplifying, one can say that there has been a
pool resources for larger tasks, if necessary. move from system and hierarchy (rational
The basic ideas were developed into game models) towards fragmentation and empower-
theoretical frameworks and applied in various ment (mutual adjustment). Institutionalists
forms of self-government (Ostrom 1990) and were concerned with how political systems fared
in intergovernmental relations in Germany and they worked to re-conceptualize the mod-
(Scharpf 1997). ernistic state apparatus into something less
Both the resource based and rational choice monolithic in processes involving various stake-
models were presented in an often-quoted holders in society. Governance scholars contin-
anthology on inter-organizational policy- ued this work and conceptualized the workings
making in 1978 (Hanf and Scharpf 1978). It of various parts of the systems and helped us
fulfilled at that time the need implementation understand better how network policy processes
scholars had for analytical models, which at took place. Discourse and deliberation scholars
the same time caught interaction among levels cashed in on further changes in society towards
of administration as well as at each level. It was involving citizens in policy processes, and they
to be the first volume of the soon after rapidly also were part of the general movement among
growing literature on the fragmentation of the some social scientists towards social construc-
state apparatus. The fragmentation was due to tivism and pragmatism.
decentralization of powers to lower levels and These three forms constitute some of the
sharing powers with various organizations in “forefront” in research during those years. But
the “gray” zone. This created new and intensi- that is not to say that every one participated.
fied possibilities to exercise influence on sepa- Of course, many policy analysts proceeded in
rate decision-makers. Following this tendency, more traditional veins and challenged the new-
the borders between public and private tend to bees, or approached the new ideas without
become blurred, and the exchanges of infor- buying them wholesale. So, first, we’ll review
mation make the various actors dependent on some of the main arguments among them.
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 105


Traditionalists (2002, 41–84). Several attempts to create a

compromise between the two schools have
The essence of the development of traditional been made over the years, and even the most
policy analysis is caught by referring to the ardent proponents of either side have con-
struggles between top-down and bottom-up ceded that a pure paradigm is not tenable
analysis. In many ways, this was a discussion (Sabatier 1986; Hjern and Hull 1987). There is
between the rational model and muddling some authority present in most systems, but it
through, between Dror and Lindblom – with- may be dormant – negotiations take place “in
out their being present in direct confrontations. the shadow of hierarchy” (Scharpf 1997,
An example of the top-down perspective 197–205). Whatever the case, most of those
is given by a model of Intergovernmental taking part in the discussion on top-down
Implementation by Van Horn (1979, 15). The versus bottom-up did fairly traditional analysis
problem is to get National priorities imple- in methodological terms; they used statistics,
mented at the local level, and the remedy is, first interviewed actors and followed the main-
of all, to get clearer policy goals and standards at stream tradition in their empirical analysis.
the federal level – the more specific, the better. Several of them also took part in the discus-
This must be supplemented by better, i.e. clear, sions leading to the new institutionalism,
accurate, consistent and timely communication. which is the subject of the next section.
All these elements are part of the rational
model, as is the distinction between policy and
implementation, which is maintained. However, New institutionalism
Van Horn does recognize that local attitudes of
political actors and interest groups are impor- The most dominant trend of the 1980s
tant, as are the skills of agencies and the need for involved new institutionalism. There are several
adequate resources. The policy problem is how versions within policy analysis – and many
to overcome such hindrances for successful more outside, which we shall ignore. Most of
implementation. One can find similar under- them share a dissatisfaction with the American
standings of the policy process in the literature behavioral revolution (Easton 1953; Simon
(Bardach 1977; Mazmanian and Sabatier 1983). 1945; Truman 1951), but they have different
This way of understanding the policy cures for the malady. The main distinction
process was countered by various scholars, relates to micro- and macro-perspectives on
claiming that enhanced control from the top actors, respectively (Scharpf 1997; March and
simply would not be enough. One must under- Olsen 1989). Many of the attempts to theorize
stand what is going on among the various about policy networks ended up with delineat-
agencies, and such understanding cannot be ing some sort of subsystem, probably with
won by focusing on the top, one has to unwrap some inspiration from Heclo (1972) and his
what goes on locally (Hjern and Hull 1982; predecessors in American analysis of sub-
Hjern and Hull 1984). The critics developed governments (e.g. Lowi 1964). One theme was
the concept of an implementation structure the degree of autonomy policy networks
(Hjern and Porter 1983), an analytical tool to enjoyed vis-á-vis more inclusive systems like
map the interactions between actors involved the political systems (Lehner 1991; Rhodes
in the policy process, inspired by – among 1986). Another theme concerned the policy
others – Elmore (1979) and Lipsky (1980). network as such: how was it organized, how
A core dictum was that actors, a priori, should were powers distributed (Scharpf 1991; Rhodes
be put on a higher position in the policy and Marsh 1992). A third theme concerned the
process if one were to truly realize what goes role of networks at a societal level: How could
on – namely the creation of a policy network one understand the politics and administration
instead of a system of authority. of societies with many policy networks
A thorough mapping and discussion of the (Lehmbruch 1991; Campbell, Hollingsworth
various positions is found in Hill and Hupe and Lindberg 1991)?
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 106


Whatever the case, policy analysts found thus consciously waiving their formal autonomy
that the formal organizational system of govern- in such fields, replacing it by mutual depen-
ment often did not adequately describe the dence. Structuralists likewise asked themselves
patterns of interaction they found in policy about the role of institutional settings, but their
formation and implementation. Moreover, the interest was, more or less, to make plausible a
alternative, “American” behavioral analysis, claim that organizational actors do not decide as
lacked a foundation in or a link to what orga- rational actors, they rather follow standard
nizations meant in political life. If one struc- operating procedures, and normative facets of
tures the field in terms of Richard Scott’s three the organization as they appear in myths, sym-
types of institutional theory – regulative, nor- bols and even rituals – in short, the organiza-
mative and cognitive (Scott 1995), three types tional culture which would define appropriate
of questions interested policy analysts. First, behavior of actors. Therefore, organizational
they found themselves confronted with ques- factors would be important, but in other ways
tions of what systems of rules that might really than traditional political science had used them
apply to the actions of both organizations and in the past (March and Olsen 1989).
individuals. Furthermore, they asked them- The main difference between the two types
selves what social obligations the actors of new institutionalism, then, was rooted in
wanted to adhere to when they set standards opposite hypotheses about the behavior of
for future policies, or when they implemented actors. In addition, there were strong norma-
policies in ways that were not always in close tive differences, in that most rational theorists
accordance with the stated, formal policy prin- did not much care about how services were
ciples. Third, they found a need for identifying provided and therefore might advocate for pri-
norms for proper behavior in networks across vatization; structuralists to a much greater
organizational boundaries – how did actors extent adhered to maintaining the particular
perceive one another, and how did they come democratic values provided by public sector
to terms when their organizational back- organization of services. In policy terms, this
grounds differed? became very visible in normative discussions,
Such questions are to some degree answered e.g. about the pros and cons of new public
by various institutional theories. Scott’s distinc- management (Hood 1991; Barzelay 1992).
tions were not part of the discussions of policy
scholars in the 1980s, at least not explicitly, so
they articulated their institutional theories dif- Governance
ferently. There was a relatively clear cleavage
between scholars working on the basis of ratio- A second major trend came in the 1990s, and its
nal choice theory and those who were more theme was labeled governance. In many ways it
interested in structural analysis. In a way, their was a natural sequel to the focus on institution-
interest was much about the same. They realized alism in the 1980s. There was an enduring com-
that it would be no use only to focus on formal petition between macro- and micro-analytical
organizations like parliaments or bureaucracies approaches to conquer the right to be called new
to analyze policy processes. Rational choice institutionalists (Selznick 1996), and there were
theorists then asked themselves how variations tensions between new and old institutionalists,
in structural conditions would affect various to say nothing about those who still saw formal
types of rational actors. Examples are various organizations as institutions (Aberbach and
ways of organizing the police force (Ostrom, Rockman 1987). So the more the field of
Parks and Whitaker 1978), or ways of organiz- analyzing policy networks developed, the more
ing local governments in metropolitan areas the search for more adequate concepts intensi-
(Oakerson 1987). The rationale behind this was fied. Increasingly, the concept of governance
that small organizations could be effective if gained momentum: it could be seen as some-
they cooperated with other ones about certain thing other than government, and it had a
tasks in a rational way, based on self-interest, processual flavor to it.
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 107


Nevertheless, governance turned out to get refers to self-organizing, inter-organizational

some comparable problems to institutional- networks characterized by interdependence,
ism. It has become a somewhat fuzzy concept, resource exchange, rules of the game, and sig-
covering a vast territory and therefore maybe nificant autonomy from the state.” (Rhodes
less useful as a discriminating concept. Rhodes 1997, 15, italics in original). Such a definition
(1997, 47) refers to six meanings: the minimal sets rather clear boundaries for the analytical
state, corporate governance (of enterprises), interest of scholars, and it invites a specific way
new public management, “good governance” of theorizing, based on inter-organizational
(for developing countries), socio-cybernetic assumptions. It puts networks into the center
system (overall characteristics), and self- of our analytical interest, and other forms of
organizing networks. More categories proba- governing are, consequently, left out of sight.
bly can be found. But let us venture to capture Policy scholars have taken part actively in
some core meanings which will then form the the development of governance theory. Early
backbone of the discussions in this chapter. on, their empirical findings pointed to prob-
The most general use of governance covers lems with traditional political theory in
new forms of government-society relations – as explaining what went on in policy formation
an example the socio-cybernetic system men- and implementation. Their findings on policy
tioned above. This comprehensive interpreta- networks called for alternatives to the received
tion of governance suggests that the principles view of the modern state. It became very clear
of modern society, with its division of labor when facets of policy networks were discussed
between state, market and civil society, is under in a management perspective – here the obvi-
siege and, in particular, hierarchical state- ous lack of traditional control instruments
society relations are being replaced by other belonging to the manager of the closed organi-
forms of interrelationships, which often imply zation (Gage and Mandell 1990; Kickert, Klijn
some “co”-action between public and private and Koppenjahn 1997); the primary role of the
(Kooiman 1993, 4–6). Such an interpretation network manager then becomes to facilitate
invites us to reconceptualize modern theories of communication.
the state; there is little agreement about how to
do this, examples are theories of reflexivity
(Beck, Giddens and Lash 1994) and of post- Deliberative policy analysis
modern conditions (Bogason 2000; Miller
2002). Such theoretical constructs open up The third major trend in policy analysis began in
possibilities for understanding the state as a the 1990s and is gaining momentum in these
network mingled with the greater society and, first years of the third millennium. It is very
consequently, political action changes in its comprehensive since it involves both theory and
meaning. Analytical interest goes away from a methodology, not to say foundations of social
focus on parliamentary and bureaucratic science. It concerns deliberation and discourse in
processes of negotiation, and instead scholars policy processes, and thus it has one leg in the
identify interaction patterns between various governance tradition, but it also reflects some-
interests, the results of which then get recogni- thing more. Echoing the linguistic turn in the
tion as public policies. The precise organiza- philosophy, one signal was the publication of the
tional pattern is not defined, it is an empirical anthology entitled The Argumentative Turn in
question within a dynamic system, much like Policy Analysis and Planning (Fischer and
Giddens’ ideas of structuration (Giddens 1984). Forester 1993), whose editors were inspired by
A second and related, but less comprehen- Deborah Stone (1988) to state that “policy-
sive meaning of governance, implies only the making is a constant discursive struggle over the
fall of clear organizational boundaries of criteria of social classification, the boundaries of
public and private organizations, and the problem categories, the intersubjective interpre-
wider context (like state theory) is not really tation of common experiences, the conceptual
addressed. One example is that “governance framing of problems, and the definitions of ideas
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 108


that guide the ways people create the shared research practices: “… a close practical and
meanings which motivate them to act.” So: conceptual connection exists between a post-
“Policy analysis and planning are practical positivist policy analysis and today’s decentered
processes of argumentation.” (Fischer and world of governance” (Hajer and Wagenaar
Forester 1993, 1–2). 2003, xiv).
No reference to networks in those sentences, This statement, then, reflects that, in most
but of course the development of policy analy- Western countries, the public sector has been
sis towards networks had unveiled processes opened up for more participation in policy
which were hidden in the organizations of the processes. One may doubt the sincerity of this
1960s and before. So they were closed to the (Miller 2002, vii–viii), but measured on the
type of scholarly scrutiny which was, after all, surface – by the sheer growth in the number of
easier to perform in the networked policy new channels for participation – this is a fact
processes of communication researchers followed (OECD 2001). In policy analysis, this has con-
in the 1970s and 1980s. And, sure enough, sequences for the role and use of expertise
about half of the articles in the anthology (Fischer 1999) which increasingly becomes
discuss various forms of deliberation in the part of an ongoing discourse with less and less
policy process, and hence indirectly network elevated status for policy analysts; instead they
settings. have to make their points of view understood
Discourse thus relates to language, and a by a broader public. Deliberation also means
primer on narrative policy analysis was written that organized interests get more legitimate
by Emery Roe (1994). But there are also roots access to the policy process, but in the light of
in institutionalism: “From this perspective acid the research on institutionalism and gover-
rain is a story-line that, potentially, brings out nance, that is hardly surprising. But the conse-
the institutional dimensions of the ecological quences for the roles of ordinary citizens may
problematique.” (Hajer 1995, 265). This means be more profound, in that citizens get access to
that the author has an analytical interest in participate in ways that earlier on might have
how discourse is structured or embedded in been seen as counterproductive to an efficient
society, while at the same time it structures public sector. Some of the development may be
society – in other words, not unlike Giddens’ conceptualized as empowerment of citizens
ideas of structuration, which has the concept (Sørensen 1997); an interesting research ques-
of institution at the core of the analysis tion is to what degree formal rights to parti-
(Giddens 1984). cipate actually are brought into use for
Speaking metaphorically, the deliberative influencing policy decisions. If that is the case,
policy analysis brings the scholar down from other researchers speak of a strengthening of
the ivory tower to the people. The institution- social capital in society (Putnam, Leonardi and
alists and most governance theorists kept the Nanetti 1993). In more radical versions, one
privileged status of researchers to analyze cur- can say that the citizens decide about the
rents in society and to work for a better theo- future of their communities (Ostrom 1995)
retical understanding of how policy came instead of, for example, relying on a benign but
about. But the 1990s gradually saw changes in bureaucratic welfare state.
the social sciences, which meant that the privi- The research, then, stresses the features of
leged and isolated status of scholars was meant deliberation, dialogue, collaboration and
to be revoked, and their roles to be changed mediation. Much of it should be understood as
from observers to participants in research part of the scientific development towards
processes that stressed dialogue instead of postempiricist social science. Empiricists, or
observation and reporting (Guba and Lincoln traditional policy analysts, have tried to mini-
1989; Erlandson, Harris, Skipper et al. 1993). mize social and interpretative judgments,
Thus, the advocates of change mirrored soci- postempiricists recognize their basic, constitu-
etal developments towards more public partic- tive role in any form of analysis (Fischer
ipation in policy processes in their own 2003, 226). Postempiricist policy analysts do
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 109


not speak truth to power, they collaborate with more and more complicated, or at least
power holders and mediate with diverse inter- comprehensive, policy process everywhere in
ests. In that sense, they have become part of the the Western world. The point, then, does not
network society. concern the exact definition of the phenomena
under scrutiny, but the general recognition
that we are observing qualitatively different
policy processes. The challenge is to show how
they differ from the past, and what measures
then should be taken.
Where is network analysis at now? We have What, then, may we find in common among
followed a historical trajectory from rational the different camps of policy network analysis?
policy analysis to analysis integrating a broader It seems to me that one main distinguishing
conception of the processes involved. So feature of the advanced policy analysts of today
scholars involved in such analysis have all par- is that they apply a new version of pragmatism.
ticipated in a battle between rational decision- The classic Deweyan pragmatist was interested
making and muddling through – maybe not in in theory as a vehicle for promoting change in
the open, and maybe not even as a conscious societal affairs. The pragmatist of today has less
choice by the analyst, but still the theme of faith in theory. To put it crudely, pragmatists of
doing things rationally or not pops up every- today are interested in conceivable practical
where. At the very least, as a pedagogic means to consequences of affirming an idea or taking
illustrate what we are not doing. More seriously, an action – consequences that are satisfying
as a theme that has to be addressed in order to and desirable in the light of power relations
persuade the reader that rationalism is or is not (Cherryholmes 1999, 124–125). They follow a
applicable in this case – and it seems that ratio- pragmatism which is anticipatory and hence
nalism is on the decline.5 That is not to say that inductive and fallible; today’s pragmatists con-
Lindblom’s mutual adjustment is the only struct their reality socially and perform analysis
answer to non-rational demands. But his ideas critically; they are skeptics and hence not believ-
are hovering over many of the solutions we face. ers of a final Truth. They see the world as con-
The outline given above about the changes tingent, and thus they are contextualists. They
in policy analysis stresses a transition towards are holists and reject distinctions like fact/
network analysis, but that does not mean that value, objective/subjective, theory/practice, ends/
former types of policy analysis are gone. means, analytic/synthetic. This credo I will call
“Network” is still a debated term, to say noth- the “new pragmatism.” It certainly covers the
ing of network analysis. Have we not seen it all postpositivists within policy analysis, and to a
before? Of course we have in some sense, our degree it covers many other network analysts –
predecessors in political analysis were not of whom some still subscribe to a distinction of
idiots. In a more narrow sense, things are new, fact and value, and of objective and subjective.
but, as Keith Dowding (1994) has shown, the New pragmatists do not see evidence in the
literature then tends to become bogged down classic sense of getting the data straight, prefer-
because of definitional fights between academic ably in some version of statistical analysis. They
camps. So more energy is used for fights than beg the question of the existence of a network
for sensible analysis of one’s own results as well and involve themselves in processes of argu-
as of the results from colleagues. Christopher mentation and power – resource exchange or
Pollitt’s critique that there are tendencies not, “science” or not. They base their action on
towards ahistorical comprehension, and that it some form of hermeneutic analysis, and many
is hardly proven that networks form a new and of them do not mind using supplementary
better type of democracy, are also worth con- information based on some strand of posi-
sidering (Pollitt 2003, 65–67). That said, I tivism. Nevertheless, they see such evidence as
think that one should interpret the focus on one out of many channels of information for
networks and process as a consequence of a their craft. Hoppe characterizes two types of
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 110


analysts which I would count under the new NOTES

pragmatists: Forensic policy analysts and partic-
ipatory policy analysts (Hoppe 1999, 207–208). 1. The author wants to thank Anders Berg-Sørensen,
The forensic policy analysts see a cacophony of Allan Dreyer Hansen, Eva Sørensen, Jacob Magnussen and
competing thinking styles, ideologies, para- Peter Triantafillou for helpful comments on a draft version.
digms, perspectives, etc. in policy analysis, and 2. Not all users of sequential models adhere to the
strictly rational version, though. For instance, Wildawsky
hence they advocate for first distinguishing
wanted us to speak truth to power, but his own model of
between the various sorts of frames of thinking decision-making was less demanding than the rational
that can be found pertaining to a policy version, he was closer to incrementalism.
problem. Then they want to create a new sort of 3. Dror is no naivist. In the second edition of his book
frame, combining plausible and robust argu- (Dror 1983), he makes explicit his history of learning as an
Israeli scholar and a Zionist in political terms, and he dis-
ments (frame-reflection, following, for example,
cusses the values that come out of such a past, thereby setting
Schön and Rein (1994)) into a new policy what he considers an example for other scholars involved in
design. This may be done with various stake- policymaking. He thus follows the stance that; although no
holders and hence the barrier between analyst one is value free – which would be desirable – one can make
and policy-maker is torn down (Guba and up for it by making values explicit to the reader.
4. From the introduction to Democracy and Market
Lincoln 1989; Erlandson, Harris, Skipper et al.
System: “I cannot think of any human accomplishment that
1993) in a creative process of finding solutions to unambiguously and undeniably could not not have been
the problems at hand. As with Lindblom, the dif- achieved without social science.” (Lindblom 1988, 21)
ferences between means and goals disappear in a 5. If one tries to get an overview of university course
world of continuous change. literature on policy analysis, it seems that rational and
statistical analysis dominates –’s three most
The participatory policy analyst may be in
popular books on policy analysis are Bardach (2000),
agreement with the forensic analyst, but does Weimer, Vining and Vining (1998) and Patton and Sawicki
not stop with the question of how to under- (1993). But in conferences and anthologies purporting to
stand policy frames. S/he broadens the perspec- mirror the state of the art, such techniques do not take
tive even more and emphasizes the importance many pages.
of involving citizens in the policy processes – to
include local knowledge, to make obvious
themes of ethics, or to cut through disagree- REFERENCES
ments among experts. Or, to make up for the
increase in distance between politicians and the
demos, and the strengthening of the adminis- Aberbach, J. D. and Rockman, B. A. (1987)
Comparative Administration – Methods, Muddles
trative apparatus (Fischer 2003, 15–16). Some
and Models. Administration and Society 18(4),
critics argue that the quality of the debate 473–506.
and/or solution is not guaranteed by citizen Bardach, E. (1977) The Implementation Game: What
involvement. The counter question of the prag- Happens After a Bill Becomes a Law. Cambridge,
matists is: Who is to be the umpire of such qual- MA: MIT Press.
ity? For new pragmatists there is no absolute Bardach, E. (2000) A Practical Guide for Policy
standard to use as truth medium. Analysis: The Eightfold Path to More Effective
Analysts following the rational paradigm Problem Solving. New York: Chatham House
cannot possibly agree with this. Followers of Publishers.
Lindblom would feel closer, particularly since Barzelay, M. (1992) Breaking Through Bureaucracy. A
science is not awarded any pedestal, but prob- New Vision for Managing in Government, in collab-
oration with B. J. Armajani. Berkeley: University of
ably they would package this form of truth
California Press.
with a conception of the political process Beck, U., Giddens, A. and Lash, S. (1994) Reflexive
which creates winners. Those winners, then, Modernization. Politics, Tradition and Aesthetics in
are in a position to define the truth for the time the Modern Social Order. Cambridge: Polity Press.
being, that is until some one else comes into Benson, J. K. (1975) The interorganisational net-
power. Cynical, maybe, but political analysis work as a political economy. Administrative
never was for someone with a feeble heart. Science Quarterly, 20, 230–49.
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 111


Benson, J. K. (1982) A framework for policy Society, eds M. A. Hajer and H. Wagenaar, Theories
analysis. In Interorganizational Coordination: Theory, of Institutional Design, pp. 209–27. Cambridge:
Research and Implementation, eds D. L. Rogers and Cambridge University Press.
D. A. Whetten, pp. 137–76. Ames, Iowa: Iowa Fischer, F. (2003) Reframing Public Policy. Discursive
State University Press. Politics and Deliberative Practices. Oxford: Oxford
Bogason, P. (2000) Public Policy and Local Governance: University Press.
Institutions in Postmodern Society, New Horizons in Fischer, F. and Forester, J. (1993) Editors’ introduc-
Public Policy. Cheltenham: Edward Elgar. tion. In The Argumentative Turn in Policy Analysis
Campbell, J. L., Hollingsworth, J. R. and Lindberg, L., and Planning, ed. F. Fischer and J. Forester,
eds. (1991) Governance of the American Economy. pp. 1–17. London: UCL Press.
Cambridge: Cambridge University Press. Friend, J., Power, J. and Yewlett, C. (1974) Public
Cherryholmes, C. H. (1999) Reading Pragmatism. Planning: The Inter-Corporate Dimension.
New York: Teachers College Press. London: Tavistock Publications.
Colebatch, H. K. (1998) Policy. Buckingham: Open Gage, R. W. and Mandell, M. P. (1990) Strategies for
University Press. Managing Intergovernmental Policies and Networks.
Czada, R. M. and Windhoff-Héritier, A., eds (1991) Praeger: New York.
Political Choice. Institutions, Rules and the Limits of Giddens, A. (1984) The Constitution of Society.
Rationality. Frankfurt am Main: Campus Verlag. Outline of the Theory of Structuration. Cambridge:
Deutsch, K. W. (1963) The Nerves of Government. Polity Press.
Models of Political Communication and Control. Guba, E. G. and Lincoln, Y. S. (1989) Fourth
New York: The Free Press of Glencoe. Generation Evaluation. London: Sage.
Dowding, K. (1994) Policy networks: Don’t stretch a Hajer, M. A. and Wagenaar, H., eds (2003) Deliberative
good idea too far. In Contemporary Political Policy Analysis. Understanding Governance in the
Studies, eds P. Dunleavy and J. Stayrer, pp. 59–78. Network Society, Theories of Institutional Design.
Exeter: Shortrun Press. Cambridge: Cambridge University Press.
Downs, A. (1957) An Economic Theory of Democracy. Hajer, M. A. and Wagenaar, H. (2003) Introduction. In
New York: Harper & Row. Deliberative Policy Analysis. Understanding
Dror, Y. (1968) Public Policymaking Reexamined. Governance in the Network Society, eds M. A. Hajer
Pennsylvania: Chandler Publishing Company. and H. Wagenaar, Theories of Institutional Design,
Dror, Y. (1971) Design for Policy Sciences. New York: pp. 1–30. Cambridge: Cambridge University Press.
American Elsevier. Hajer, M. (1995) The Politics of Environmental
Dror, Y. (1971) Ventures in Policy Sciences. New York: Discourse. Ecological Modernization and the Policy
Elsevier. Process. Oxford: Clarendon Press.
Dror, Y. (1983) Public Policymaking Reexamined. Hanf, K. and Scharpf, F. W., eds (1978) Inter-
New Brunswick, N.J.: Transaction Books. organizational Policy Making. Limits to Coordina-
Easton, D. (1953) The Political System. An Inquiry tion and Central Control, SAGE Modern Politics
Into the State of Political Science. New York: Alfred Series, 1. London: Sage.
A. Knopf. Heclo, H. (1972) Review article: Policy analysis.
Easton, D. (1965) A Framework for Political Analysis. British Journal of Political Science, 2, 83–108.
Englewood Cliffs, NJ: Princeton University Press. Heclo, H. and Wildawsky, A. (1975) The Private
Elmore, R. E. (1979) Backward mapping: Implem- Government of Public Money: Community and
entation research and policy decisions. Political Policy Inside British Politics. London: Macmillan.
Science Quarterly, 94(4), 601–16. Hernes, G., ed. (1978) Forhandlingsøkonomi og bland-
Erlandson, D. A., Harris, E. L., Skipper, B. L. and ingsadministration. Oslo: Universitetsforlaget.
Allen, S. D. (1993) Doing Naturalistic Inquiry: A Hill, M. and Hupe, P. (2002) Implementation.
Guide to Methods. London: Sage. London: Sage Publications.
Fischer, F. (1995) Evaluating Public Policy. Chicago: Hjern, B. and Hull, C. (1982) Implementation
Nelson-Hall Publishers. research as empirical constitutionalism. European
Fischer, F. (1999) Technocracy and the Politics of Journal of Political Research, 10(2), 105–17.
Expertise: Managerial and Policy Perspectives. Hjern, B. and Porter, D. O. (1983) Implementation
London: Sage. structures: A new unit of administrative analysis.
Fischer, F. (2003) Beyond empiricism: Policy analysis In Realizing Social Science Knowledge, eds
and deliberative practice. In Deliberative Policy B. Holzner, K. D. Knorr and H. Strasser,
Analysis. Understanding Governance in the Network pp. 265–77. Wien-Würzberg: Physica-Verlag.
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 112


Hjern, B. and Hull, C. (1984) Going interorganisa- Lindblom, C. E. (1990) Inquiry and Change. The
tional: Weber meets Durkheim. Scandinavian Troubled Attempt to Understand and Shape
Political Studies, 7(3), 197–212. Society. New Haven: Yale University Press.
Hjern, B. and Hull, C. (1987) Helping Small Firms Lindblom, C. E. and Cohen, D. (1979) Usable
Grow: An Implementation Approach. Beckenham, Knowledge. Social Science and Social Science
Kent: Croom Helm. Solving. New Haven: Yale University Press.
Hood, C. (1991) A public management for all sea- Lipsky, M. (1980) Street-Level Bureaucracy: Dilemmas
sons? Public Administration, 69 (Spring), 3–19. of the Individual in Public Services. New York:
Hooghe, L. and Marks, G. (2001) Multi-Level Russell Sage.
Governance and European Integration. Boulder: Lowi, T. J. (1964) American business, public policy,
Rowman & Littlefield. case studies, and political theory. World Politics,
Hoppe, R. (1999) Policy analysis, science and poli- 16 (July), 686–715.
tics: From ‘speaking truth to power’ to ‘making March, J. G. and Olsen, J. P. (1989) Rediscovering
sense together’. Science and Public Policy, 26(3), Institutions. The Organizational Basis of Politics.
June, 201–10. New York: Free Press.
Kickert, W. J. M., Klijn, E.-H. and Koppenjahn, J. F. M., Mazmanian, D. A. and Sabatier, P. A. (1983)
eds (1997) Managing Complex Networks. Strategies Implementation and Public Policy. Glenview, IL:
for the Public Sector. London: Sage. Scott & Foresman.
Kooiman, J. (1993) Social-political governance: Miller, H. T. (2002) Postmodern Public Policy.
introduction. In Modern Governance: New Albany, NY: State University of New York Press.
Government-Society Interactions, ed. J. Kooiman, Mills, C. W. (1956) The Power Elite. New York:
pp. 1–6. London: SAGE Publications. Oxford University Press.
Kuhn, T. S. (1962) The Structure of Scientific Oakerson, R. J. et al. (1987) How Fragmentation
Revolutions. Chicago: Chicago University Press. Works – St Louis Style. ACIR: Washington DC.
Lakatos, I. (1974) Falsification and the methodology OECD (2001) Citizens as Partners. Information,
of scientific research programmes. In Criticism Consultation and Public Participation in Public
and the Growth of Knowledge, ed. I. Latakos and Decision-Making. Paris: OECD.
A. Musgrave, pp. 91–196. Cambridge: Cambridge Olsen, J. P., ed. (1978) Politisk organisering. Oslo:
University Press. Universitetsforlaget.
Lehmbruch, G. (1991) The organization of society, Ostrom, E. (1990) Governing the Commons: The
administrative strategies, and policy networks. In Evolution of Institutions for Collective Action.
Political Choice. Institutions, Rules and the Limits of Cambridge: Cambridge University Press.
Rationality, eds R. Czada and A. Windhoff-Héritier, Ostrom, E. (1995) Self-organization and social capital.
pp. 121–58. Frankfurt am Main: Campus. Industrial and Corporate Change, 4(1), 131–59.
Lehner, F. (1991) The institutional control of orga- Ostrom, E., Parks, R. and Whitaker, G. (1978)
nized interest intermediation: A political perspec- Patterns of Metropolitan Policing. Cambridge,
tive. In Political Choice. Institutions, Rules and MA: Ballinger Publishing Company.
the Limits of Rationality, eds. R. Czada and Ostrom, V. and Ostrom, E. (1977) Public goods
A. Windhoff-Héritier, pp. 233–56. Frankfurt am and public choices. In Alternatives for Delivering
Main: Campus. Public Services, ed. E. S. Savas. Boulder, CO:
Lindblom, C. E. (1959) The science of muddling Westview.
through. Public Administation Review, 19, 79–99. Patton, C. V. and Sawicki, D. S. (1993) Policy Analysis:
Lindblom, C. E. (1965) The Intelligence of Democracy. Concepts and Practice. Englewood Cliffs, NJ:
Decision Making Through Mutual Adjustment. Prentice-Hall.
New York: The Free Press. Pedersen, O. K. and Nielsen, K. (1988) The Negotiated
Lindblom, C. E. (1977) Politics and Markets. Economy: Ideal and History. Scandinavian Political
New York: Basic Books. Studies, 11(2), 79–101.
Lindblom, C. E. (1988) Bargaining: The hidden Pennings, J. M. (1980) Interlocking Directorates. San
hand in government. In Democracy and Market Francisco: Jossey Bass.
System, C. E. Lindblom, pp. 139–70. Oslo: Pollitt, C. (2003) The Essential Public Manager.
Norwegian University Press. Maidenhead: Open University Press.
Lindblom, C. E. (1988) Introduction. In Democracy Pressman, J. L. and Wildawsky, A. B. (1973) Imple-
and Market System, C. E. Lindblom, pp. 9–21. mentation. Berkeley, CA: University of California
Oslo: Norwegian University Press. Press.
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 113


Pressman, J. L. and Wildawsky, A. B. (1979) Schön, D. A. and Rein, M. (1994) Frame Reflection.
Implementation. Berkeley, CA: University of Toward the Resolution of Intractable Policy
California Press. Controversies. New York: Basic Books.
Putnam, R., Leonardi, R. and Nanetti, R. Y. (1993) Scott, R. W. (1995) Institutions and Organizations.
Making Democracy Work: Civic Traditions in Modern London: SAGE Publications.
Italy. Princeton, NJ: Princeton University Press. Selznick, P. (1957) Leadership in Administration.
Ranney, A. (1968) Political Science and Public Policy. New York: Harper & Row.
Chicago: Markham. Selznick, P. (1996) Institutionalism “old” and “new”.
Rhodes, R. A. W. (1979) Central-Local Government Administrative Science Quarterly, 41, 270–77.
Relationships. London: Social Science Research Sharkansky, I. and Hofferbert, R. I. (1969) Dimensions
Council. of State Politics, Economics and Public Policy.
Rhodes, R. A. W. (1986) The National World of Local American Political Science Review, 63, 867–79.
Government. London: Allan & Unwin. Simon, H. A. (1945) Administrative Behavior.
Rhodes, R. A. W. (1997) Understanding Governance. New York: Free Press.
Policy Networks, Governance, Reflexivity and Sjöblom, G. (1968) Party Strategies in a Multiparty
Accountability. Buckingham: Open University Press. System. Lund: Studentlitteratur.
Rhodes, R. and Marsh, D. (1992) New directions in Sørensen, E. (1997) Democracy and empowerment.
the study of policy networks. European Journal of Public Administration, 75(3), 553–68.
Political Research, 21, 181–205. Stone, D. (1988) Policy Paradox and Political Reason.
Roe, E. (1994) Narrative Policy Analysis. Theory and Glenview, IL: Scott, Foresman and Co.
Practice. Durham, NC: Duke University Press. Streeck, W. and Schmitter, P. C., eds (1985) Private
Rosenau, J. N., ed. (1969) Linkage Politics. Essays on Interest Government. Beyond Market and State.
the Convergence of National and International London: Sage Publications.
Systems. New York: Free Press. Truman, H. S. (1951) The Governmental Process.
Sabatier, P. A. (1986) Top-Down and Bottom-Up Political Interest and Public Opinion. New York:
approaches to implementation research: a critical Alfred E. Knopf.
analysis and suggested synthesis. Journal of Public Van Horn, C. E. (1979) Policy Implementation in the
Policy, 6(1), 21–48. Federal System: National Goals and Local
Scharpf, F. W. (1991) Political institutions, decision Implementors. Lexington, MA: Lexington Books.
styles, and policy choices. In Political Choice. Weimer, D. L., Vining, A. R. and Vining, A. (1998)
Institutions, Rules and the Limits of Rationality, Policy Analysis: Concepts and Practice. Englewood
ed. R. Czada and A. Windhoff-Héritier, pp. 53–86. Cliffs, NJ: Prentice-Hall.
Frankfurt am Main: Campus. Wildawsky, A. (1979) Speaking Truth to Power. The
Scharpf, F. W. (1997) Games Real Actors Play. Actor- Art and Craft of Policy Analysis. London:
Centered Institutionalism in Policy Research. MacMillan.
Boulder, CO: Westview Press. Williamson, O. (1975) Markets and Hierarchies:
Schmitter, P. C. (1974) Still the century of corpo- Analysis and Antitrust Implications. New York:
ratism? The Review of Politics, 36, 85–131. Free Press.
Peter-3383-Chapter-05.qxd 6/8/2006 7:32 PM Page 114
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 115

Concepts and Theories of Horizontal
Policy Management


Coordination and coherence are familiar the vertical nature of governing has changed
themes in the discussion of shortcomings of substantially in the UK (6, 2004). The Finnish
public administration and public policy. government has developed an elaborate system
Governments have long sought to discover for managing cross-cutting policy priorities
means of making the policies adopted in one (Peters, 2006). Likewise, the Canadian govern-
department or agency correspond with, or at ment has come to recognize the need for
least not conflict with, those adopted in other enhanced coordination but there has not yet
departments. Likewise, governments have been the level of change desired, despite invest-
sought for mechanisms to ensure that there are ment of a good deal of energy by senior offi-
not major lacunae in their policy regimens, so cials through a variety of organizations and
that all potential clients are served and citizens procedures. We could go on adding examples
do not fall through holes in the various safety of government attempts to create more effec-
nets of government, or escape adequate eco- tive horizontal structures but the general point
nomic regulation. Unfortunately, as Pressman remains that governments are faced with sig-
and Wildavsky (1984) argued some years nificant challenges in creating greater policy
ago, much more has been said about creating coherence.
coordination than actually has been done Asian governments are no different from
about it, and coordination remains a principal those on other continents in confronting the
“philosophers’ stone” in the analysis of good challenge of coherence. Attempting to typify
public administration (Jennings and Crane, government in Asia is in some ways ridiculous,
1994). given the diversity of governing forms and
It is indeed difficult to do anything about experiences, so we will need to examine differ-
coordination, given the way in which govern- ences perhaps more than similarities. For some
ments tend to be structured and given the Asian governments the need to create coordi-
entrenched patterns of thinking about public nation may be more pressing for Asian govern-
policy and about governing. The Blair govern- ments, given that the lower levels of economic
ment in the United Kingdom, for example, has development produce fewer slack resources
made a great ruckus about creating “joined up to be wasted on redundant programs. On
government” but there is no real evidence that the other hand, some governments in Asia
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 116


may have greater capabilities for producing within government. Perhaps most importantly,
coordination, given the existence of strong the vertical character of policy and administra-
political parties (China) or strong central agen- tion is maintained by clientele politics, as
cies within the bureaucracy (Japan). Likewise, well as by the desire of government to serve
the absence of strong civil society institutions in particular components of the public. Clientele
most Asian countries (Callender and Johnston, groups that are involved in the public sector
1998) eliminates another source of effective if often believe that they have the right to their
informal coordination of public programs. own organization to serve them. Government
organizations that stray too far from serving
their clientele encounter the risk of losing nec-
MANAGING HORIZONTAL GOVERNMENT essary political support. Sometimes maintain-
ing the separate organizations to serve the
clientele appears important, even if the ser-
Governments need to find mechanisms to man-
vices themselves might be retained, or even
age more effectively in the horizontal as well as
enhanced, in a larger and more heterogeneous
the vertical direction. Reforms such as those
associated with the New Public Management
The organizations serving clients have many
have improved vertical management, but if any-
of the same interests in maintaining their
thing have weakened the horizontal manage-
exclusive position as do the clientele groups,
ment capacity of governments (see below).
and perhaps even more. Bureaucratic organi-
There are any number of examples of gover-
zations want to maintain their client base and
nance failures resulting from inadequate policy
their close relationships with their clients in
coordination and a reluctance to expend the
order to maintain their budgets and their
political capital necessary to manage govern-
influence within political circles. In this part of
ment in a more horizontal manner. These
the game of bureaucratic politics clients are a
failures include excessive costs because of dupli-
resource and a source of power. Therefore,
cation of programs, firms having to go to
coordinating programs, and possibly diluting
numerous different regulatory organizations in
that relationship between clients and the orga-
order to get needed permits to go into business,
nization, is usually not considered good poli-
“poverty traps” for less advantaged citizens
tics in these circles.2 Somewhat paradoxically,
resulting from an unwillingness to coordinate
some research has indicated that coordination
taxation and expenditure programs, and citi-
with less-closely allied activities is easier than
zens not receiving needed services because their
with more similar programs – the more
particular set of socio-economic characteristics
remote types of programs are not conceived of
were not covered by a patchwork of laws
as being the potential threat that the more sim-
attempting to provide for all groups within the
ilar programs are. In coalition governments
society. All these failings result in governments
these differences among ministries may be
costing more, and/or providing less high-
accentuated by ministers from different parties
quality services, than they might under a more
controlling ministries that need to cooperate.
coordinated system.
Although it is easy to characterize bureau-
cratic politics in the above totally self-serving
The vertical nature of government manner, we should also remember that most
employees of public programs believe in the
Politically and administratively public policy benefits of their program for clients. The orga-
appears to function better vertically, so that the nizations therefore argue (and often believe)
“stovepipes” that define policy within govern- that they protect their programs as much in
ments are perpetuated and reinforced. It is the perceived interest of the clients as that of
crucial to recognize that this vertical structur- the organization itself. This belief in the cen-
ing of the public sector is as much a function trality and efficacy of their programs by
of political as administrative characteristics employees of public organizations makes
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 117


achieving coordination more difficult, given but many of these same vertical patterns emerge
that members of an organization believe that without the peculiar features of American gov-
they are in actuality supplying what the citizen ernment.3 Indeed, the pressing need to manage
needs. Members of a public organization may government in a more horizontal fashion
realize that their clients do require a variety of appears in virtually every country for which
services, but may still believe that the most effi- I can find information. Thus, even in
cient way to provide for those needs is to place Scandinavian countries with corporate pluralist
additional funds and responsibility within the structures and networks surrounding their
one organization. ministries (Olsen, 1987), coordination often is
Further, in addition to the commitments to identified as a significant administrative issue.4
serve their clients, the vertical nature of These systems are cooperative and to some
administrative structures is reinforced by the extent network based in how they administer
different types of expertise held by different programs (Marton, 2000), but still have diffi-
ministries, and the linkage of those organiza- culty in working across ministries, or at times
tions with different professional networks. As working across divisions within the same min-
Martin Painter (1987, 12) put it: istry. On the latter point, one common strategy
… powerful political and intellectual reasons contribute for attempting to improve horizontal manage-
to perpetuating and reinforcing the self-containment of ment has been to create “superministries”
functional and organizational compartments. Each within which numerous closely related pro-
policy sector – health, transport, town planning indus- grams are all included. In many cases all this
trial development, employment and so forth – exists in
a jurisdictional shelter of organizational structures
effort has done is to move the coordination
where actors develop their own partial perspectives and problems within the one ministry, and to make
where specialized relationships of support and opposi- the issue less visible and therefore less likely to
tion develop connecting, bureaucratic and outside be addressed directly.5
interests in distinct arenas of sectoral policymaking.

Finally, legislative politics also helps to main-

tain vertical politics and to make coordination Alternative Views of Coordination
among programs, even closely related pro-
grams, more difficult. Some of this effect is a Although coordination is often discussed as a
result of the same type of politics described for single concept, it actually may mean several
clientele groups and bureaucratic organiza- different things. The most common distinc-
tions. Legislatures have similar contacts with tion made is between positive and negative
clients, as do the bureaucratic organizations, versions of coordination, with the former term
although the clients are usually referred to as implying avoiding direct conflicts among pro-
“constituents” in the legislative milieu. The grams, while the latter implies a more active
vertical nature of these relationships may be stance of assuring that the programs work
exacerbated through having strong legislative together effectively, and support each other. So,
committees that correspond directly with the negative coordination might be achieved
ministerial structure of government. Vertically through attempting to reduce the number of
defined ministerial structures also are consid- conflicting regulations on businesses, while the
ered important by legislatures for maintaining latter might be achieved by providing business
accountability for programs and funds. If the opportunity to do “one-stop shopping”
funds and responsibility are commingled and receive all their needed licenses at once at
through horizontal structures then those over- a single location.
sight functions become more difficult, and While the former conception of coordina-
accountability for both programs and finances tion is much easier to attain, it is a rather
may be diminished. minimalist version of horizontality, and may
The above discussion sounds much like tra- not produce the types of benefits expected.
ditional discussions of “iron triangles” in That is, eliminating overlaps may be desirable
American government, and to some extent it is, politically given the amount of complaint and
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 118


Table 6.1 Thinking about horizontal emphasis in coordination and horizontality in

government the public sector should be on policy or on
Activity administration (Regens, 1988, 138). The title of
Means/ends Coordination Integration this conference contains both the words “policy”
Joined-up (ends) Joined-up Joined-up
and “management”, and both of these concepts
coordination integration
Holistic (means) Holistic Holistic have some relationship to the capacity to make
coordination integration government work better horizontally. These two
issues are indeed certainly related, but they also
have important differences. Further addressing
one issue without the other can solve only a por-
obvious difficulties created, but may not address
tion of the coordination problems usually iden-
the more fundamental questions of achieving
tified in public service delivery. Administrative
integrated and coherent conceptions of public
coordination is in essence coordination from the
policy. Achieving that level of integration will
bottom-up, and is focussed on service delivery
require more positive approaches to coordina-
issues. This bottom-up orientation toward mak-
tion, involving implementing common concep-
ing government more effective assumes that the
tions and negotiating differences among
important questions about governing are imple-
policymaking organizations. Achieving that level
mentation questions. A policy orientation to
of coordination tends to involve imposition of
coordination assumes, on the other hand, that if
political power from above, or the creation of a
policies are formulated well initially then there
strong network of actors that develop shared
will be few (or at least fewer) problems in
conceptions of appropriate policy.
putting them into effect. The policy perspective
The ideas of “joined up government”
is more of a “top-down”, politically-centered
advanced by the Blair government in the
conception about how to make government per-
United Kingdom is an indication of the desir-
form better than is administrative coordination.
ability of achieving more positive levels of
The choice between administrative and
coordination. Perri,6 a scholar who has worked
policy coordination is to some degree a false
in Number 10, has pointed to two dimensions
dilemma; to be truly effective in generating
in the analysis of coordination activities in
coherence governments will require both forms
government (see Table 6.1). One dimension is
of coordination. The question then becomes
the simple question of whether government is
one of the balance between coordinating the
attempting to have its numerous departments
two elements of the policy cycle: formulation
merely take into account the activities of other
and implementation. Some scholars (Elmore,
organizations, or whether there is more of an
1979; Barrett and Fudge, 1981) have argued that
attempt to force real integration of the pur-
policy formulation should be guided by imple-
poses and actions of the programs. The second
mentation concerns and that policy should be
dimension is a bit more complex, implying the
“backwards-mapped”. In the context of coordi-
existence of a means/ends dichotomy in the
nation, this strategy involves thinking about the
management of programs. That is, it may be
potential duplication problems at the imple-
easier to agree on ends for programs and for
mentation stage implied by the policies that are
coordination than it is to agree upon the
being designed, and designing around the
mechanisms that will generate that coordina-
implementation problems. Other analysts have
tion. Politically, agreement on means may
argued (Linder and Peters, 1987; Hogwood and
involve one or more parties having to cede
Gunn, 1984) that, although implementation is
some aspects of their program (and budget).
important, it should not be so dominant in
initial policy formulation. Governments should
Policy and administration
first decide what they want to do and then
Following from the above classification of decide how those goals can be achieved effi-
actions meant to produce enhanced coherence ciently and effectively (Bogason, 1991). For the
in government, we can ask whether the coordination question these decisions will
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 119


require deciding what priorities the governing of government. It appears that the highest
system as a whole has, in contrast to the multi- reaches of government can not avoid responsi-
ple priorities that exist with each individual bility for creating and delivering integrated
program and organization. policies, especially if there is to be an attempt
Asian governments may have less capacity to to produce some redefinition of a policy area
depend upon coordination at the bottom than to include the cross-cutting dimensions. That
do governments in European or North having been said, the importance of emerging,
American systems. To depend upon lower level cross-cutting issues, and the absence of as yet
bureaucrats to coordinate, or to engage in what clear definitions of those issues, will require
is to some extent self-effacing behavior, when high levels of interaction between top and
they are not as well integrated as career struc- bottom of organizations.
tures, or may depend upon external sources of Again, the dichotomy between top-down
income to survive economically, is to ask per- and bottom-up being discussed here is to some
haps too much. Descriptions of several of these extent false. In many concrete examples of
bureaucracies (Indonesia, Philippines) indi- coordination, the need for greater cooperation
cates that there is less commitment to public among policy-making organizations at the top
service goals than might be expected for civil first becomes apparent as difficult cases pre-
servants in other political systems. Further, in sent themselves in the field. Likewise, the fail-
some systems (Thailand, India) there is a great ures of coordination at the bottom of the
deal of internal fragmentation in the adminis- structures become apparent as the results of
trative system that may make coordination policy-making and administration are assessed
more difficult. by evaluators at the centers of the political sys-
tems. As well as becoming apparent of the
problems through interactions of top and
Top-down versus bottom-up bottom, it is likely that the problems can only
be addressed through interactions up and
Related to the above question about politics down the hierarchies within government.
and administration is a second question of
whether coordination can best be achieved at
Horizontal or Vertical?
the beginning of the policy process or whether
it should be more focussed at the “bottom” of The coordination question is generally, con-
government. In other words, do the people ceptualized in terms of making government
who actually deliver services know more about function in a more “horizontal” manner, and,
those services, and their clients, than individu- indeed, many of the most important contem-
als at the top of organizational hierarchies? If porary issues concerning coordination are
they do, then it would make sense to have ser- those of working across programs within a sin-
vices coordinated around their targets, gle level of government. These problems of
whether they are individuals, organizations, or managing horizontally are compounded when
areas, rather than impose the coordination the issue of coordination among levels of gov-
from the center of government. If the bottom- ernment is added, especially in federal regimes
up approach to coordination is emphasized such as Canada, Germany, or the United States
then bargaining over coordination would be (Derlien, 1991). Even in unitary regimes, how-
done not by ministers and senior civil servants ever, many of the same inter-governmental
in central agencies but rather by lower level coordination problems among political and
administrators. administrative levels of government arise,
This latter strategy corresponds well to the albeit usually without the political intensity
“empowerment” ideologies now being imple- that can characterize federal-provincial dis-
mented within many governments (Peters, putes in a federal regime (Toonen, 1985).
1996), but it also requires the initial creation of These problems may be confounded to an even
clear policy frameworks at the highest reaches greater extent when, as in most countries,
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 120


central government ministries have field United States (Katz, 1995), then there is no
structures that do not correspond to the struc- particular cause for concern. If the central gov-
ture of sub-national government. ernment does care, as most do, then vertical
The fundamental root of the coordination coordination becomes a sine qua non for suc-
problem in federal systems is that most federal cessful horizontal coordination (Derlien,
regimes have evolved in ways that permit all 1991). The reverse is probably also true, and it
levels of government to be involved in almost would be difficult for sub-national implemen-
all policy areas. Thus, to be capable of address- tors to compensate for fundamental design
ing problems of redundancy and incompatibil- errors occurring at the central government
ity requires that all governments agree on some level (Linder and Peters, 1987). Perhaps the
basic approaches to the policy and work to most important point here is that almost all
make their means of service provision more governing now is multi-level governance,
compatible. This agreement is often easier said requiring blending not only the perspectives of
than done, but it has been done. For example, different levels of the public sector, but also
several respondents in a series of interviews on blending different functional policy communi-
coordination in Canada pointed out that the ties existing at all levels of the system.
federal government and the provinces have
come to agreements to coordinate and inte-
grate their efforts at food inspection as a pre-
cursor to further attempts to coordinate PRESSURES FOR HORIZONTALITY
policies among the levels of government. These
agreements may appear to be simple matters –
Coordination and coherence have been
negative coordination only – but should be
problems for as long as there have been gov-
seen as an accomplishment in light of the diffi-
ernments, but the need for improved coordi-
culties involved in achieving those agreements.
nation appears to be more manifest in the early
One respondent pointed out that these agree-
21st century than in the past. This emphasis on
ments were gained only after days of discussion
coordination appears to have arisen for a num-
and then had to be validated at the level of the
ber of reasons. These pressures for horizontal
provincial prime ministers. Further, these verti-
government reflect political demands for a
cal coordination concerns were compounded
range of actors, including changes in political
by the need to coordinate food safety issues
ideologies about the role of public sector. In
across various ministries, such as agriculture,
addition, administrative reforms over the past
trade and industry, and increasingly foreign
several decades have resulted in the need for
affairs. Thus, one of the more common func-
increased coordination interventions in order
tions of government – ensuring that the food
to address new problems that have emerged as
sold in the market – is safe for the consumer
a result of attempts to solve problems of effi-
actually involves a significant level of inter-
ciency and effectiveness in government.
organizational and intergovernmental bargain-
ing to become compatible.6
Further, in most federal systems the central Fiscal pressures
government utilizes sub-national governments
to implement many or most of their policies. Perhaps the dominant pressure for enhanced
This implementation strategy means that, even organizational coordination in the public sector
if policies are effectively coordinated in the comes from the need to save public money. For
national capital, that integration may fall apart both political and strictly financial reasons gov-
once those policies begin to be implemented. If ernments have less money to spend than in the
the central government is not particularly con- past and must attempt to control public spend-
cerned about how their policies are imple- ing as effectively as possible. One way of con-
mented, as appears true under the block grant trolling expenditures is through eliminating
provisions becoming so common in the redundancy and ensuring that services will be
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 121


provided in the most cost-efficient manner. The accomplish their goals. These programs are
examples of contradictions and duplications in never easy tasks to implement, and the mecha-
public programs are familiar and make good nisms as yet in place are often inadequate to
copy for the exposes of inefficiency and outright create genuine strategic management. Still,
incompetence in the public sector. Further, for these represent movements in the appropriate
governments intent on saving money, eliminat- direction for enhanced policy coordination. As
ing redundant programs is a way to do so with- governments continue to develop programs of
out necessarily reducing the level of services performance management and stress the
being delivered to the public. results of their programs, then questions of
Although familiar to people inside and out- the contributions of numerous programs to
side of government, the existence of these the achievement of the desired outcomes will
coordination problems and their fiscal impacts become more apparent.
are often exaggerated for political purposes.
These coordination failures certainly do exist,
but generally represent very minor amounts of
money, compared with the political benefits
The issues that governments must now con-
that the programs involved may generate for
front also call for greater attention to horizon-
their governments. Further, coordination itself
tal management. Governments have invested a
is far from a costless activity. To impose coor-
great deal of effort at finding ways to make
dination on an existing ministerial structure
individual programs work more effectively, but
requires utilizing resources to monitor over-
the policy issues that are emerging tend to
laps and perhaps even more political capital to
transcend the usual boundaries of those min-
impose sufficient political “clout” to eliminate
istries and programs. For example, a principal
those overlaps. Creating coordination may also
economic issue confronting contemporary
require creating new programs to close gaps in
governments has been defined as “competitive-
program structures, and with that will generate
ness”. This is certainly an economic issue but
increased expenditures by government.
also involves education, labor, social policy,
regulation and a host of other policy consider-
ations. Likewise, issues are being defined in
Strategic management
terms of client groups – women, the elderly,
immigrants – whose needs cut across conven-
One positive consequence for coordination
tional boundaries of ministries and require
arising out of the management reforms during
greater integration of existing programs. There
the past several decades is the emphasis on
is every reason to expect issues to continue to
strategic management and the selection of
defy the boundaries of conventional minister-
clear objectives for the public sector. One of
ial structures, so that the needs for coordina-
the first things that any politician or adminis-
tion are not likely to diminish. Likewise, it does
trator engaged in such an undertaking dis-
not seem that simply moving “boxes” in order
covers is that most of the important strategic
to capture a particular set of problems will
objectives for governments cut across con-
solve the problems – the next issue may only
ventional organizational boundaries and
cut across the new administrative boxes that
therefore require working horizontally across
have been created to solve the previous
the “stovepipes” of government. So, exercises
such as implementing Strategic Results Areas
in New Zealand, Strategic Portfolios in Finland
and (to a more limited degree) the Govern- Earlier administrative reforms
ment Performance and Results Act in the
United States7 require those governments to Although some earlier reforms have empha-
think collectively and horizontally about what sized strategic management and integration of
they want to do, and how they are going to policies, a more common pattern of reform has
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 122


been to decentralize, devolve and disaggregate and therefore greater need for horizontal
the public sector (deMontricher, 1999; Peters government structures.
and Pierre, 2000). These reforms may have The impact of international forces on coor-
generated some efficiency benefits for the dination requirements is nowhere clearer than
public sector (although those benefits are often in the European Union. These fifteen countries
difficult to demonstrate) but they have made are finding that they must develop coordina-
governments that were already fragmented tion mechanisms in their national capitals, as
vertically even more diverse and fragmented well as in Brussels, if they are to be successful
institutions. As well as creating autonomous or participants in the game of European policy-
semi-autonomous agencies within the public making (Kassim, Peters and Wright, 2000;
sector, these reforms have created a number of Kassim, Menon, Peters and Wright, 2001).
organizations – referred to as quangos and Although some countries have chosen not to
quagos – that have some characteristics of both invest heavily in coordination activities and to
public and private sector organizations. permit ministries to bargain relatively auto-
Therefore, a subsequent round of reform has nomously in a variety of settings, the typical
been instituted that stresses the need for response to EU membership has been to create
restoring some greater integration among significant structures for coordination. Some,
the component structures of the public such as the United Kingdom, have regarded
sector, and creating greater coherence. These policy coherence vis-à-vis Europe as essential
reforms include strengthening central agen- to their capacity to extract as much as possible
cies, strengthening offices of presidents and from the policy system, and also to be able to
prime ministers (Peters, Rhodes and Wright, defend the national interest effectively in what
2000) and creating “super-ministries” that pull is often assumed by the British to be a relatively
together a range of relevant ministries. In some hostile environment.
instances the simplest response has been to
reinstitute the traditional ministerial struc-
tures and reduce emphasis on decentralization. Cross-cutting concerns

Coordination is also being driven by a range of

Globalization/Europeanization cross-cutting concerns of modern govern-
ments and contemporary public opinion. For
Globalization is something of a cliché, but it is example, the environment has become a com-
also a reality for governments. Even govern- mon concern for a range of policies, and envi-
ments that have been relatively insulated from ronmental agencies have become in essence
international pressures find that almost all another set of central agencies. Just as all poli-
their policies have an international dimension, cies must be approved by ministries of finance
and that those international pressures tend to for their fiscal implications so too must these
force broader consideration of the issues. policies also be vetted by environmental agen-
Education may have been primarily a national cies (Doern, 1993). Rights and equality for
concern, even in the recent past, but it is now minorities, women and other designated
centrally connected with issues of interna- groups are also becoming concerns for coordi-
tional competitiveness, thereby requiring nation and common consideration across a
closer associations with labor, industry and range of specific policy choices. Finally, as
foreign affairs ministries. Agriculture is now noted above, the international implications of
centrally connected with international affairs, most if not all issues are also now a matter of
certainly in Europe, but also in most other common concern.8 In short, policies are no
countries. The list of connections among min- longer seen as just operating within their own
istries could be extended but the basic point defined policy domain; the full range of their
remains the same – international involvement implications is now understood and becomes
creates more interconnections among policies part of a broader consideration of the issues.
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 123


MANAGING HORIZONTAL GOVERNMENT acceptable price level for both. Thus, while
most of the means of producing coordination
that we discuss conventionally require the
The above discussions of the nature of policy
imposition of coordination on government
coordination and the contemporary motiva-
structures by law and authority, the market
tions for enhancing coordination then lead
approach to coordination would substitute
rather naturally to question how this Holy Grail
more indirect and virtually autonomous
of administration can be created in the real
modes of coordination. The actors will be
world of governing. The conventional answer
involved with each other for purposes of
would be to impose greater control from the
mutually advantageous exchange, and their
upper reaches of government. This is certainly
involvement will be episodic and partial, rather
one solution, and often an effective one, but it
than the more comprehensive style of interac-
is by no means the only available solution to the
tion typical of actors entirely within the public
problem. The general style of governing has
sector itself.
changed in the majority of industrialized
In the case of coordination through markets
democracies, with much greater reliance on
within the public sector, the necessary bargain-
indirect instruments and diminished reliance
ing may come about, with money being the
on direct government provision and command
medium of exchange. One clear example is the
and control regulations. As those changes have
creation of internal markets (Jerome-Forget,
occurred so have changes in the ways in which
White and Wiener, 1995; OECD, 1993) as a
coordination is approached. Therefore, we will
means of coordinating actions and imposing
look at markets, networks and hierarchies as
market discipline on organizations that other-
alternative means of achieving coordination.
wise would be governed strictly through hierar-
This classification of social processes has by
chy. The assumption of internal markets is that
now become quite conventional in the social
there are some components of service delivery
sciences (see Thompson, Frances, Levacic and
that can be conceptualized as “selling” their ser-
Mitchell, 1991) but is still a useful scheme
vices, while other actors within government are
through which to approach issues of categoriz-
the “buyers” of those services. These artificial
ing and assessing coordination mechanisms.
markets function as mechanisms of getting
Further, this threefold classification emphasizes
these actors together in the most efficient man-
the extent to which contemporary governance
ner, without having to use formal authority.
ideas have moved away from thinking about
Also, we should remember that, to some extent,
hierarchy as the only, or even the dominant,
the budgetary process has always been some-
approach to coordination (not to mention
thing of a coordination process based on
other aspects of governing (Pierre and Peters,
money, but the increasing reliance on internal
markets has made the role of money and
exchange in coordination more explicit than in
Markets conventional patterns of governance.
Likewise, contracts among public organiza-
The economic theory of markets assumes that tions are increasingly being employed as a
coordination will occur almost automatically if means of coordinating their activities, replac-
competitive forces are permitted to function ing previous coordination through hierarchy
without interference. The “hidden hand” that with mutually acceptable “deals” (Fortin,
is assumed to function in other aspects of 2000). Contracts in the public sector combine
market relationships is also expected to oper- some features of markets with an otherwise
ate in the coordination of programs. Indeed, in legally based instrument (Peters, 2002). As
many ways, markets are fundamentally institu- with internal markets the bargaining in inter-
tions for coordination, with the institution nal contracting is also “quasi”, given that it is all
presumed to assure that sellers and buyers will public money and there may be a legal require-
find each other, and find each other at an ment for the delivery of the service.9 Further,
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 124


contracting in the public sector may be perhaps the tops of hierarchies responsible for
restricted in the number of possible bidders implementing the law. The very decentraliza-
and in the range of alternative patterns of tion that makes markets so valuable in some
service delivery possible under the law. Despite situations also may limit their effectiveness.
those constraints, contracting is a market- Those limits on markets may be especially
based idea for coordination that may be damaging for policies that contain a strong
applicable to a wide range of service areas. legal or entitlement basis for citizens, and
Finally, the market can be utilized to coordi- hence will not be effective for the vast range of
nate economic policy, especially within the social policies in which government is placed
European Union. The increased linkage of in the position of providing services to clients.
these countries and their fiscal and monetary On the other hand, market-based instruments
policies means that the operations of the such as taxation may be able to replace com-
market places pressures on all the countries to mand and control regulation in a variety of
make their policies compatible (Jacquet, 1998). areas.11
Further, the Maastricht Treaty and the
Amsterdam Treaty place strong demands on
the member governments to coordinate their Networks
fiscal policies. Arguably, some of the same
pressure for coordination is arising, perhaps Networks are another bargained mechanism
less directly than in the EU, in all the industri- for producing policy coordination. Rather
alized democracies as they must compete with than the exchange relationships implied by the
one another in an international marketplace. market, networks themselves are defined much
This coordination may extend not just to mon- as individual organizations would be, though
etary and fiscal policy but also to tax policy patterns of interaction. For organizations
(see Hallerberg and Basinger, 1998) social those interactions occur primarily among
policy (Adema, 1997) and a range of other individuals, while for networks they are among
public policies contributing to economic other organizations as well as individuals.
competitiveness.10 Further, the close financial Networks have many virtues as mechanisms
connections now are necessitating closer coor- for coordination, and to some extent depend
dination of financial and even criminal laws upon natural patterns of interaction that
(Laronche, 2000). emerge among organizations and individuals
In summary, the market does provide some concerned with the same policy issues. These
instruments to promote the coordination of may be “epistemic communities” (Zito, 1999)
policies, but it may not be an all-purpose solu- defined by common intellectual position and
tion. For the market to be effective there must common patterns of training. Networks
be something for the participants to exchange, depend upon the interests and commitment of
and that is not always the case in public pro- individuals and groups to be successful; most
grams. Thus, this approach may work when the of these participants (inside and outside the
programs involve goods and services that are public sector) want to do their jobs as well as
in principle marketable, but almost certainly possible and find networks convenient for
not for all. Not all relationships among multi- enhancing their effectiveness (Chisholm,
ple organizations can be coordinated effec- 1989).
tively through markets and exchange. In some The term “network” is to some extent a
instances there are mutual complementary short-hand term for a variety of patterns of
goals, rather than the somewhat contradictory interactions between state and society, and
goals implied in market exchanges. Further, within the public sector itself (Olsen, 1987;
the goals that might be achieved through Kickert, 1995). In general, these network rela-
mutual adjustment among the interested par- tionships involve the State relinquishing some
ties might be different from those sought of its authoritative powers in order to achieve
by the legislators who wrote the law, or even greater agreement among the interested and
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 125


affected parties within a policy area (Marin, rather than organizations in the network
1990). This is a natural mechanism for coordi- solving problems among themselves. This is
nation, given that bargaining tends to create one emerging role for central agencies that
agreement on proposals and tends also to may eschew their traditional role of imposing
involve a wide range of actors. Network bar- coordination in favor of bargaining over issues
gaining can be effective at both the formula- with a range of organizations. Further, certain
tion and implementation stages of the policy public sector organizations have an emerging
process (see Considine, 1992). In both aspects role in creating coordination, particularly
of the process of governing networks can be when utilizing popular ideas and issues, partic-
used to avoid turf-battles over policies and ularly environmentalism, as a means of pro-
clients, and to create a common perception of moting common values (Doern, 1993). Ideas
the policy issues and a common reaction to the like environmentalism, even in the absence of
issues. Networks need not exist only in the an institutional basis, can be used to produce
national capital around the design of policy, coordination across programs.
but often are crucial at the lowest level. That is, Despite their virtues, networks also have
networks organized around clients (classes or some weaknesses intellectually (Dowding,
even individuals) can be the most effective 1995) and as a means for producing public
form of coordination in the social services. For sector coordination. One of these is an analytic
many Asian countries the absence of networks problem; once you have said that a network
of organizations in the civil society that exists in a policy area, what do you say next? It
can provide the external network may be a seri- is difficult to argue that networks do not exist,
ous impediment to using non-governmental but it also appears difficult at times to say much
methods for coordination. Perhaps especially more about them and to use them in any pre-
important for coordination and coherence is dictive manner. This weakness is especially true
the relative absence of social groups that cut given that networks can have rather different
across conventional policy sectors. internal dynamics. For example, Paul Sabatier
Professionalism creates a ready-made net- (1988) conceives of multiple networks existing
work for coordinating some types of public around many policy areas, with the principal
policies. One virtue of the professions in dynamic being conflict over the definition of
their classical definition (see Wolgast, 1992) is the policy problem and over finding the appro-
their function as a reference group for their priate solution to the issue. This clash of ideas
members and, in most instances, professionals is, however, a form of coordination, since it
will have their own network of fellow profes- tends to eliminate conflicting ideas about
sionals that can supplement the networks cre- policy and with that conflicting and probably
ated through the organization itself. Although wasteful duplication. Still, the behavior of net-
in many ways beneficial, professional networks works, especially those structured without a
also can limit coordination. In the first place, central position to government, may be too
these networks are relatively closed to out- indeterminate to permit government to be par-
siders, so that there is less capacity for objective ticularly effective in coordinating programs.
scrutiny of policies than is true for other forms
of decision-making. In addition, each profes-
Interest groups
sion tends to define problems and solutions in
their own terms so that there may be very Political groups advocating the interests of
effective coordination within each profession those segments of society presenting govern-
and therefore (usually) within each single pro- ment with cross-cutting policy issues can func-
gram, but coordination across programs may tion as a means of identifying needs and
actually be more difficult.12 pressing for their solution. In many countries
Other networks could be structured more the target populations for major cross-cutting
vertically, with most interactions being upward policy issues – the elderly, women – are well-
to the relevant government organization, organized and are positive political symbols.
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 126


Other populations, however, are less well- encounter greater difficulties in coordinating
organized and in the case of immigrants are policies than will other countries. Civil service
often conceptualized as pariah groups with few systems such as those of the United Kingdom
political rights and resources. Depending upon and other Westminster democracies, in which
the political power of the groups requiring there are relatively frequent movements
service may not be the most effective means of among departments as a civil servant works his
generating coordination and policy coherence. or her way up through the hierarchy during
For example, one of my British respondents their career, should produce a somewhat better
pointed to the recent attempts to coordinate possibility of adequate policy coordination.
and integrate government responses to racial Civil servants who have worked in a variety of
attacks in British cities. In this case the leadership different programs should have a better idea of
had to be from within the bureaucracy itself – the perspectives of other departments, and
the Home Office and the Crown Prosecution tend to have a sense of belonging to a govern-
Service – as opposed to coming in response to ment rather than to a particular organization
the political power of the groups in question. within that government. They should also have
The various immigrant groups and ethnic a better idea of the total range of services deliv-
organizations simply did not have sufficient ered by government and how they could be
legitimacy and political “clout” to provoke the made available to clients.
desired response from the political system Canadian respondents in a set of interna-
themselves. tional interviews about coordination expressed
Interest groups may have many of the same growing concern about the loss of this impor-
problems as political parties in coping with tant informal mechanism for generating coordi-
cross-cutting issues. Many political parties nation within the public sector. They noted that
work with particular definitions of the issue the down-sizing of the public service is reducing
areas that have been functional for them in the the opportunities for movement within govern-
past but which have outlived their utility, or ment so that individuals tend to remain in one
which are not widely shared by other actors post for much longer. Their vision of what gov-
involved in the issue areas. To be successful ernment does and is about has narrowed
they may have to broker deals with other accordingly. Further, the increasing technical
groups with complementary if not contradic- content of most programs means that greater
tory definitions of issues (Sabatier, 1988; Page, expertise is required, and with that individual
2003). This need to compromise and negotiate civil servants may have fewer fungible skills that
often contrasts with their need to serve their can be applied in other settings. While the public
members directly. That service may be ori- service may be becoming more expert it also
ented toward the separation of separate may be lessening the chances of effective coordi-
programs for their constituents rather than nation from within the public service itself.
accepting the interdependencies among poli- The absence of a career, professional civil
cies and issues. service may be an impediment to effective
coordination within government. The prob-
lems are as significant for European and North
The civil service network
American countries that have opened their
Another important network mechanism for senior civil service to non-career appoint-
producing greater coordination within gov- ments as it is for Asian, African, and Latin
ernment is structuring the careers of civil ser- American countries that have not successfully
vants so that they have broad experience and a institutionalized civil service systems. The
broad conception of government and policy. absence of a civil service may be compensated
Countries such as the United States, Finland or for by the connections of the “in and outers”
Norway, in which civil servants spend most with policy networks that can provide them
or all of their careers within a single agency or with both substantive policy information and a
department will, everything else being equal, range of connections within government.
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 127


Hierarchy executives are now attempting to provide

themselves with the means for promoting
We will now turn to the most common mech- coordination, whether it is done primarily by
anism used to achieve coordination in the the executive or through the use of cabinet and
public sector – hierarchy. The most common powerful administrative agencies (see Peters,
pattern is for coordination problems to be Rhodes and Wright, 2000).
addressed from the top by authority and legal Chief executive staff. By themselves prime
provisions. Even after several decades of inten- ministers and presidents do not have much
sive administrative reforms, driven to a great capacity to produce effective coordination.
extent by ideas of the market (Peters, 1996), They tend to be extremely overworked and
when issues of coordination arise the most have little time to spend on coordinating the
common response is still to use authority and activities of the numerous ministries under
hierarchy. This is to some extent a recognition their overall control. They can, however,
of established patterns, or path dependency, of develop staffs and organizations that can assist
using this type of coordination device. This them in coordination. The most developed
also reflects the relative effectiveness of these organizations of this type are in the Executive
devices when dealing with the full range of Office of the President in the United States.
policy issues, while markets tend to be This office contains not only the personal
restricted in their applicability, and networks staff of the President but a number of moni-
depend perhaps excessively upon the good will toring and coordinating organizations such
of the participants. as the Office of Management and Budget, the
Another virtue of the hierarchical approach Council of Economic Advisors and the
to coordination is that there is a wide range of National Security Council as well. All recent
responses available within the broad category presidents also have had some organization in
of hierarchical methods. Governments have the White House for coordinating domestic
been rather creative in the ways in which they policy, although the name and responsibilities
respond to the necessity of coordinating, and of those organizations have varied. Although
have a repertoire of ways to confront overlap similar offices exist in other governments, e.g.
and duplication. As noted above, some of these the Bundeskanzlersamt in Germany and
depend upon political power and others Austria (Mueller-Rommel, 2000), the Kansli in
depend more upon administrative procedures Sweden (Larsson, 1988), and the Department
and inter-organizational relationships, but all of Prime Minister and Cabinet in Australia
have at times enjoyed some success in bringing (Davis, 1998), they tend not be as fully articu-
together program that were by structure and lated as in the United States.
culture different, competitive and perhaps One virtue of these executive organizations
even hostile. for managing cross-cutting policy issues is that
they tend to be flexible and do not have to be
concerned with delivering services to existing
The core executive
clients (other than advice to the chief execu-
The locus for horizontal policy coordination tive). They can thus create internal task forces
and issues management is usually assumed to or temporary structures to cope with changing
be at the very center of government – the chief issues and interpretations of issues. Further,
executive and the central agencies that serve they do not have as much policy “turf ” to
that executive. The ultimate responsibility for defend as do line agencies. On the other hand,
policies, and the coordination of those poli- relying on this level of government for coordi-
cies, lies with prime ministers in parliamentary nation is likely to be highly centralizing.
regimes. The situation of the American Further, it can overload the office of the chief
President is somewhat more complex, but even executive at a time when the prevailing ethos of
there the President bears ultimate responsibil- governance is decentralization. Those prob-
ity for the execution of policy. These chief lems can be compounded if definitions of
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 128


issues and policies are more clearly identified of control and accountability. Again, however,
at the lower echelons of government. to be fully effective, the central agencies must
Central agencies. A more general strategy for be strongly supported by the prime minister.
achieving coordination from the center of gov- Cabinet itself, especially with strong Prime
ernment is to rely upon central agencies. By Minister or Minister of Finance. Cabinet itself is
this term we refer to budgetary, policy, and another locus for the management of cross-
personnel management organizations that cutting policy issues. In some ways it is the
report directly to the chief executive, or which most logical institution to perform this task; all
are assigned principal responsibility for policy the principal actors in policy-making and
coordination and central management of service provision are represented. On the other
issues (Campbell and Szablowski, 1979; Savoie, hand, cabinet may be a place in which the min-
1995b). Examples of central agencies are the isters must protect the interests of their depart-
Treasury in Britain, the Treasury Board ments. Those interests may well not be best
Secretariat in Canada, and Departments of the served (in the short term at least) by excessive
Public Service and Ministries of Finance in a cooperation with other agencies, or by exam-
number of countries. These organizations can ining the broader implications of groups of
be employed to enforce the priorities of the policies. The cabinet can serve as the locus for
chief executive, but they also tend to develop the examination of cross-cutting issues if there
priorities and managerial styles of their own, is adequate leadership, both from the Prime
and to develop substantial power over policy. Minister and from the civil service that serves
Central agencies can play a significant role the cabinet. With that leadership there can be a
in creating coordination, but they also can capacity to redirect the discussion of issues and
generate substantial conflict with the line orga- enhance policy coordination. Even with a
nizations actually providing public services. strong prime minister, a close link with the
These frictions reflect the conflicts between minister of finance appears to be crucial in
“line” and “staff ” organizations that are typical creating coherent government, so that, as
of inter-organizational politics within the Donald Savoie (1999) said, there is a need to
public sector. The former type of organizations ensure that “no light shines between them”.
resent the power exercised by control organiza- Cabinet committees. A cabinet may be too
tions that do not directly serve the public and large an organization to coordinate programs
which, it is argued, know little about the pro- effectively. This is especially true given that
grams being delivered. Staff organizations each minister usually will feel compelled
(including central agencies) tend to believe to defend the interests of his or her own
that line agencies have extremely narrow views department, and this need may make the nec-
on policy and do not understand the need to essary cooperation difficult to obtain. Many
impose overall priorities on government. cabinets have been reduced drastically in size
The role of central agencies has been (Bouckaert, Ormond and Peters, 2000), but
increasing in most contemporary political sys- may still be too large to function effectively as
tems, despite the general Zeitgeist of decentral- a single decision-making entity. In the case of
ization and deconcentration. The political problems that are not well defined and which
dynamic has been that as programs are decen- cut across a range of ministries ministers may
tralized the conventional mechanisms of polit- feel compelled to defend the claims of their
ical control are devalued. The one major department over control of the issue, with
control instrument that remains in place is the some loss of necessary cooperation across
budget, so that ministries of finance in partic- departments.
ular become crucial in the process of control- Most cabinet systems therefore have devel-
ling administration. Further, as performance oped working “inner cabinet” systems, or some
management becomes a central component of committees within cabinet that can establish
managing the public sector, then ministries of collective priorities and coordinate policies
finance become even more central to processes across portfolios (Mackie and Hogwood,
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 129


1985). One approach to achieving this goal is to for coordinating a range of other programs,
create an overarching “priorities and planning this is unlikely to receive the same priority as
committee” within cabinet, as in Canada. This running the programs within his or her own
approach can coordinate policies across the department. In the case of a minister without
entire range of public programs, but often will portfolio assigned primary responsibility for
push too many decisions upwards to a few coordinating programs, such an individual
senior officials of government. The alternative may have more time to spend on this activity
approach is to develop a series of cabinet com- but may not have other necessary resources. In
mittees, each responsible for a particular seg- particular, cabinet ministers without a depart-
ment of policy. This approach has the mental power base may not have sufficient
advantage of bringing the relevant departmen- clout within the cabinet to bring his or her col-
tal ministers to the table to coordinate their leagues along if there is a need to coordinate
own activities. This has been particularly evi- their policies. This, however, may be counter-
dent in the budgetary process, with envelope acted by assigning politicians with strong
budgeting in Canada and Sweden as examples. political links to the prime minister these roles.
This approach also has its disadvantages. In Junior ministers. Rather than have a minister
particular, the boundaries between policy areas, accept additional responsibilities and add to an
and therefore between cabinet committees, are already extensive range of duties, governments
not always clear. The boundaries between can instead develop a system of junior minis-
policy areas may be becoming even less clear; ters that can help coordinate their ministries,
for example social policies, labor market policies, and perhaps accept responsibility for services
and even education policies, have become inter- to designated groups or for other special func-
twined through competitiveness concerns to a tions. To some degree junior ministers will
degree not previously experienced. Therefore, have some of the same problems encountered
there may be a proliferation of coordinating by ministers without portfolio. Being desig-
committees, and the consequent need to coor- nated “junior” these officials almost certainly
dinate the coordinators. have less power in government than will min-
Ministers without portfolio, or with additional isters, or probably senior civil servants. If these
coordinative portfolio. Another means of gener- aspiring political leaders are asked to coordi-
ating improved coordination within a cabinet nate a range of services and manage cross-
system is to utilize ministers without portfolio cutting issues controlled by powerful ministers
tasked to coordinate programs within a broad they may have only limited success. Further,
policy area. Another related method would be they may be placed in confrontational posi-
to assign departmental ministers additional tions with the senior ministers and may per-
coordinating portfolios. For example, in the ceive the job as a political detriment rather
Netherlands one minister has been assigned the than a step up the political career ladder.
additional responsibility for coordinating all Ministerial organizations themselves. We
programs being delivered to immigrants, as well have been discussing the need to coordinate
as programs designed to regulate their entry and across cabinet portfolios or their equivalents,
their participation in the labor market. In other but cabinet departments can themselves
cases ministers have been assigned the responsi- develop mechanisms for policy coordination.
bility for integrating the services provided to One that has been tried in a number of coun-
women, or in one case to provide a range of ser- tries is the creation of “superministries” that
vices for the middle class. would incorporate within their own structures
While this system has the advantage of des- a wide range of programs that otherwise
ignating someone to be responsible for coordi- would have to be made compatible across
nation of a policy area, it also has several departmental structures. At one extreme the
important drawbacks. The most obvious is Swiss government is limited to seven govern-
that it can overload an already busy minister. ment departments, so that if their portfolios
Further, although the minister is responsible are relatively homogenous they should be able
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 130


to produce substantial internal congruence of interests of relevant programs. This can be

policy. At less of an extreme, in Australia, the done through the creation of broad advisory
Hawke government in 1983 reorganized gov- committees for departments or bureaus con-
ernment to create a smaller number of large taining representatives of other organizations.
ministries and created an inner cabinet that For example, in the Scandinavian countries
had some capacity to coordinate policies (Norway in particular) each ministry will have
across the entire range of government services. an advisory committee composed of represen-
The British government had tried a similar tatives of interest groups, as well as from other
strategy much earlier, and the Nixon adminis- ministerial departments. Any significant policy
tration in the United States had proposed initiative by the ministry must be referred to
creating four “super-departments” in the fed- this advisory committee. This system works
eral government (Nathan, 1975). well in these countries, with their traditions
While it may appear logical to locate as of consensual decision-making and well-
many similar programs as possible within a developed interest group universes. Even with-
single ministry, the coordination gains from out that tradition this method can at least
that structural decision may be more apparent inform interested departments of actions and
than real. First, there will have to be a signifi- perhaps allow them to be settled (in cabinet or
cant sub-ministerial structure which may by other means) earlier than they might other-
engender its own difficulties in coordination. wise be. Further, like most other existing mech-
Likewise, if a minister has a too large a min- anisms for coordination, the agenda for these
istry with too many internal divisions, he or committees is set by existing organizations
she may encounter the same problems in pro- using conventional conceptualizations of policy.
ducing coordination as might a prime minister One variant of the advisory committee
with an equal number of ministries to coordi- mechanism is the use of management boards.
nate. Finally, the location of all the apparently With the increasing use of disaggregated gov-
related programs within the single department ernment organizations (e.g. agencies in the
may lead to complacency and the assumption United Kingdom) in a number of countries
that the problems have been solved while the there may be a need to use the same gover-
problems actually persist. nance system as was developed for organiza-
Placing a number of programs together tions of this type in Scandinavia. In the
within a single ministry also may have other Scandinavian countries from which this model
effects on policy and management. By placing of organization was derived, the use of boards
the principal coordinative responsibility composed of government and lay personnel is
within a department, the decisions tend to be a means for providing a broad perspective on
taken more by career officials than by political the functions and role of the organization, and
officials. If departments remained more frag- hence a broader perception of the policies
mented, politicians would have to debate those being developed. To the extent that other gov-
issues in at the cabinet level in order to pro- ernment organizations are represented on
duce better policy coordination. Developing these boards they can help produce enhanced
larger departments, in turn, may free up the coordination. For example, the boards used for
cabinet to make more fundamental decisions policy direction and oversight in Sweden con-
about policy priorities. On the other hand, tain a variety of government officials who can
however, ministers will always have priorities, advance the ideas and interests of their own
so creating the large departments may assist organization and hence produce a certain
some interests and leave many others without amount of coordination without formalized
advocates in cabinet. interventions.
Advisory committees. One way to approach Agencies with portfolios relevant to coordina-
the problem of coordination of programs is to tion. Ministries or agencies can be developed
have a means of mutually representing the that have direct responsibility for coordinating
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 131


services for a specific target population or Task forces, working groups, etc. When
geographical area. At a minimum these organi- government is going to enter a policy area for
zations can act as advocates within govern- the first time, or when there is a great deal of
ment for the interests of those segments of the confusion about the best way to conceptualize
population. Examples of organizations of this a cross-cutting policy issue, a standard
type serving particular demographic groups response is to create a temporary “task force”
are the Administration on Aging in the United or “working group”. These are sometimes given
States, the Ministry for Family and Seniors and executive authority, for example some projets
the Ministry for Women and Youth in Germany, de mission in France or Projektgruppen in
and the former Ministry for the Middle Class in Germany, but generally these organizations are
France. Examples of these organizations serving oriented toward problem identification and
geographical areas are the “regional ministries” clarification – a central need for cross-cutting
in Canada, for example the Atlantic Canada policy issues (Timsit, 1988). A major recent
Opportunities Agency, the Ministry for example is the establishment of major agency,
Macedonia and Thrace in Greece, and the program and expenditure reviews in Canada
Ministry for the Mezzogiorno in Italy. (including one in social policy on pensions).
The development of organizations of this These appear to be very much recognitions of
type does bring attention to the needs of the need to think more broadly about the
demographic or regional groups, but it is far issues facing an aging population and the gov-
from a guarantee that those interests will be ernments that provide them services.
served in the way in which they need to be. If the cross-cutting issues can be “solved” in
These ministries and agencies often are not a limited period of time, or if a clear definition
perceived as central players in government, so of the issues can be developed in the limited
that, even though they may sit at the cabinet time span allowed for most of these special
table, they may not have much influence over organizations, then this approach is perhaps
major players such as the principal large social the most desirable manner to address the coor-
and economic ministries. In addition, these dination problem. They can provide a clear
ministries may provide some services for the focus and perhaps clear answers to a limited
target groups but still must ensure that services problem. If that success is not possible – and
provided from other ministries are compati- that is usually the case – then these organiza-
ble. In other words this may be just another tions either go out of business with little being
version of the division of services among accomplished or become simply another set of
departments. Further, as cross-cutting issues players in the complex network that will sur-
become more significant, the more traditional round these issues.
definitions and limitations characteristic of Another way to think about interministerial
existing programs may not push consideration organizations is as “virtual organizations” –
of the issues ahead quickly enough. organizations that may have no permanent
Interministerial organizations. Another obvi- structure and/or membership. This style of
ous means for coordinating the activities of organization has been advocated by some
existing programs and to explore the needs for Canadian civil servants interviewed in this pro-
new structures to cope with cross-cutting ject as a mechanism for generating coordination
issues is to develop organizations within the without creating yet another permanent struc-
interstices of existing organizations. All gov- ture. The argument is, in part, that creating
ernments have some forms of inter-ministerial another permanent organization with the goal
governance, although they differ in the extent of enhancing coordinate will itself soon require
to which those structures are articulated and additional coordination as issues change and
the power they can exercise over policies. What new patterns of interaction among organiza-
follows here is a brief enumeration of some of tion become the dominant concerns. Still,
those mechanisms. reaching agreements about when and under
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 132


what circumstances an organization will cease they could cooperate more effectively with
to exist is not the most pleasant thing for most organizations having similar objectives.
public officials – for them it is thinking about
Interministerial committees. Another flexible Processes
means for attempting to deal with cross-
cutting issues is to employ committees of the The above discussion has centered on the
organizations affected. Almost all governments impacts of structural remedies for creating
use some form or another of these committees. greater effectiveness of coordination activities.
This practice is perhaps best developed in Coordination also could be enhanced by a
France with committees existing at the level of variety of procedures. Procedures may not
officials, ministers (or their cabinets), and appear to be hierarchical in the same way as
finally to coordinate between the Prime structural changes, but do definitely rely on
Minister and the President. Coordinative authority for their capacity to make organiza-
committees of this sort have also been well- tions comply. If nothing else, procedures can
developed in the Antipodean systems. Like all force organizations to consider the implica-
committees formed to link existing organiza- tions of their policy choices for other organiza-
tions, committees of this type will have a diffi- tions and for clients. Just as structures cannot
cult time in advancing the definitions of policy guarantee success in coordination, these
far beyond those that already exist. If, as processes depend upon the commitment of the
argued, there is a need in many policy areas for principal participants to the goals of coordina-
some potentially sweeping redefinition of the tion. Otherwise, the processes may only per-
issues to be considered then these committees petuate or reinforce the independence of
are unlikely to change policies significantly programs and justify that perpetuation as the
(Schon and Rein, 1994). Granted broader pow- result of careful policy analysis.
ers than is usually the case, committees of this Budgeting. Budgeting reflects the priorities
type might be able to advance more innovative of government in dollar terms. Therefore, it
ideas about policy, but would tend to be only can be a central process for improving coordi-
as effective as their most committed member. nation of government priorities and programs.
Coordinating organizations. Another approach Given the tight fiscal constraints under which
to policy coordination is to develop special governments now function, budgeting may be
organizations with the task of ensuring coordi- the most important mechanism for setting pri-
nation for clients. One example of this was the orities and coordinating activities. The goals of
Model Cities program in the United States priority setting and policy coordination can be
which, during the War on Poverty in the 1960s achieved in at least two ways. One would be
and early 1970s, sought to identify the range of through the use of relatively technocratic
services available to residents of poor inner approaches such as those associated histori-
city neighborhoods and to coordinate them in cally with program budgeting. This involves
order to provide the full range of services to assessment of the relative costs and benefits of
clients. The time at which Model Cities was in any expenditures and their relationship with
operation was in many ways very much like the other spending programs. The alternative
present, in terms of the perceived need to approach is “Star Chamber” proceedings in
rethink an area of policy and to attack social which senior political and/or administrative
policy questions differently. For a variety of officials examine expenditure requests, requir-
reasons (financial, bureaucratic among others) ing the advocates of programs to justify their
Model Cities enjoyed only limited success, but expenditures, and then impose some collective
it was one means of incorporating both some priorities on public spending.
rethinking of the problems with service deliv- Budgeting in the contemporary political
ery. In particular, it forced a number of organi- and fiscal environment implies reducing
zations delivering services to think about how spending as well as allocating resources among
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 133


competing purposes. This change in the other ends. Evaluation tends to be directed at a
culture of budgeting, in turn, tends to reduce single program rather than at complexes of
the willingness of organizations to invest in programs. Even then, it can point to programs
policy coordination. When there is reduced whose effectiveness is limited by failure to
funding organizations tend to retreat to their coordinate with other programs, or by the
“heartlands” (Downs, 1967) and do not want absence of needed programs. If evaluation can
to use resources to extend their domains or to be oriented around target populations (Rossi
help achieve broader, government-wide, goals. and Freeman, 1989), rather than to specific
The constant threat of reductions and cutbacks programs, it can be a means for pointing to the
makes agencies hunker down and wait for bet- needs for coordination. Further, the definition
ter days. On the other hand, the changing of the relevant target population should, per-
managerial culture in government also means haps, be made external to the program itself if
that it is difficult for public sector organiza- the greatest benefits for managing cross-
tions to avoid demands for greater attention to cutting issues is to be obtained (Schneider and
coordination and policy coherence. Ingram, 1993).
Regulatory review. The Office of Manage- Evaluation will tend to be less useful in the
ment and Budget in the United States exercises case of cross-cutting policy issues because the
regulatory reviews over the activities of the goals and interdependencies within the con-
executive branch (McGarrity, 1991; Stevens, stellation of policies may be less clearly estab-
1995). Whenever an agency wants to issue reg- lished than in more linearly-defined policy
ulations (secondary legislation) OMB reviews areas. Conventional program evaluation may
these regulations in terms of their compatibil- find that a program is working effectively,
ity with the program of the president, their while from a broader, systemic perspective it is
cost, and their relationship to other, existing seriously deficient. Existing social insurance
sets of regulations. This is but one of several programs, for example, may provide pensions
mechanisms that governments use to monitor to the elderly efficiently and effectively but yet
and control secondary legislation to both not address at all the range of services a gray-
ensure the protection of individual rights and ing population requires, nor effectively relate
to coordinate regulations being issued by gov- the skills of aging population to a changing
ernment organizations. labor market.
Central agencies in several other countries “Coordination comments.” In Australia the
also exercise similar forms of regulatory review procedure of “coordination comments” has
and attempts to coordinate the activities of been designed and institutionalized in order to
their bureaucracies (see Pullen, 1994). This prevent members of cabinet from proceeding
control is often not as great a problem, given with departmental policies without adequate
that in cabinet governments a good deal of this coordination with their peers. Cabinet members
regulatory clearance is done at the cabinet level are required to circulate for comment any pro-
itself, or through the Prime Minister’s office. In posals they will bring to cabinet at least several
general, the greater the autonomy granted to days prior to the meeting. Other cabinet sys-
administrative agencies, as is the case for the tems have rules to avoid surprises in cabinet,
Scandinavian countries, the greater will be the but this method in Australia goes the furthest
need to institutionalize some mechanisms of in generating coordination. Although occur-
coordinating issuance of secondary legislation. ring at a lower level, one Canadian respondent
The question then becomes whether eco- in our study pointed out that, in the large
nomic, policy or political criteria will domi- departments created after the 1993 reorganiza-
nate the coordinating decisions, and how will tion, one of the emerging forms of coordina-
the values be structured. tion among sections within some departments
Evaluation. The evaluation of public policies is a formalized comments procedure for the
can be another process for producing coordi- component organizations. Again, we can see
nation, although it is usually directed toward that creating very large structures tends to
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 134


induce managerial problems of its own, even if coordination and coherence in government.
the strategy does make some contributions to That is certainly the conventional stance in
enhanced coordination. public administration and it is for the most
Another not unrelated, procedure can be part also the most appropriate stance from my
found in the Finnish government. In that perspective on governing. On the other hand,
system every month the cabinet meets infor- however, there are some questions that should
mally in a so-called “evening school” to discuss be asked about coordination as a value in admin-
collectively a major policy issue. Given the istration, especially about the cost/benefit ratio
informality of that setting, and the absence of that program designed to enhance coordina-
formal decision-taking at the meeting, this tion may have for government. That is, there
procedure promotes full discussion of the are multiple goals that must be pursued
issues and makes substantive coordination through the public sector and coherence may
possible in a less threatening manner. The for- be only one among many.
mal decision-making for government still
must take place within the Council of State,
Is Coordination Always the Answer?
but some of the more difficult issues may be
decided ahead of time in a setting in which
We have been pursuing coordination as if it
informal bargaining rather than defense of
were always the answer to problems facing the
departmental turf can be more possible.
public sector. We should, however, at least
entertain the possibility that enhancing coor-
Summary dination and coherence is not always a positive
contribution to resolving policy problems.
The relatively greater attention to mechanisms There may be some circumstances in which
of hierarchical control in this paper does not competing and incoherent approaches are
imply any greater intellectual appeal of these functional, rather than dysfunctional, for both
mechanisms of coordination. What it does indi- government and the governed. For example,
cate, however, is the relative level of attention of even though government funds a great deal
governments to these as opposed to other forms of scientific research, it almost certainly should
of coordination. Governments continue to not attempt to impose a single line of research
develop and implement means of coordination or establish an orthodoxy (Salbu, 1994).
using their own powers, even when there are Drawing the line between funding “good
network and even market mechanisms operat- science” and establishing such an orthodoxy
ing at the same time. On the one hand this may may be difficult, but also may be necessary. For
appear wasteful, but it does indicate both that example, the “War on Cancer” in the United
governments retain rather old-fashioned com- States demonstrated that settling on a single
mitment to authority and hierarchy and that line of research too quickly can waste resources
they also recognize their legal and political with little result.
responsibilities to provide the best possible gov- In addition to research funding, there may
ernment to their citizens. Their emphasis on law be other policy areas in which coordination is
and authority also indicates that they continue not especially desirable for government. For
to think of the public as citizens as well as example, in many policy areas there is far less
merely economic consumers of services. than certain knowledge (Dror, 1992) about
how to produce desired changes in the behav-
ior of individuals in society. The debates over
welfare policy and criminal justice in the
United States and numerous other countries
are extreme but not isolated examples of this
To this point in the paper I have been rather knowledge problem. Therefore, government
unambiguous in singing hymns of praise to may be well-advised at times to adopt an
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 135


explicitly experimental approach to policy and about accountability in the emerging world of
to minimize coordination and coherence. the more complex delivery of public service,
Some scholars have advocated such an exper- and if not, what principles are better suited for
imental approach (Campbell, 1982; 1988), this task?
and at times even some politicians have
argued that there is insufficient evidence to
make a long-term commitment to any partic- SUMMARY AND CONCLUSIONS
ular policy option. In addition, some program
areas can benefit from redundancy and dupli-
We began by noting that coordination is one
cation and excessive coordination can make
of the major goals of public administration,
the policy area more prone to error (Landau,
and has been so for some time. This paper has
discussed a number of the issues that arise
when that seemingly simple goal of improving
Accountability and Coordination coordination is advanced. These problems are
intellectual as well as practical, and require
We also must consider the classical question thinking about a number of alternative ways
of public accountability and the impact of of achieving the same policy and administra-
attempts to enhance coordination on the tive goals. Formal mechanisms of coordina-
capacity of governments to enforce account- tion are often the best ways of achieving those
ability. While greater coordination generally goals, but more informal methods also have
should enhance the efficiency and effective- their place, even within the public sector
ness of government programs, there are itself. The changing nature of government,
instances in which it does not, and within a with the emergence of the “New Public
complex multi-organizational policy environ- Management” and a range of other innova-
ment it may be difficult to identify where the tions in public policy and administration,
system has broken down. Accountability, at tend to place a greater emphasis on the infor-
least in its ex post facto sense, depends upon mal and market-based ideas, rather than the
the capacity of politicians and the public to formal.
identify who is responsible for any failures in One of the challenges – analytically and for
a program. The danger with coordination is government itself – is to determine when dif-
that by making everyone to some extent ferent approaches to coordination are most
responsible for programs in the end no one appropriate, and develop an understanding
is actually responsible and it becomes much of the contingencies that may be operative.
easier to evade responsibility. The three approaches we have discussed here
The dangers to accountability may arise reflect different concepts and different loca-
through several of the methods used for tions in the political process. The hierarchical
improving coordination. For example, finan- approach, the most common approach,
cial accountability becomes difficult to enforce appears crucial in program areas that are pecu-
when funds from several departments are min- liarly public, e.g. defense, foreign affairs,
gled in order to create adequate resources for revenue collection and the like. The other two
comprehensive attacks on major cross-cutting approaches appear more applicable to that
policy problems, for example drug enforce- other large range of public programs that are
ment or urban regeneration. How does gov- (at least in principle) marketable or which are
ernment ensure that money is being spent in surrounded by significant networks of private
the ways intended when it was appropriated? If actors. This is as yet a gross generalization, and
it is not, who should be held responsible for substantial refinement is needed to make it at
the misallocation? Perhaps even more funda- all useful. Still, it may point to ways of address-
mentally, are conventional ideas about parlia- ing coordination and improving both the
mentary accountability the best ways to think practice and the analysis.
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 136



1. For example, President Jimmy Carter attempted to 6. P. (2004) Joined-Up Government in the Western
divide the (then) Veterans Administration by moving the World in Comparative Perspecitve: A Preliminary
educational components into the Department of Literature Review and Exploration, Journal
Education, the health and social service components into of Public Administration Research and Theory
the Department of Health and Human Services, and the 14, 103–39.
residual elements into the Department of Defense. He 6, P. (2001) Joined Up Government Around the
failed rather abjectly at this rationalization and coordina-
World, paper presented at Conference on Joined-
tion in the face of pressure from veterans’ organizations.
2. In a study I did with several colleagues over 25 years
Up Government, Oxford, October 30.
ago we found that social service agencies tended to hang Adema, W. (1997) What Do Countries Really Spend
on to clients and not to refer them for needed services. on Social Policies: A Comparative Note, OECD
This was true even in the context of the Model Cities Economic Studies 28, 153–67.
Program, an attempt to provide more comprehensive ser- Alexander, E. R. (1993) Interorganizational
vices to the residents of poor neighborhoods in American Coordination: Theory and Practice, Journal of
cities. When interviewed agency leaders argued that it was Planning Literature 7, 328–43.
too “dangerous” for them to risk losing clients. Allard, C. K. (1990) Command, Control and the
3. For example, the “Resortsprinzip” in German govern- Common Defense (New Haven: Yale University
ment is another indication of this segmentation of
4. For example, in recent consulting work for the
Barrett, S. and C. Fudge (1981) Policy and Action:
Finnish government creating policy integration emerged Essays on the Implementation of Public Policy
as a dominant issue. This was true even though this gov- (London: Methuen).
ernment has a long tradition of cooperative management Bogason, P. (1991) Control for Whom?: Recent
and policy development. Adventures on Governmental Guidance and
5. One common pattern has been to create very large Control, European Journal of Political Research
ministries for health and social services. This single min- 20, 189–208.
istry may have several ministers and be responsible for up Bouckaert, G., D. Ormond and B. G. Peters (2000) A
to half of the total public budget. This creates a tremendous New Governance Agenda (Helsinki: Ministry of
managerial overload for the senior public servants, and
may do so with little real improvement in coordination.
6. The Canadian Food Inspection Agency was created in
Callender, G. and J. Johnston (1998) Governments
1997 to bring together related activities from Agriculture, and Governance: Examining Social and
Fisheries, Health and Industry ministries. It also absorbed Economic Autonomy in Malaysia and Singapore,
some functions that had been largely provincial. Asian Journal of Public Administration 20, 151–72.
7. The Office of Management and Budget prepares each Campbell, C. and G. Szablowski (1979) The
year a government-wide set of priorities to correspond to Superbureaucrats: Structure and Behaviour in
the more specific priorities coming from the agencies. The Central Agencies (Toronto: Macmillan of Canada).
OMB document tends to be long on pious hopes and short Campbell, D. T. (1982) Experiments and Arguments,
on specifics. Knowledge: Creation, Diffusion, Utilization 3, 327–37.
8. In some ways ministries of foreign affairs have not
Campbell, D. T. (1988) The Experimenting Society,
become central agencies so much as all ministries have
developed their own international sections and interna-
in Campbell, Methodology and Epistemology in
tional political views. This is in contrast to the role of envi- the Social Sciences: Selected Essays (Chicago:
ronmental agencies. University of Chicago Press).
9. In a “real” contracting arrangement the actors involved Chisholm, D. (1989) Coordination Without Hierarchy
would have more of an opportunity to walk away if an (Berkeley: University of California Press).
acceptable deal could not be reached. In the public sector Commission on Social Justice (1994) Social Justice:
the legal mandates restrict the exit option. Strategies for National Renewal (London: Vintage).
10. There is, however, evidence that countries are still Considine, M. (1992) Alternatives to Hierarchy: The
behaving more autonomously than the proponents of Role and Performance of Lateral Structures
globalization might lead us to believe.
Inside Bureaucracy, Australian Journal of Public
11. This can be seen most clearly from the “instru-
ments” perspective on public policy (Salamon, 2000).
Administration 51, 309–20.
12. For example, in my earlier research on the Model Davis, G. (1998) Executive Coordination Mechanisms,
Cities program in the United States, professionals tended in P. Weller, H. Bakvis and R. A. W. Rhodes, eds.,
to refer clients to others in the same profession rather than The Hollow Crown: Countervailing Trends in Core
serving the “whole client” as intended by the program. Executives (London: Macmillan).
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 137


de Montricher, N. (1999) L’amenagement du The Brussels Dimension (Oxford: Oxford

territoire (Paris: La Decouverte). University Press).
Derlien, H.-U. (1991) Horizontal and Vertical Kickert, W. J. M. (1995) Public Governance in the
Coordination of German EC-Policy, Hallinnon Netherlands: An Alternative to Anglo-American
Tutkimus, 3–10. Managerialism, Public Administration 75, 731–53.
Doern, G. B. (1993) From Sectoral to Macro Green Landau, M. (1969) Redundancy, Rationality and the
Governance: The Canadian Department of the Problem of Duplication and Overlap, Public
Environment as an Aspiring Central Agency, Administration Review 29, 346–58.
Governance 6, 172–94. Laronche, M. (2000) Quarante pratiques fiscales
Dowding, K. (1995) Model or Metaphor? A Critical deloyales repertoirees au sein de la seule Union,
Review of the Policy Network Approach, Political Le Monde, 23 Mai.
Studies 43, 136–58. Larsson, T. (1986) Regeringen och des kansli
Downs, A. (1967) Inside Bureaucracy (Boston: (Stockholm: Studentlitteratur).
Little, Brown). McGarrity, T. (1991) Reinventing Rationality: The Role
Elmore, R. F. (1979) Backward mapping: of Regulatory Analysis in the Federal Bureaucracy
Implementation Research and Policy Decisions, (Cambridge: Cambridge University Press).
Political Science Quarterly 94, 601–16. Mackie, T. T. and B. W. Hogwood (1985) Unlocking
Dror, Y. (1992) Policymaking Under Adversity (New the Cabinet: Cabinet Structures in Comparative
Brunswick, NJ: Transaction Books). Perspective (London: Sage).
Fortin, Y. (2000) Contractualisation dans le secteur Marin, B. (1990) Generalized Political Exchange:
public (Paris: Decouvert). Preliminary Considerations, in B. Marin, ed.,
Greer, P. (1994) Transforming Central Government: Generalized Political Exchange: Antagonistic
The Next Steps Initiative (Buckingham: Open Cooperation and Integrated Policy Circuits
University Press). (Frankfurt: Campus Verlag).
Hall, R. et al. (1978) Interorganizational Marton, S. (2000) The Mind of the State
Coordination in the Delivery of Social Services, (Gothenberg: University of Gothenberg).
in L. Karpik, ed., Organization and Environment: Metcalfe, L. (1994) International Policy Co-
Theory, Issues and Reality (Beverly Hills: Sage). ordination and Public Management Reform,
Hallerberg, M. and S. Basinger (1998) Inter- International Review of Administrative Sciences
nationalization and Changes in Tax Policy in 60, 271–90.
OECD Countries, Comparative Political Studies Miller, G. J. (1992) Managerial Dilemmas: The
31, 321–52. Political Economy of Hierarchy (Cambridge:
Hogwood, B. W. and L. Gunn (1984) Policy Analysis Cambridge University Press).
for the Real World (Oxford: Oxford University Mueller-Rommel, F. (2000) Germany, in Peters,
Press). Rhodes and Wright, Administering the Summit
Jacquet, P. (1998) L’union monetaire et la coordina- (London: Macmillan).
tion des politiques macroeconomiques, in Nathan, R. P. (1975) The Plot That Failed: Nixon and
Conseil d’Analyse Economique, Coordination the Administrative Presidency (New York: Wiley).
europeenne des politiques economiques (Paris: La Niskanen, W. (1994) Bureaucracy and Public
documentation francaise). Economics (Cheltenham: Edward Elgar).
Jennings, E. T. and D. Crane (1994) Coordination OECD (1993) Internal Markets (Paris: Organization
and Welfare Reform: The Quest for the for Economic Cooperation and Development),
Philosopher’s Stone, Public Administration Market–Type Mechanisms, Vol. 6.
Review 54, 341–8. Olsen, J. P. (1987) Organized Democracy (Oslo:
Jerome-Forget, M., J. White and J. M. Wiener (1995) Universitetsforlaget).
Health Care Reform Through Internal Markets Page, S. (2003) Entrepreneurial Strategies for Managing
(Washington, DC: The Brookings Institution). Interagency Collaboration, Journal of Public
Katz, J. L. (1995) Key Members Seek to Expand State Administration Research and Theory 13, 311–40.
Role in Welfare Plan, Congressional Quarterly Painter, M. (1987) Steering the Modern State:
Weekly Report 53 (2), 159–62. Changes in Central Coordination in Three
Kassim, H., B. G. Peters and V. Wright (2000) Policy Australian State Governments (Sydney: Sydney
Coordination in the European Union; The National University Press).
Dimension (Oxford: Oxford University Press). Peters, B. G. (1996) The Future of Governing; Four
Kassim, H., A. Menon, B. G. Peters and V. Wright Emerging Models (Lawrence: University Press of
(2001) Policy Coordination in the European Union: Kansas).
Peter-3383-Chapter-06.qxd 6/8/2006 7:47 PM Page 138


Peters, B. G. (2002) Contracts and Resource Savoie, D. J. (1999) Governing from the Centre.
Allocation in Public Governance, In A. Heritier, (Toronto: University of Toronto Press).
ed., Common Goods: Reinventing European and Schneider, A. B. and H. Ingram (1993) Social
International Governance (Lanham, Md: Rowman Construction of Target Populations: Implications
and Littlefield). for Politics and Policy, American Political Science
Peters, B. G. (2006) Horizontal Governance: The Review 87, 334–47.
Search for Coordination and Coherence (Lawrence: Schon, D. and M. Rein (1994) Frame Reflection:
University Press of Kansas). Resolving Intractable Policy Issues (New York:
Peters, B. G. and J. Pierre (2000) Administrative Basic Books).
Reform and the Civil Service (London: Routledge). Smith, G. (1991) The Resources of a German
Peters, B. G., R. A. W. Rhodes and V. Wright (2000) Chancellor, in G. Jones, ed., West European Prime
Administering the Summit: Support for Presidents Ministers (London: Frank Cass).
and Prime Ministers (London: Macmillan). Smith, M. J., D. Marsh and D. Richards (1993)
Pierre, J. and B. G. Peters (2000) Governance, Policy Central Government Departments and the Policy
and the State (Basingstoke: Macmillan). Process, Public Administration 71, 567–94.
Pressman, J. L. and A. Wildavsky (1984) Imple- Stevens, L. N. (1995) Regulatory Reform: How Can
mentation, 2nd ed. (Berkeley: University of Congress Assess the Administration’s Initiatives
California Press). (Washington, DC: General Accounting Office,
Pullen, W. (1994) Eyes on the Prize, International July 18) GAO/T-95–206.
Journal of Public Sector Management 7, 5–14. Taylor, S. (1984) Making Bureaucracies Think: The
Regens, J. L. (1988) Institutional Coordination of Environmental Impact Statement Strategy of
Program Action: A Conceptual Analysis, Inter- Administrative Reform (Stanford: Stanford
national Journal of Public Administration 11, 135–54. University Press).
Sabatier, P. A. (1988) An Advocacy-Coalition Model Thompson, G., J. Frances, R. Levacic and J. Mitchell
of Policy Change and the Role of Policy-Oriented (1991) Markets, Hierarchies and Networks
Learning Therein, Policy Sciences 27, 123–35. (London: Sage).
Salamon, L. M. (2000) Handbook of Policy Instruments Timsit, G. (1988) Les autorites independents admin-
(New York: Oxford University Press). istratives (Paris: Presses Universitaires de France).
Salbu, S. R. (1994) Regulation of Drug Treatments Toonen, T. A. J. (1985) Implementation Research
for HIV and AIDS, Yale Journal of Regulation and Institutional Design: The Quest for
11, 401–52. Structure, in K. Hanf and T. A. J. Toonen, eds.,
Savoie, D. J. (1995a) Globalization, Nation States and Policy Implementation in Federal and Unitary
the Civil Service in B. G. Peters and D. J. Savoie, eds., Systems (Dordrecht: Martinus Nijhoff).
Governance in a Changing Environment (Montreal: Wolgast, E. (1992) Ethics of an Artificial Person: Lost
McGill/Queens University Press). Responsibility in Professions and Organizations
Savoie, D. J. (1995b) Central Agencies: Looking (Stanford: Stanford University Press).
Backward (Ottawa: Canadian Centre for Zito, A. (1999) Making European Environmental
Management Development). Policy (London: Macmillan).
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 139



Political science has long wrestled with the States has certainly had its share of both of these
question of whether the characteristics of types of solutions, but is not alone in its attempt
decision-making processes affect the outcomes to find a solution for one of the central set of
of those decisions. There are usually many policy decisions for government.
other factors contributing to outcomes besides
processes, which means that any given process
may have different outcomes at different times
or in different circumstances. As a result, it has
been difficult for political scientists to demon-
strate that decision-making processes impact
outcomes. This chapter examines the question What are budget process rules? What is their
of the impact of process on outcomes, using as function? Budget process rules describe the
an example budget process in the United States order in which budget decisions will be made,
at the national level, with particular attention the timeliness of the decision-making, which
to the post 1998 changes and recent budgetary actors make which decisions, and the con-
outcomes. While this case does not provide straints under which the decisions will be
an airtight design, it does provide a kind of made. These rules provide definitions, for
natural experiment. example, how balance will be defined. Budget
Although the focus will be on the United rules determine what budget claims will be
States, there will be some comparisons with compared with what other claims, what evi-
process reforms in other industrialized democ- dence and whose evidence will be brought to
racies. All the countries in the OECD world, bear on the decision, and how much time there
and indeed in the world more broadly, are will be for deliberation on those decisions.
struggling to find budget processes that produce A good budget process is stable from year to
efficient allocation of funds but yet are not year, it provides necessary definitions that are
dominated by technical considerations. The known to and used by all parties, and it ensures
“ideas in good currency” about budgeting discipline, so that revenues do not exceed
appear to cycle between highly rationalistic expenditures over the long haul. Budgets may
solutions, such as PPBS, to simple, if effective, include borrowing, but are not balanced unless
mechanisms to reduce public expenditure there is a plan and revenue sources to pay off
(Gray, Jenkins and Segsworth, 2002). The United that debt as it comes due. To ensure discipline,
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 140


the process normally outlines steps to estimate for all the participants, who must have a chance
(and revise) revenue estimates, and breaks up to learn and use the process. The process must
pieces of the decisions on spending and gives be visible to the public.
them separate limits, so that, when added up, In evaluating the consequences of rules
they do not exceed the total revenue available. changes, one has to look initially at the rules
A good process is reasonably neutral among themselves. They can be described in terms of
policy goals and political interests, providing a how they compare to the idealized process
forum for the articulation, discussion and res- described here. For example, how neutral are
olution of necessary policy issues without bias- they with respect to outcomes? How consistent
ing the results in advance. It combines and predictable are they from year to year?
information coming up from programs and What steps are included in the budget process
professional staff with centrally determined to assure that spending is controlled by a
policy and discipline. Further, in a democracy, revenue estimate? Is there a procedure for
budget process must assure that minority par- prioritization of expenditures? Second, one has
ties are heard, and that decisions are not made to look at the changes. Did new rules supple-
that prevent different factions that might come ment or supplant the old ones? Were the old
to power in the future from also making policy rules kept but their intent evaded? Were the
decisions. rules “gamed”? Was it a minor rule that was
Maintenance of neutral rules allows the avoided or voided, or a major one, was it one
decision-making to proceed in a politically rule or a cluster of them, was the change infor-
charged environment. Without it, conflicts may mal and temporary, or formal and long term?
bring the budget to a standstill, as often occurs Budget process change is a complex variable,
in less stable democratic systems. Without but it must be understood in some detail if its
neutral rules, minorities may feel that they consequences are to be examined.
cannot win or influence decisions, and hence
cease to participate or work at disrupting the
process. Politicians are often tempted to bias
the rules to facilitate some immediate policy
goal rather than maintaining neutral rules. For
example, the requirements of the stability pact
in the Eurozone have been used not only to This study focuses on the federal budget
stabilize the budget but also to disadvantage process in the United States, with particular
some social interests (Stolfi, 2006). attention to the period from 1998 to the end of
Some rules can be changed without damag- 2004. However, in order to see a rule change,
ing the collective policy framework necessary one must also examine what budget rules pre-
for democratic decision-making, others ceded the change and what outcomes were
cannot. There must be time to consider legisla- associated with the period in which the pre-
tion, there must be thoughtful prioritization, ceding process was in effect. One also needs to
there must be a role for the minority as well as know the degree to which those prior rules were
majority; there must not be decisions that lock implemented. Thus, the study of budgeting in
in current policy priorities and prevent future the United States compares three periods, one
democratically elected officials from making from 1986 to 1990, when Gramm Rudman
changes. Perhaps one exception to that general- Hollings was in place, one from 1990 to 1998
ization is the use of earmarked taxes and trust (when the Budget Enforcement Act was in
funds as a means of justifying certain types place), and the final one, from 1998 to 2004,
of taxation (Patashnik, 2000). There must be during the decline and termination of the BEA
good quality information provided in a timely and the rise (return?) of ad hoc budgeting.
fashion, to decision makers and to the media Innovations in budget practices in the United
and public. The process must be firmly enough States have been diffused widely in the past,
in place that it is predictable from year to year and some of these same ideas have been used
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 141


in other countries to address similar issues. be demonstrated, such that the process addresses
Thus, this one case study is just that but it also issue A and the outcome for issue A changes
illustrates some generalities about budgets and when the process changes, this should be con-
the impact of rules on performance. strued as further evidence supporting a relation-
ship. Finally, if alternative explanations besides
the changes in budget process can be ruled out,
Data Sources
in part or in whole, the case would be even
The information on budget process and imple-
A second assumption is that processes can be
mentation comes from laws, journal articles,
more or less formal, with informal processes
newspaper accounts, newsletters from public
easier to change and less durable or predictable.
interest research groups, and reports from think
Informal processes may carry out the intent of
tanks and advocacy groups. Efforts were made
the formal rules or contradict them or supplant
to draw on both sides of the political spectrum
them. Budget rules may be more or less actively
as well as some sources closer to the middle.
enforced, mechanisms of enforcement may
Data also comes from government publica-
exist, but be seldom used, or may be actively
tions, such as the Historical Tables of the U.S.
evaded, through distorted data (gaming the
Budget and budget hearings, and from on-line
system) or definitional changes or various
news magazines focusing on the federal gov-
kinds of budget gimmicks or accounting rules
ernment, such as Government Executive and
changes. The impact of budget rules on out-
The Hill.
comes depends to some extent on the degree
of enforcement in practice. Budget processes
Assumptions do not enforce themselves, they require the
consent and will of the actors to give them life
Several assumptions underlie the analysis. and force. For the purposes of this study, a shift
First, budget processes do not determine out- from rule enforcement to rule non-enforce-
comes, they influence them, making certain ment is considered a change in budget process,
outcomes easier to achieve and others harder. even without a formal, legal change.
This influence may be greater and more visible
at some times than others. Since it is difficult
to either prove or falsify influence or measure The Intervention
it in any precise fashion, the conclusions in this
essay are inferential. This study is about a natural experiment, a set of
The logic of the analysis depends on four changes in budget rules. Budget processes are
kinds of argument. First, there were three phases often overlaid on each other, rather than com-
of budget process change, each associated with a pletely removed and started afresh, but
set of outcomes. If deficits increase during the key features may be added or subtracted. All
first phase, decrease and disappear during later budget processes were overlaid on the 1921
the second, and reappear, almost instantly, in the Budget and Accounting Act, which created
third, even if there are many other things going executive budgeting, so that the president and
on during those time periods, there would be his budget office prepared a budget request for
evidence of a likely relationship between process legislative consideration. Similarly, the changes
and outcomes. Second, if the actors who want to inherent in the 1974 Congressional Budget
accomplish a particular policy goal first change and Impoundment Control Act, in which the
the process and then accomplish their goal, it Congress enhanced its ability to question that
looks as if they believe they had to change the presidential proposal and come up with its
process, formally or informally, to accomplish own if it wished, underlay the later changes.
their goal. Such behavior provides additional From the point of view of budget process, the
strength to the argument for a linkage between central feature of the 1974 Congressional
process and outcomes. Third, if mechanism can Budget and Impoundment Control Act was the
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 142


preparation by Congress of an annual budget Dependent variables

resolution to guide committee decisions and
assure some fiscal discipline. This study of budgeting in the United States
In 1986, a new set of procedures was laid on examines two sets of outcomes of the budget
top of the preceding laws to try to reduce the level process. One is traditional budgetary out-
of deficits. Its central feature was an effort to force comes, such as budgetary balance and the size
spending reductions in order to reduce the of and the composition of public spending.
annual deficit by a targeted amount. To do this, The other is the degree to which the budget
the process, named after its legislative sponsors process is consistent with, and implements,
Gramm, Rudman, and Hollings, provided for an democratic governance.
automatic, across-the-board cut if more targeted One of the most basic aims of budget process
and selective cuts were not made or were not is to match resources to expenditures. On one
deep enough. The threat of such across-the- end of this continuum would be budgetary bal-
board cuts, affecting both Republican and ance, defined in some consistent manner, and at
Democratic priorities, was intended to force the other, uncontrolled and huge deficits, repre-
spending reductions. Gramm, Rudman, and senting not so much spending for real emergen-
Hollings did not work well from a variety of cies as a loss of discipline. Somewhere near the
perspectives, including the exemption of many midpoint would be controlled deficits, for spec-
major programs from the threat of across-the- ified policy purposes and true unforeseeable
board cuts, so that so called across-the-board cuts emergencies. Indicators of this continuum
fell on relatively few programs, and those were hit include changes in the size of the deficit, in
disproportionately hard, and could not effec- absolute and relative terms, and the amount and
tively be cut year after year. The rules of Gramm, purpose of borrowing.
Rudman, and Hollings were generally evaded. A second outcome can be labeled the private
In 1990, a new process was instituted, versus public benefit continuum. It is anchored
the Budget Enforcement Act, which had a dif- by spending for narrow interest or constituency
ferent approach to deficit reduction. It was group benefits or short-term partisan purposes,
more even-handed, in that it included controls on one end, and by broad-based benefits for
not only over spending but also over revenue. society on the other. Most outcomes fall between
The central features of the Budget Enforcement the ends of the continuum, but different budget
Act were spending caps in the discretionary processes or differential enforcement of the rules
side of the budget and required offsets for can facilitate a shift more in one direction or the
increases in entitlement spending or decreases other. It may be difficult to come to agreement
in revenues (Meyers, 1992). It was forbidden on what a broad-based policy outcome for the
to reduce revenue without increasing it some- public good looks like, but, even with some noted
where else or reducing mandatory spending to differences in definition, it has been possible to
compensate. track the amount of pork in the budget over
By 1998, partly as a result of a booming time, where pork is an indicator of narrow, often
economy and partly the result of the budget partisan, benefits. Earmarking in legislation is
discipline required by the budget enforcement used here (and by policy analysts more generally)
act, the federal government achieved bud- as an indicator of these narrower benefits.
getary balance, (more or less), removing the A third outcome of the budget process is the
pressure to maintain the spending caps, which size of and composition of public spending,
were then widely ignored (see Palazzolo, 1999). the scope of government. On the one end is
In 2002, the Budget Enforcement Act lapsed, small government, low taxes, limited regulatory
and it was not renewed. During this period, powers, and few entitlements; on the other,
Congress has also had difficulty enforcing or larger government, higher taxes, more active
carrying out some of the key provisions of the regulatory functions, and many substantial enti-
1974 act. What were the consequences of these tlements. Budget rules can influence the ease of
rules changes? moving in one direction or the other.
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 143


These three outcomes are traditionally Comparative Perspectives

thought of as budgetary outcomes, but there is
a sense also in which the budget process is itself In 2005 the federal deficit in the United States
an outcome. Because it is itself such an impor- was among the highest in the world, perhaps a
tant part of public policy-making, the way the function of the absence of clear rules for con-
budget is put together says much about gover- trolling expenditure increases and revenue
nance. On the one hand, budget process has a reductions. A number of other countries that in
direct effect on the quality of decision-making the past had had significant budget deficits –
and, on the other hand, it has a direct effect on for example Italy, Belgium and Canada – are
democratic processes. now managing their budgets much better and,
Budget process influences the quality of in the case of Canada, have been amassing sig-
decisions to the degree that it frames appropri- nificant budget surpluses. Is this just good
ate comparisons and trade-offs and does an fortune, or does the presence of clear controls
effective job at prioritization. The quality of and rules give some backbone to public officials
information provided to decision-makers and who might otherwise find deficits more palat-
the timeliness of that information, combined able politically? Or have there been changes in
with the time allowed for decisions to be made, the fiscal cultures of countries that now stress
all influence the quality of budgetary decision the importance of maintaining fiscal discipline
making. If assumptions are watered down, and using market-type forces to control the
revenue projections distorted up or down, the public sector (see HM Treasury, 2005)
degree of feedback from decisions on the The Maastricht Treaty and the Stability and
economy, and hence on future revenues, is Growth Pact have helped governments in many
pulled out of the air, then the quality of deci- European countries stand up to special inter-
sions will suffer. If accounting rules are ests and to control spending. Even countries
changed or definitions blurred, or if there is a who are not members of the Eurozone have
difference in assumptions or rules between had their budgeting behaviors influenced by
decision-makers, so that all are not acting the central place of Brussels in public finance
from the same set rules or assumptions and (Albert-Roulhac, 1998). One major study of
are talking past each other, then the quality budgeting (Wanna, Jensen and deVries, 2003)
of decision-making declines. If the balance argues, however, that rules and structures have
between political policy goals on one hand become less important than in the past in con-
and technical input on needs on the other gets trolling budgets. Rather, their central finding is
out of whack, in either direction, the quality that evidence and argument tend to dominate
of decisions will suffer. over formality and power. If this is the case then
If the rules are fair, neutral, and allow all par- the failures of American budgeteers in recent
ties to introduce legislation, propose amend- years may be more a function of their inability
ments, and have a say, if the process does not in coming up with good arguments, or, perhaps
lock in outcomes that prevent future political even more so, their commitment to ideas about
coalitions from making changes according to reducing taxation rather than to ideas about
the wishes of their constituencies, if there is controlling deficits.
sufficient accurate information presented in a It may well be, of course, that the United
timely fashion and there is time to deliberate States is different. The complexity of the political
on proposals, then the process helps imple- system and the magnitude of the budget
ment democracy. If it fails to do those things, if process may make reliance on argument and the
decisions are made in secret, if the process is culture of organizations a less viable solution
not open and not accountable, if the rules are than in parliamentary democracies, especially
not made known in advance, if some are kept small, homogenous, parliamentary democra-
out, then the processes of representation and cies. Somewhat paradoxically for a country
accountability are corroded, and democratic often priding itself on being opposed to strong
governance is compromised. regulations and bureaucracy, its own internal
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 144


governance processes may require just those Budget Enforcement Act quickly eroded. Spending
types of controls if they are to resist the exceeded the caps by a wide margin. The key
centrifugal pressures within the political system. features of the BEA rules were ignored or
gamed. One common practice was to treat some
routine expenditures as if they were emergen-
CHANGES AND OUTCOMES cies, since emergencies were not required to be
offset by revenue increases or other decreases in
1986 to 1990: Gramm Rudman spending. By calling the expenditure, such as the
Hollings and Deficit Reduction Targets census, an emergency (which it was not except
in political terms) budgeteers did not have to
Gramm Rudman Hollings set annual targets for find revenue to cover the expense, which then
deficit reduction, but those targets were rou- added to the deficit.
tinely missed. Efforts to avoid sequestration, the As budget scholar Allen Schick observed,
technical term for across-the-board cuts, led to
The arrival of a surplus a few years ago triggered a spend-
intense gaming of the system, pushing expendi-
ing frenzy that vitiated the discretionary spending caps
tures into following years to make the budget established by the 1990 Budget Enforcement Act and
look more balanced than it was, and then push- made a mockery of the BEA requirement that increased
ing them back when the next year arrived, for spending be offset by cuts in other spending or by
example. The legislation was not even handed, it revenue increases. In 2000 and 2001, discretionary spend-
ing soared more than $200 billion above the legal limits
approached the issue of deficits from the per-
on annual appropriations. The caps expire at the end of
spective of forcing deep spending cuts, but there 2002 – which will at least enable politicians to be more
was little consensus on which programs to honest about what they are doing. They no longer have to
cut or how deeply to cut them, so the unthink- pretend that the census and other ongoing operations of
able, the sequester process, was supposed to be government are national emergencies. (Allen Schick, The
Brookings Review, Spring, 2002. Vol. 20 No. 2 pp. 45–48.
invoked. The sequester process was supposed
The Deficit That Didn’t Just Happen: A Sober Perspective
to be neutral, in the sense of affecting all pro- on the Budget)
grams, but while it didn’t have a political bias, it
tended to cut some programs disproportionately Huge surpluses turned nearly immediately
because they hadn’t been exempted. The result into large and growing deficits. This process
was illogical, without serious prioritization. The began before 9/11/2001 and hence cannot be
process might have reduced deficits if it had in attributed solely to responses to that event.
fact been implemented, but, with widespread OMB reports in the Historical Tables of the U.S.
evasion of its rules, Gramm Rudman Hollings Budget for 2005 that the on-budget surplus for
proved unable to control deficits. fy 2000 was 86.626 billion dollars, but by 2001
From 1990 to 1998, under the Budget there was a deficit of 33.257 billion (fiscal year
Enforcement Act, the norms of budgetary 2001 spending is controlled by decision-making
balance were strengthened. Spending caps that occurs the previous year). Figures for
were generally treated seriously, forcing a deficits for 2002, 2003, and 2004 were reported
number of difficult trade-offs, many in full as 317.456 billion, 536.128 billion, and 674.766
public view. The always present temptation to billion. These numbers are so huge and growing
reduce taxes was curtailed; taxes were actually so rapidly as to suggest they are out of control.
increased to help balance the budget. By 1998, Deficits are listed in Table 7.1, along with the
with the help of a growing economy, the budget process in effect at the time.
budget was more or less balanced; surpluses As former CBO staffer Phil Joyce observed,
were projected for many years in the future. there was insufficient consensus after 1998 to
Public debate focused on what to do with the come up with congressional targets and adhere
surpluses, whether to reduce taxes or fix social to them. The lack of agreement meant that
security, or buy down the national debt. the budget process was operating without a
After 1998, because of the surpluses, the con- notional budget constraint. As a result, Joyce
sensus to adhere to the spending caps in the claimed “no one knows how much is enough – or
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 145


Table 7.1 Deficits/surpluses, on budget, from wars in Afghanistan and Iraq were paid for
1980 through 2004 (millions) primarily through supplemental appropria-
Budget process in place year deficit/surplus tions, which were not offset by reductions in
Congressional budget 1980 −72,710 spending elsewhere, and added to the deficit,
and Impoundment Control 1981 −73,948
but they were in the range of 60 to 80 billion a
1982 −120,040
1983 −208,014 year, nowhere near the total of the deficit.
1984 −185,629 Three sources are often cited as the cause of
1985 −221,671 deficits, a slow growing economy, increases in
Gramm Rudman Hollings 1986 −237,946 spending, and reductions in tax rates. Of these,
1987 −169,298
the slow growing economy had the least impact,
1988 −193,951
1989 −205,910 because while revenue growth may have
Budget Enforcement Act 1990 −277,786 slowed down some, so did interest rates, which
1991 −321,525 lowered the costs of government borrowing.
1992 −340,463 Expenditures did increase, but not just for the
1993 −300,434
wars and intelligence efforts; entitlements
1994 −258,904
1995 −226,387 also grew rapidly, especially for Medicare and
1996 −174,061 Medicaid. Other spending increased as well.
1997 −103,322 Overall, spending increased from 18.4 percent
1998 −29,982 of GDP in fiscal year 2000 to 19.8 percent
BEA key provisions not 1999 +1,873
in 2004, a startling increase in the size of gov-
enforced 2000 +86,626
2001 −33,257 ernment spending. But this increase would
BEA expires 2002 −317,456 have been more than offset if revenues had
2003 −536,128 been allowed to remain the same percent of
2004 −674,766 (est) GDP they had been in 2000. If taxes had not
Source: Historical tables of the U.S. Budget, 2005, been reduced as a proportion of the economy,
table 1–1. there would have been no deficit in 2004,
instead there would have been a surplus of more
than 100 billion dollars (Center for American
too much – spending. And nobody knows – or Progress, An analysis of the recent deterioration
everybody knows, but nobody agrees – when of the fiscal condition of the U.S. Government.
the deficit is too large or the surplus too small” Washington, DC September 2004, p. 9, accessed
(Philip Joyce “Federal Budgeting After online at
September 11th: A Whole New Ballgame, or is Increases in entitlement spending for Medicare
it Déjà vu All Over Again?” in Public Budgeting and Medicaid should not be laid at the feet of
& Finance (Winter, 2005). the budget process, nor probably should the
To the extent that the increase in deficits increases in spending on the military post
since 1998 resulted from 9/11 and terrorism 9/11; but the reductions in taxation were made
related expenditures, one could argue that they decidedly easier by the removal of the con-
are not related to the budget process and would straint that required either spending reduc-
have occurred no matter what the budget tions or alternative revenue increases to
rules were. However, as Brian Reidl from compensate for tax reductions. These tax
the conservative Heritage Foundation has reductions contributed in a major way to the
argued, most of the spending increase since rise of deficits.
2001 was not for terrorism related expenses. While some of the spending increases were
The recent round of deficits was the result of not related to the budget process and would
tax cuts that were not offset by reductions in have occurred even if there had been no changes
spending, and by spending increases not only in the budget process, some of it may well
for the war in Afghanistan, and later in Iraq, have been related to the changing process. Many
but also for domestic spending, including observers noted the increase in the amount of
pork, which began to surge after 1998. The pork type spending, or earmarks after 1998.
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 146


Brian Reidl of the Heritage foundation reported “People say earmarks are part of the grease that makes
the number of pork projects skyrocketed from government work, and to some extent, that’s true,” said
John Cogan, a Stanford University economist and
under 2,000 five years ago [1998] to 9,362 in the former OMB deputy director who headed the initial
2003 budget. Total spending on pork projects drafting of Bush’s budget blueprint as a member of the
has correspondingly increased to over $23 transition team. “But 6,183 earmarks is an awful lot of
billion. This trend continued in the fiscal year grease.” Government Executive Magazine – 2/13/01
(FY) 2004 appropriations bills, which include Bush team eyes earmarks, may push biennial budget
plan, February 13, 2001, James Barnes, National Journal
approximately 10,000 earmarks (Another 2002: Expiration of the BEA and Ad Hoc Budgeting
Omnibus Spending Bill Loaded with Pork. The
Heritage Foundation by Brian M. Riedl When the budget enforcement act of 1990
WebMemo #377 December 2, 2003). expired in 2002, it was not renewed. Congress
Reidl described what he saw as the legions of could have fallen back on the 1974 process, but
interest group lobbyists asking for these funds, with lack of consensus on balance, tax cuts,
and a shift from program funds for general and offsets, it was difficult to put a budget
programmatic purposes, often awarded on the resolution together that both houses and both
basis of technical merit through open competi- parties could agree on. The result can best be
tions, to these earmarked projects, which he described as ad hoc budgeting, unpredictable,
argued went to the highest bidder. To finish up marked by stalemates, and noted for huge
the process, he argued, Congress then had to omnibus appropriations that no one could
appropriate more money for the initial program read in the time allotted. The omnibus appro-
and then in turn earmarked more of that money. priations were laden with pork and special
During the period of the Budget Enforcement interest legislation. War was funded by supple-
Act, while the caps were closely observed, appro- mental appropriations that added to the
priations subcommittees had to make painful deficits, and obscured the fact that the budget
trade-offs for each item of spending. Pork pro- as passed seriously and systematically underes-
jects were obvious and required a publicly timated expenditures and the size of the
acknowledged cut in some other worthy project. deficit. Deficits continued to grow, despite
Because the caps were binding, the amount of promises to curtail them.
pork was low during this period. When pork The quality of budgetary decision-making
spending no longer required an explicit trade- deteriorated in this environment. The admin-
off, the budget process made it easier to include istration supported new budget rules that were
them. Moreover, policy deadlocks, particularly one sided, that required offsets for new expen-
over whether the rules still required offsets to tax ditures or program expansions, but not for tax
reductions, delayed appropriations, resulting in reductions. This proposed rule would make it
omnibus appropriations, easy vehicles to slip easy to reduce taxes almost at will and without
pork into. The omnibus appropriations bills regard for budgetary consequences. This pro-
were often so large and so quickly passed that posed rule was attached to the House version
legislators did not have time to read them, let of the budget resolution for the 2005 fiscal year
alone pull back the pork projects from them. budget. It was not only biased toward one
The successes of the BEA exacerbated the policy, it was unlikely to actually reduce or
demand for pork, but the failures of budget control the burgeoning deficit since both
process after 1998 facilitated this growth in revenues and expenditures need to be con-
spending and moved the budget more in the trolled to reduce or eliminate deficits. The
direction of short-term, personal or partisan Senate refused to go along with the House
benefit and away from collective benefit. rules proposal and hence was unable to pass
The growth in earmarks was noted by many the budget resolution. Consideration of
other observers, and had begun well before budget issues in the Senate went forward with-
9/11. In fiscal year 2000, OMB counted 6,183 out a resolution but few of the appropriations
legislative earmarks. That number represented bills were passed by the end of the fiscal year,
a considerable increase from prior years. necessitating an omnibus appropriation of the
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 147


remaining appropriations bills. The omnibus rules, they would only have to count a few years
legislation was huge, no one was given a chance of the consequences, even though the financial
to read it before passage, violating some basic impact increased after the end of that period.
democratic values and process. Moreover, Then, a couple of years later, the administration
there is little indication that the omnibus legis- argued that the tax cuts should be treated as
lation represented considered evaluation of permanent, so that an extension of the tax cuts
competing projects, programs, or budgetary would not be scored as having any increased
claims. The pork that was stuffed into this costs. The administration sought to dodge the
mammoth appropriation act prompted mock- implications of the tax cuts first one way, and
ing editorials in newspapers. then serpentined back to dodge the implica-
Under the 1974 act, both houses of Congress tions the other way (Budget Rule Change
agree to not only totals in the budget resolution, Would Make The Cost Of Extending The Tax
and revenues necessary to fund those totals, but Cuts Disappear, 2/27/04 Robert Greenstein
also to how those totals will be allocated to dif- and Joel Friedman The Center for Budget and
ferent functions of government. The commit- Policy Priorities, on line at http://www.cbpp.
tees work within these targets, so they will not org/2-27-04bud4.htm).
have to do their work only to find out at the end It was not only the executive branch that
that their decisions and carefully worked out engaged in creative scoring and rules changes
compromises would be undone by further cuts to make deficits look smaller or vanish in order
to come in under limits. By allocating targets to justify the policy of tax cuts without replace-
up front, the process provides some financial ment of the revenues or compensating reduc-
discipline. In addition, Congress collectively tions in spending. House Republicans adopted
prioritizes spending at the macro rather than what has been called dynamic scoring as a way
the micro level. The subcommittees make prior- of measuring the impact of revenue cuts.
itization decisions within their areas, deciding Dynamic scoring tries to take into considera-
between program requests as long as they stay tion the increase in economic activity, and
within their allocations. Prioritization is the hence the increase in federal revenue, that
heart of budgeting; when Congress does not would result from any tax cut. The problem is
come up with a budget resolution or automati- that no one knows how big that effect might
cally accept the president’s proposal, the com- be, and the temptation is to make it as big as
mittees may go over the totals (which may not necessary to wipe out the apparent negative
even be known or agreed to in advance) and effects of a tax cut. CBO, the Congressional
then make across-the-board cuts later, to get Budget Office, has not adopted dynamic scor-
down to the maximum which has by then been ing, and neither did the Senate, meaning that
determined. This process allows for a minimum budget estimates are likely to differ between
of prioritization. This is in fact what happened committees and between houses.
in 2004, for the fy 2005 budget. In fact, Congress
failed to pass a budget resolution in three of the
last five recent years.
The deficits were largely a function (though
not exclusively) of the tax reductions ardently
desired by the president. The president wanted
the tax cuts, but did not want to take responsi- Budget processes do not stand alone, they do
bility for either programmatic reductions or the not reduce spending, or increase deficits, they
deficit. The result was a series of rules that would are a set of rules used by people making
make it look as if the tax reductions cost less than decisions. When the consensus that underlies
they actually do. Budget observers call these the rules evaporates, the rules may remain in
ad hoc accounting rules “gimmicky scoring.” place while implementation erodes. When that
The administration first argued that its tax happens, the effect of the rules is either watered
cuts would be temporary, so that, under the down or eliminated. Gramm Rudman Hollings
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 148


did not eliminate or reduce deficits in the manner gamed; deficit reduction targets were missed.
predicted by the rules, in large part because BEAs caps and offset requirements were imple-
Gramm Rudman Hollings was not enforced. mented and, even though they did not control
Similarly, the Budget Enforcement Act had major the cost increases due to existing legislation,
impact on holding down expenditure growth they provided a kind of discipline and con-
and reducing the amount of pork in the budget, sciousness of the impact of spending decisions.
as long as there was consensus that the process BEA also forced trade-offs, making earmarks
was needed and should be enforced. When that not only visible but politically expensive, as
consensus evaporated in 1998, the rules were no legislators had to say what they would cut (and
longer enforced and spending surged, tax who else’s earmarks they would cut) in order
reductions were passed, and deficits grew at a to fund their constituency-based projects.
very fast pace. When decision-makers agree that When these constraints ended, spending,
they want a particular outcome, particular fea- including for earmarks, surged. BEA required
tures of budget processes can make it easier to explicit offsets for revenue reductions; when
bring about those outcomes. the BEA fell into disuse, the president pro-
This chapter has argued that there are two posed and Congress passed large and repeated
classes of outcomes of the budget process. The tax breaks. House efforts to reform the budget
first is this indirect impact or influence on the process by asking for offsets for increased
size of the deficit, the growth of expenditures spending, but not for tax reductions, suggests
(as a measure of the size of government) and that they felt such rules facilitated their policy
the degree to which spending is constituent goals of reducing tax levels. There would be no
related (to satisfy requests from either individ- procedural obstacles to doing so.
ual or companies in a legislator’s district) or Are there any rival hypotheses that could
for narrow partisan purposes. The second is a explain these results other than the influence
direct effect on the quality of decision-making of budget process? There is little doubt that the
and on democratic decision-making. growth of the economy in the later 1990s helped
The chapter makes the case that deficits and to balance the budget, but the growth of the
spending respond to budget rules, since deficits economy was not sufficient in and of itself to
grew during GRH and, after a couple of years reach balance. In addition, spending rates were
of growth, began to and continued to decline controlled. The two together helped restore
under the Budget Enforcement Act, and then balance. Similarly, the slow down of the econ-
when BEA was no longer enforced and allowed omy in the early 2000s had some effect on reduc-
to lapse, and the features of the 1974 Congres- ing revenue, but tax reductions and spending
sional Budget and Impoundment Control Act increases had a much larger impact. A second
were only used episodically, spending surged possible rival hypothesis is that the amount of
and deficits rebounded. The effectiveness of the pork spending was increasing anyway, possibly
BEA in controlling earmarks may have built up due to a changeover in leadership and determi-
demands to the point that, when the controls nation to build new coalitions of supporters in
were relaxed, earmarking too took a major many tightly contested races. However, the long
upswing. term trend was a reduction in geographically
It is not only the case that these outcomes cor- based spending, and although there have always
related with the periods when various budget been some earmarks in the budget, BEA seemed
process rules were in place, and implemented to hold down the level, which then jumped after
or not implemented, but also that the specific the BEA constraints were over. It may well be
features of the process that were or were not that the reduction in the number of programs
being implemented spoke directly to the that can be used distributively has increased the
results. In other words, there is a mechanism demand for pork, but, even within this model,
linking the budget process with these three BEA spending controls and explicit and public
outcomes. The spending controls in GRH were trade-offs held it down for a while. In the post
not implemented fully and were frequently BEA world, trade-offs, when they occur, are
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 149


invisible; tax reductions occur seemingly without aware that there are minimum requirements for
the necessity of cutting spending programs. a budget process, that collective rules are neces-
The effect of budget process rules on the sary for functioning, that they must be agreed
quality of decision-making is less inferential on by all parties, and be perceived as fair. Both
and there is no need to rule out alternative Democrats and Republicans have become more
hypotheses. The process is the outcome. The ideological in recent years, the differences
deterioration of scoring rules, altering them to between them have become more acute. The
gain short term policy advantages, and then resulting policy quarrels need to be fought on
changing them back again, to get additional a level playing field so the public trusts and
benefits in the short run, the adoption of accepts the outcomes. If the process does not
accounting rules that obscure present and seem fair, legitimate, or neutral with respect to
future costs, the huge and quick omnibus partisan policy outcomes, or if it is not open to
spending bills that bypass the committee scrutiny, public mistrust grows, participation
process and that legislators are unable to read, continues to decline, and democracy itself
the inability to implement the budget resolu- erodes. In this sense, budget process not only
tions that force some prioritization, all suggest a matters, it may be fundamental to governing in
deterioration of the quality of decision-making. the United States. That having been said, how-
Perhaps most seriously, the budget process ever, rules are but one part of the equation of
has become less democratic. The processes have budget control in most industrialized democra-
become ad hoc, to produce a budget in the cies, and a range of factors must be considered
face of policy stalemates. Rules are made and for a more complete understanding.
remade, often in the service of particular imme-
diate policy goals, not known in advance, not
neutral, and not agreed to by all parties. They do
not allow for a forum in which disagreements REFERENCES
can be discussed and worked out, exaggerating
the tendency toward stalemate, contributing to Albert-Roulhac, C. (1998) The Influence of EU
the lack of budget resolutions. The administra- membership on Methods and Processes of
tion is not content merely to lower taxes, but Budgeting in Britain and France, Governance 11,
also seeks to lock in those reductions for years to 209–30.
come, and has increased borrowing to an extent Gray, A., W. I. Jenkins and B. Segsworth (2002)
Budgeting, Auditing and Evaluation (New
that will burden the next generation, and reduce
Brunswick, Transaction).
its ability to prioritize. HM Treasury (2005) Public Expenditure Planning
With the end of divided government as a and Control in the United Kingdom (London: HM
result of the elections in 2004, policy stalemates Treasury).
should be reduced. It should be easier to pass Meyers, R. (1992) Federal Budgeting and Finance;
budget resolutions and hence to prioritize the The Future is Now, Public Budgeting and Finance,
budget because the administration will control 4, 2–14.
a majority of both Houses of Congress. It should Palazzolo, D. (1999) Done Deal?: The Politics of the
be easier to pass appropriations bills, and hence 1997 Budget Agreement (New York: Chatham
the need for huge, last minute omnibus appro- House).
priations should be reduced. The quality of Patashnik, E. M. (2000) Putting Trust in the US Budget
(Cambridge: Cambridge University Press).
decision-making should therefore improve and
Stolfi, F. (2006) Reforming the Italian State:
the process should become more predictable Administrative Modernization and Fiscal
and open. But when the same political party Management (Unpublished Ph.D. Dissertation,
controls the White House and both Houses of Department of Political Science, University of
Congress, there is an increased chance of pass- Pittsburgh).
ing budget rules with marked policy biases. As Wanna, J., L. Jensen and J. de Vries (2003) Controlling
yet, there is no evidence that political actors are Public Expenditure (Cheltenham: Edward Elgar).
Peter-3383-Chapter-07.qxd 6/8/2006 7:34 PM Page 150
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 151



Although the field of implementation research is stage for later implementation research. Most
barely 30 years old, implementation has already implementation research has focused on
been analyzed from many different perspectives implementation problems, barriers, and fail-
representing different research strategies, evalua- ures, and this pessimistic view of implementa-
tion standards, concepts, focal subject areas, tion was already reflected in the subtitle of this
and methodologies. This chapter first performs seminal work, ‘How great expectations in
a critical review of some of the major contribu- Washington are dashed in Oakland; or, Why it’s
tions to the literature. This examination will amazing that federal programs work at all…’.
follow the development of the field over three This case study focused on the local imple-
generations of research (Goggin, 1986). mentation of a federal economic development
Second, based on a critical examination of the program to increase employment among
development and status of the research field, the ethnic minority groups in Oakland in
chapter will suggest promising ways of moving California. Its guiding research questions were:
ahead. It claims that implementation research ‘How well was this authoritative mandate (law,
can be improved by accepting theoretical diver- regulation, program, official pronouncement)
sity and partial theories and hypotheses, rather implemented?’ and ‘How might it have been
than looking for one common and general better implemented?’ Later research redefined
theoretical framework, seeking conceptual clari- the question to focus on achieving the explicit
fication, including focusing on output (perfor- or implicit values in a given mandate rather
mance of implementers) as well as outcomes as than its prescriptive details (Bardach, 2001).
dependent variables in implementation research, Accordingly, goal achievement has been the
and applying more comparative and statistical dominating standard and dependent variable
research designs, rather than relying on single for implementation research since the 1970s.
case-studies in order to sort out the influence of Pressman and Wildavsky focused on the
different implementation variables. ‘complexity of joint action’ as the key imple-
mentation problem. In their Oakland economic
development case – as in many others – federal,
regional, state, and local government actors,
courts, affected interest groups, private firms,
In several respects the book, Implementation, and media had a role and stake in policy
by Pressman and Wildavsky (1973) sets the implementation. Implementation problems
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 152


were amplified not only by the many actors but SECOND GENERATION MODEL
also by the many decision and veto points, BUILDERS: TOP-DOWN, BOTTOM-UP
which must typically be passed during the AND SYNTHESES
implementation process. Although they proba-
bly over-emphasized the lack of conflict in Second generation implementation studies began
their case, Pressman and Wildavsky convinc- in the early 1980s. While the first generation
ingly showed that merely slightly different per- studies had been explorative and theory generat-
spectives, priorities, and time horizons among ing, the ambition of the second generation was to
multiple actors with different missions in take a next step in theory development by con-
repeated and sequential decisions could cause structing theoretical models, or rather frame-
delay, distortion, and even failures in policy works of analysis, which could guide empirical
implementation. analysis. Some of these studies had more opti-
However, the two authors also demon- mistic views on successful implementation.
strated that failures are not only caused by bad The construction of models and research strate-
implementation but also by bad policy instru- gies, however, immediately led to a major con-
ments. Many of the problems in the Oakland frontation between the so-called top-down and
case would have been avoided had policy makers bottom-up perspectives on policy implementation.
chosen a more direct economic instrument The predominant top-down researchers
that would ex post have tied spending of public focused on a specific political decision, nor-
expenditures to the actual number of minority mally a law. On the background of its official
workers employed, rather than relying on end- purpose they followed the implementation
less ex ante negotiations with affected parties down through the system, often with special
and authorities. interest in upper level decision-makers. They
Pressman and Wildavsky are good represen- would typically assume a control perspective on
tatives for the first generation of implemen- implementation, trying to give good advice on
tation studies, which were typically explorative how to structure the implementation process
and inductive case studies with a theory- from above in order to achieve the purpose of
generating aim. Very few central theoretical the legislation and to minimize the number of
variables were in focus, in this case the number decision points that could be vetoed.
of actors and decision points and the validity The best-known and most frequently used
of the causal theory. Another outstanding (Sabatier 1986) top-down analysis framework
example is Eugene Bardach’s (1977) The was developed by Mazmanian and Sabatier
Implementation Game, which placed more (1981). It contains 17 variables placed in 3 main
emphasis on the aspects of conflict in imple- groups concerning the tractability of the prob-
mentation, seeing implementation as a contin- lems addressed by the legislation, the social
uation of the political game from the policy and political context, and the ability of the leg-
adoption stage, although partly with other islation to structure the implementation
actors and other relations among actors. process. This structuring can be made by
Bardach analyzed the types of games that means of, for example, hierarchy, appointing of
various actors apply in the implementation authorities and staff with a positive attitude
process in order to pursue their own interests. towards the legislation/program, and use of
However, these games tend to distort imple- incentives including competition among
mentation from the legislative goals. Among providers. By adding a long-term perspective
other representatives from what has later been of 10–15 years to implementation, the authors
called the first generation of implementation show that, over time, start-up problems are
research we find Erwin Hargrove (1975), who often ameliorated by better structuring of the
called implementation research ‘the missing implementation by policy advocates (see also
link’ in the study of the policy process, and Kirst and Jung, 1982). This gave rise to much
Walter Williams and Richard Elmore (1976). more optimistic views of implementation in
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 153


contrast to the pessimism introduced by public policies, their similar working conditions
Pressman and Wildavsky (1973) and joined by make them all apply similar behavior. This
most implementation analysts. means that street-level bureaucrats even
Mazmanian and Sabatier’s framework was met across policy-types, tend to apply similar types
by two different kinds of criticism. According of practices, whether they are teachers, police-
to one strand, the model was naive and unreal- men, nurses, doctors, or social workers. It also
istic because it overemphasized the ability of means that their individual attitudes are not
policy proponents to structure implementa- expected to have important implications for
tion, thus ignoring the ability of policy oppo- their behaviors.
nents to interfere in this structuring process Although trying to do their best, street-level
(Moe, 1989). Often policy opponents are able bureaucrats experience a gap between the
to make policy goals less clear and to increase demands made on them by legislative man-
their own long-term influence in the imple- dates, managers and citizens on one side and
mentation process in order to avoid some of their high workload on the other. In this situa-
the effects intended by policy proponents. tion they all apply a number of coping mecha-
Conceptually, the model ignored the politics of nisms that systematically distort their work in
policy-formulation and policy design (Winter, relation to the intentions of the legislation. They
1986b; May, 2003). ration services, make priorities between their
Another strand of criticism came from the tasks, apply simple standardized processing of
bottom-up researchers who took special interest clients, which tends to be in the favor of more
in ‘the bottom’ of the implementation system, easy cases and resourceful clients (creaming),
the place where public policies are delivered to they seek to control clients, and over time street-
citizens or firms. They all emphasized the influ- level bureaucrats develop more cynical percep-
ence that front-line staff have on the delivery of tions of clients and modify the policy objectives
policies such as social services, income transfers that are the basis of their work. According to
and law enforcement in relation to citizens and Lipsky, increasing staff resources is no cure
firms. Front-line workers are crucial decision- to coping as more resources will merely lead to
makers in these studies that emphasize the more demand for services.
weak control that politicians and administrative Other bottom-up researchers go the whole
managers have of control front-line staffs. length, rejecting the objective of policy man-
Like top-down researchers (and also most eval- dates as an evaluation standard. Instead their
uation researchers), some bottom-up researchers analysis departs from a specific problem
use the official objectives of a given legislation as (Elmore, 1982) such as youth unemployment
the standard of evaluation (Lipsky, 1980; Winter, (Elmore, 1982) or small firms’ conditions of
1986a). Michael Lipsky (1980) developed a growth (Hull and Hjern, 1987). In practice it is
theory on ‘Street-level Bureaucracy.’ It focuses the researcher himself who in most cases
on the discretionary decisions that each front- defines the problem and thereby his evaluation
line worker – or ‘street-level bureaucrat’ as standard. This is acceptable if done explicitly,
Lipsky prefers to call them – make when deliv- and it can be fruitful if the researcher is able to
ering policies to individual citizens. This discre- convince his readers about the appropriateness
tionary role in delivering services or enforcing of his problem definition.
regulations makes street-level bureaucrats The next task in Hull and Hjern’s (1987) bottom-
essential actors in implementing public poli- up approach is to identify the many actors that
cies. Indeed, Lipsky (1980) turns the policy are affecting the problem in question and to
process upside-down by claiming that street- map relations between them. In these network
level bureaucrats are the real policy makers. analyses – using a ‘snowball method’ – both
However, one ironic aspect of the theory is that, public and private actors become essential,
although Lipsky emphasizes the individual role and the analyses often include several policies
of street-level bureaucrats in implementing that affect the same problem. In this way,
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 154


the analysis maps the informal, empirical one approach might be more relevant than the
implementation structure around a given other. Sabatier (1986) claims that the top-down
problem, while top-down research tends to look perspective is best suited for studying the
at the formal implementation structure related implementation in policy areas that are domi-
to one particular policy program. According to nated by one specific legislation, limited
Hull and Hjern, empirical implementation research funds, or where the situation is struc-
structures tend to be far less hierarchical than tured at least moderately well. Bottom-up per-
formal ones, and they often cross organizational spectives, on the other hand, would be more
borders in forming collaborative networks at relevant in situations where several different
the operational level. The bottom-up analyses policies are directed towards a particular
by Hjern and associates are important in draw- problem, and where one is primarily interested
ing attention to implementation activities and in the dynamics of different local situations.
structures at the local operational level, but the Attempts were also made to synthesize the
perspective has more the character of guidelines two models. Richard E. Matland (1995) suggests
for an inductive research strategy and method- that their relative value depends on the degree of
ology than a development of theory and ambiguity in goals and means of a policy and the
hypotheses that can be empirically tested. A degree of conflict. Traditional top-down models,
recent example of the bottom-up approach is based on the public administration tradition,
Bogason’s (2000) study of local governance. It is present an accurate description of the imple-
inspired by Hull and Hjern but adds elements mentation process when a policy is clear and the
of institutional and constructivist analyses and conflict is low. However, newer top-down
points to the fragmented character of the mod- models, such as the Mazmanian-Sabatier frame-
ern state in policy making and implementation. work, are also relevant when conflict is high and
ambiguity is low, which makes the structuring of
the implementation particularly important. In
Suggested Syntheses contrast, bottom-up models provide an accurate
description of the implementation process when
The top-down and bottom-up perspectives were the policy is ambiguous and the conflict is low.
useful in drawing attention to the fact that When conflict as well as ambiguity is present,
both top and bottom play important roles in both models have some relevance according to
the implementation process, but in the long Matland.
run the battle between the two approaches was Other attempts at synthesizing the two
not fruitful. Each tended to ignore the portion approaches were made by the former main
of the implementation reality explained by the combatants. The bottom-uppers, Hull and Hjern
other (Goggin et al., 1990: 12; Hill and Hupe, (1987), proposed a synthesis – called ‘an induc-
2002). Elmore (1985) actually recommends tive approach to match outcomes of politics and
using both forward mapping – which is essen- their intentions,’ which calls for systematic
tially a top-down analysis – and backward interview analysis of relevant actors from the
mapping for policy analysis because each tends bottom to the very top. The approach would
to offer valuable insights for policy makers. He require immense research resources, and I am
claims that policy designers need to consider not aware of any such study performed in prac-
the policy instruments and the resources they tice. In addition, their proposed synthesis suffers
have at their disposal (forward mapping), as from being methodological recommendations
well as the incentive structure of the target rather than theoretically based expectations,
group and street-level bureaucrats’ ability to which can be tested systematically.
tip the balance of these incentives in order to Also Sabatier (1986) has suggested a
affect the problematic situation of the target synthesis – the so-called Advocacy Coalition
group (backward mapping). Framework (ACF). However, although making
Other scholars have tried to solve the con- an important contribution to the public policy
troversy by specifying the conditions where literature, Sabatier and his later associate,
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 155


Jenkins-Smith (Sabatier and Jenkins-Smith, A policy design typically contains a set of

1993), actually moved the focus of analysis goals, a mix of instruments for obtaining these
towards policy change and formation and away goals, a designation of governmental or non-
from implementation. governmental entities charged with carrying out
Another kind of synthesis was suggested by the goals, and an allocation of resources for the
Winter (1990; 1994; 2003a) in his ‘Integrated requisite tasks (May, 2003). Policy design and
Implementation Model.’ Unlike previous policy instruments have received substantial
attempts, the purpose here is not to make a research interest since the 1980s (Linder and
true synthesis between top-down and bottom- Peters, 1989; Salamon, 2002). The basic claim of
up perspectives, but rather to integrate a num- this literature is that any policy can be disaggre-
ber of the most fruitful theoretical elements gated to one or a mix of a limited number of
from various pieces of implementation generic policy instruments. The research inter-
research – regardless of their origin – into a est, however, has not led to agreement on any
joint model. As dependent variable and stan- typology of instruments (Vedung, 1995). One
dard for evaluating the results of the imple- simple classification consists of mandates, eco-
mentation process the model focuses on nomic incentives, and information, which aim
performance as well as outcome in relation to at affecting the behavior of either target groups
the official policy objectives. This standard was or intermediaries (implementers).
selected from a democratic point of view, Policy design affects the implementation
because goals formulated in parliament and in process and results in various ways. Different
laws have a particular legitimate status and are mixes of instruments are not equally effective
relevant for holding government accountable. in obtaining a given policy objective. Policy
The first set of factors, which affects imple- design is important in affecting the incentives
mentation results, is the policy formulation of intermediaries to carry out their requisite
process and the policy design. Too many imple- tasks, particularly through affecting their com-
mentation researchers have erroneously put the mitment and capacity and by signaling desired
whole blame for any lack of goal-achievement actions (May, 2003). While the validity of the
on implementation. As noted by Peter May causal theory linking instruments to objectives
(2003) well designed policies are necessary but certainly is important, however, the research
not sufficient for improving implementation documentation of instrument effects is still
prospects. Other implementation scholars meager. One reason is that effects of instru-
have ignored or failed to conceptualize the ments on implementation are often deter-
connections between policy formulation, mined by the context, including the political
policy design and implementation. context. Consequently, designing good policies
The roots of implementation problems can is not a simple, technocratic process like select-
often be found in the prior policy formulation ing the best types of materials for building a
process. For instance, conflicts in that process bridge (May, 2003).
often create a policy design that is marked by In addition, the chosen instruments may
ambiguous goals as well as an invalid causal affect the overall implementation structure
theory with a lack of connection between goals and process, as certain instruments tend to
and means. Sometimes even symbolic policies favor the formation of particular implementa-
are adopted to (appear to) address a problem tion structures. Mandates aimed at regulating
without actually offering the means that could the behavior of target groups normally require
achieve the stated objectives. And, as men- a staff for inspecting and enforcing the man-
tioned by Bardach (1977), the conflicts in date and a set of sanctions. Information strate-
policy formulation often continue in the sub- gies and use of economic incentives such
sequent implementation process. Not only as environmental taxes can sometimes be
conflict but also lack of attention among the implemented with fewer staff, although there
coalition partners passing a law can lead to is no one-to-one relationship between instru-
implementation failures (Winter, 1986b). ments and staff requirements. Some taxes are
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 156


relatively automatic and easy to collect, such as can produce and deliver implementation
an environmental tax per liter gasoline sold, outputs independently of each other, can pro-
while others require a substantial staff for vide relatively good implementation results,
inspection and enforcing, for example, taxing although the coordination may not be optimal.
diffuse pollution. O’Toole (2003) and May (2003) show how inter-
It is important to understand that ineffec- organizational coordination problems can be
tive policy designs are not always due to lack of reduced by using policy design to increase com-
knowledge on the part of the policy designers. mitment, build and use a common interest, and
Policy design of instruments and organiza- facilitate cooperation via exchange.
tional structure is first of all a political process, The behaviors of street-level bureaucrats are
in which political actors – both policy propo- also crucial for the implementation of most
nents and opponents – try to maximize their policies, and Lipsky’s (1980) insights above on
interests, including selecting an organizational ‘street-level bureaucracy’ are included in the
structure, which will allow themselves to max- Integrated Implementation Model. Street-level
imize long-term control of the implementa- bureaucrats are making important discre-
tion process (Moe, 1989). tionary decisions in their direct contact with
The next set of factors of the model focuses citizens and firms. Because such bureaucrats
on how the implementation process affects the work in situations characterized by many
results. Implementation processes are charac- demands and limited resources, they respond by
terized by organizational and interorganizational resorting to coping behaviors. These short-cuts
behaviors representing different degrees of com- systematically bias the delivery behavior in
mitment and coordination. Interorganizational relation to the policy mandates. While Lipsky’s
implementation settings seem to become ever contribution was important for understanding
more important (O’Toole, 2003). As mentioned implementation, the theory needs more specifi-
above, Pressman and Wildavsky (1973) focused cations of the causal mechanisms that can
on the ‘complexity of joint action,’ according to explain variation in coping-behaviors and their
which successful implementation is likely to be consequences (Winter, 2002; Heinesen et al.,
negatively related to the number of actors and 2004), cf. below. For a review of the literature on
decision and veto points. street-level bureaucrats, see Meyers and
However, as shown by O’Toole and Montjoy Vorsanger, 2003).
(1984; O’Toole, 2003) this insight only applies to According to the Integrated Implementation
certain kinds of interorganizational implementa- Model also target groups of public policies,
tion settings. Decision points are not indepen- that is, citizens or firms, play important roles,
dent of each other, but successful implementation not only on the effects of the policy, but also
results can be stimulated by an early agreement in affecting the performance by street-level
on basic understandings, which can promote bureaucrats through positive or negative
‘bandwagon effects’ in later decisions, and deci- actions in co-producing public services and
sions can be merged by crafting ‘package deals.’ regulation (Hill and Hupe, 2002: 134–36).
The implementation prospects also depend Finally, socio-economic contexts form impor-
on the type of resource-dependency among tant framework conditions for implementa-
participating organizations. The ‘complexity of tion. For example, in employment policies
joint action’ best applies to a chain of sequen- delivery behavior (the types of employment
tial relations, in which one organization offers) and effects depend heavily on ups and
depends on outputs from another as input for downs in the business cycle.
its own contribution to implementation. The Integrated Implementation Model is
However, reciprocal relations, in which two obviously not a model in the strict sense of a
organizations depend on each other for inputs simple causal model. It is rather a framework
can decrease the likelihood of veto points of analysis presenting key factors and mecha-
because both have incentives to cooperate. nisms that affect implementation outputs
Pooled relations, where multiple organizations and outcomes. For each set of factors a number
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 157


of more specific hypotheses can be developed Later, Lester and Goggin (1998), in making a
(Winter, 1990, 1994, 2003a; Jensen et al., 1991; status for implementation research, called for
May, 2003; O’Toole, 2003; Meyers and the development of ‘a parsimonious, yet com-
Vorsanger, 2003). plete, theory of policy implementation.’ They
suggested that such meta-theory might be devel-
oped by combining the insights of communi-
cations theory, regime theory, rational choice
theory (especially game theory), and contin-
gency theories. As dependent variable for imple-
mentation studies they proposed to focus on
While the first and second generations of implementation processes rather than outputs
implementation studies have been helpful in and outcomes.
directing attention to implementation prob-
lems and identifying implementation barriers
and factors that might ease implementation,
the research had not succeeded in sorting out
the relative importance of the explanatory
variables. A substantial part of the studies
could be criticized as merely presenting – often While agreeing with Goggin’s (1986) call for
long – checklists of variables that might effect more comparative and statistical research
implementation. designs based on quantitative methods, I dis-
Malcolm Goggin (1986) pointed out that, agree with several of the later methodological
because implementation research had been and theoretical recommendations made by
dominated by single case studies, it was plagued him and his colleagues. As recognized by
by the problem of ‘too few cases and too many one of these authors, O’Toole (2000), to follow
variables’ or by ‘overdetermination,’ where two the methodological suggestions given by
or more variables explain variation in the Goggin, Bowman, Lester, and O´Toole (1990)
dependent variable equally well. The fact that would involve at least outlining a research
the single case-study approach does not allow career’s worth of work. This work would
for any control of third variables had hampered require applying research designs that involve
the development of implementation theory, numerous variables, across different policy
according to Goggin. He therefore called for types, across 50 states, over at least 10 years, as
a third generation of implementation studies well as measuring the relevant variables by a
that would test theories on the basis of more combination of content analyses, expert pan-
comparative case-studies and statistical research els, elite surveys, and expert reassessment of
designs, which could increase the number of the data from questionnaires and interviews.
observations. As such a research strategy is too demanding;
Goggin followed up on these recommenda- less demanding research strategies, which can
tions with his associates (Goggin, Bowman, still secure a sufficient number of observa-
Lester, and O´Toole, 1990) in a study, which was tions, would be more realistic.
based mainly on a communications theory per- My suggestions for further development
spective on intergovernmental implementation, of implementation research can be summarized
but also included many variables from previous in six points: (1) providing theoretical diversity,
top-down and bottom-up research. The study (2) focusing on partial rather than general
focused especially on variation among states in implementation theories, (3) seeking conceptual
the way and extent they implement federal poli- clarification, (4) focusing on the implementa-
cies in three different social and regulatory poli- tion output (performance of implementers) as
cies. The authors tried to encourage further a dependent variable, (5) including studies of
research involving multiple measures and mul- outcomes, and (6) using more comparative and
tiple methods, including quantitative methods. statistical research designs (Winter, 1999).
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 158


While the last point has been developed above, I Lynn, Heinrich and Hill (2001; Heinrich and
will elaborate on the other ones in the following Lynn, 2000). They have proposed a ‘Logic of
and illustrate them by some of my recent Governance.’ They define governance as ‘regimes
research on implementation of Danish agro- of laws, rules, judicical decisions, and adminis-
environmental regulation with Peter May and of trative practices that constrain, prescribe, and
Danish integration policy towards refugees and enable the provision of publicly supported
immigrants. goods and services.’ The logic can be expressed
as a basic ‘reduced form’ according to which
outputs/outcomes are a function of environ-
Theoretical Diversity and mental factors, client characteristics, treatments,
Partial Theories structures and managerial roles and actions.
A more complex form allows for interrelations
Given the many exploratory variables, which between the explanatory factors. This frame-
have already been identified by various imple- work broadens the scope of traditional
mentation scholars, the suggestion by Goggin implementation research. It focuses on the inter-
et al. of developing a ‘parsimonious, yet complete relations and complexities of administrative
implementation theory’ by combining theoreti- phenomena, and calls for sound empirical
cal elements from at least four different theories, yet theory-driven research. While its main theo-
appear to be a contradictio in adjecto and is more retical inspiration is political economy, other
likely to lead to theoretical mismatch. Rather theoretical perspectives are welcomed.
than looking for the overall and one for all The main contribution of the framework is
implementation theory, as has been the utopian its emphasis on systematic empirical testing of
objective for many implementation scholars, we theory driven hypotheses and for generating a
should welcome diversity in both the theoretical framework that allows many studies to talk to
perspectives and methodologies applied. Such each other. The authors claim that both top-
diversity will give us new insights. It also down and bottom-up considerations are
strikes me as unrealistic – and probably not very included, but the top-down control orienta-
fruitful – that many scholars could agree on tion seems to be stronger than bottom-up per-
applying one common theoretical framework. spectives. The motivations of street-level
Although the general implementation frame- bureaucrats have not been conceptualized as
works presented by model builders so far have part of the framework, which focuses more on
been helpful in giving an overview of some cru- managerial behaviors.
cial implementation variables, the generality of A related framework and research program
such models may in fact be an obstacle for on management and performance has been
further development of our understanding established by Meier and O’Toole (2004; 2006).
of implementation. This is due to the fact that
generality inhibits precise specification of vari-
ables and causal mechanisms (May, 1999). Need for Conceptual Clarification and
Consequently, it seems more fruitful to use Focusing on Implementation Outputs as
research resources on developing partial a Dependent Variable
theories and hypotheses about different and
more limited implementation problems and on As pointed out by Peter May (1999) most con-
putting these to serious empirical tests. ceptual frameworks in the implementation
Some of the different implementation per- literature are weakly developed, lacking ade-
spectives may be integrated into broader analyt- quate definitions of concepts and specification
ical frameworks or models (Mazmanian and of causal mechanisms. The most important
Sabatier, 1981; Winter, 1990; Goggin, Bowman, issue for the development of implementation
Lester and O’Toole, 1990). A new promising research may be to reconsider what constitutes
research collaboration around a common the object of the study. There has been some
analytical framework has been initiated by disagreement in the literature on the term of
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 159


‘implementation’ and on what is the important goal-achievement based on analysis of the

dependent variable in implementation research. behavior or performance of implementers is
One problem is that the concept ‘implemen- dependent on the goal variable having a certain
tation’ is often used to characterize both the value. The generalization may become invalid
implementation process and the output – and if the goal changes. Therefore, generalizations
sometimes also the outcome – of the imple- about implementation output are extremely
mentation process. Lester and Goggin (1998) relativistic because statements are conditioned
view implementation as a ‘process, a series of by the goals that are formulated. This is prob-
subnational decisions and actions directed lematic when it is recognized that policy makers
toward putting a prior authoritative federal are often more interested in making decisions
decision into effect.’ Thereby, they reject focus- on means or instruments than goals; goals are
ing on the output of the implementation often invented after decisions on the means
process as ‘a dichotomous conceptualization of have been made in order to legitimize the means
implementation as simply success or failure.’ adopted, and goals are not always expected, or
Although agreeing that the success/failure even intended, to be achieved.
dichotomy is problematic, I suggest that the The second problem of using goal-achievement
most important focus of implementation as the dependent variable of implementation
research would not be the implementation research is that such goals can be difficult
process but the output of that process in terms to operationalize. Much has already been writ-
of delivery behavior. This would be much more ten in the implementation and evaluation
in line with the classic focus of public policy literatures about the vagueness and ambiguity
research on the content of policy, its causes, and of policy goals and the difference between
consequences (Dye, 1976). Implementation out- official and latent goals. In addition, while
put is policy content at a much more operational most policy statutes state some kind of goal
level than a law. It is policy as it is being delivered for the outcome of the policy, many fail to
to the citizens. However, we should conceptual- specify goals or standards for the behavior of
ize output in other ways than the common the implementers. This is often the case in
success/failure dichotomy or interval. regulatory policies that tend to specify the
The most common dependent variable in behaviors of regulatees rather than those of the
implementation research so far has been the regulators.
degree of goal-achievement, whether defined Because of the problems of using goal-
in terms of output or outcome. The first achievement as a dependent variable, I suggest
problem, however, is that goal-achievement is that we look for behavioral output variables to
a fraction. Output in terms of performance characterize the performance of implementers
of the implementers or outcome in terms of in delivering services or transfer payments to
effects on target population is the numerator, the citizens or enforcing regulations. The first
and the policy goal is the denominator. Yet, aim of implementation research then should
using a fraction as the dependent variable ren- be to explain variation in such performance.
ders theory building problematic when differ- This will require substantial effort in concep-
ent factors explain variation in the numerator tualizing and categorizing the performance of
and the denominator. While the policy forma- implementers at the levels of agency as well as
tion process is likely to account for variation that of the individual street-level bureaucrat.
in goals, the implementation process is likely However, as specified below, focusing on
to account for variation in performance, and implementation output does not mean that
additional factors are likely to account for vari- outcomes are unimportant.
ation in outcomes. This renders the construc- One very intriguing question is whether
tion and accumulation of implementation we can find behavioral output dimensions and
theory very complex. classifications that are universally applicable
Pushing it to extremes, the problem is that in all policy areas, or if we should generate
any attempt to make generalizations about concepts and classifications that are different
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 160


from one policy area to another. In order to 2004) However, in addition, the dimension of
stimulate theory building we should avoid professional distance (versus close personal
concepts that are very policy specific, because involvement with clients) is included, which
generalizations based on these would have a seems to be particularly relevant in social
rather narrow sphere of application. policy implementation.
At the other extreme, Lipsky’s (1980) street- One advantage of creating such conceptual-
level bureaucracy theory represents an ambi- ization of the behavior of implementers is that
tious attempt to offer a universally applicable it is well suited for testing hypotheses for
set of concepts for describing the coping explaining variation in implementation behav-
behavior of street-level bureaucrats in all ior across time and space. Variables from
policy areas. Although Lipsky’s coping behav- implementation theory characterizing aspects
iors might seem to have a better fit for social of the implementation process would be an
policies with weak clients, it has been demon- important basis for the development and test
strated that they are also relevant for a regula- of such hypotheses. However, another advan-
tory policy with strong clients in a study of tage of focusing on outputs as a dependent
Danish agro-environmental and integration variable in implementation research is that we
policies (Winter, 2002). However, coping can integrate the study of implementation
focuses on dysfunctional behaviors, and we much more with theory on bureaucratic poli-
need more concepts for characterizing agency tics and organization theory. Implementation
and front-line staff behaviors adequately. research can thereby gain inspiration from
Other concepts have been developed for these research fields that have a long tradition
classifying the behavior of implementers in of studying the behavior of agencies and
almost any kind of social regulation policy bureaucrats (see also Lynn, Heinrich and Hill,
(Kagan, 1994). May and Winter (1999; 2000; 2001). In return, these sub-disciplines can
Winter and May, 2001) have developed con- benefit from implementation concepts that are
cepts for regulatory enforcement at both much more policy relevant than those behav-
agency and individual street-level bureaucrat ioral variables applied in most bureaucracy
levels. Agency enforcement choices are concep- and organization theory.
tualized as (1) tools (use of different enforce- As an example, Winter (2003b) analyzes the
ment measures: sanctions, information, and discretion of street-level bureaucrats in imple-
incentives), (2) priorities (whom to target and menting agro-environmental regulation and
what to inspect for), and (3) effort (use integration policy in Denmark by applying a
and leveraging of enforcement resources). The modified principal-agent perspective and its
enforcement style of individual inspectors is notion of information asymmetry in examin-
defined as the character of the day-to- ing the extent to which local politicians control
day interactions of inspectors with the target their street-level bureaucrats (Moe, 1984;
group. May and Winter expect and verify, in Brehm and Gates, 1999). Regression analyses
a study of agro-environmental regulation in of 216 local inspectors and 388 social case-
Denmark, that enforcement style has two workers show that local politicians’ policy pref-
dimensions comprised as the degree of formal- erences have very little direct impact on the
ity of interactions and the use of threats and behaviors of street-level bureaucrats. However,
other forms of coercion (May and Winter, to some extent the politicians do control rela-
2000; see also May and Burby, 1998; May and tively visible kinds of performance, such as the
Wood, 2003). number of inspections and the number of
In a study of the implementation of Danish timely processed cases, through funding capa-
integration policy for refugees and immigrants city for implementation. On the other hand,
the same two implementation style dimen- when it comes to less transparent front-line
sions have been found relevant for examining behaviors – such as the implementation styles
a social policy (Winter, 2003b; Heinesen et al., and the strictness front-line staffs apply in
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 161


reacting to violations of the rules – politicians’ to make a distinction between explaining

policy preferences and their funding of staff implementation outputs and outcomes.
resources have little or no influence on these We do not have a complete understanding of
practices. On the contrary, the latter are domi- the policy process unless we know how target
nated by the street-level bureaucrats’ own groups respond to public policies. Despite the
values. fact that ‘the authoritative allocation of values
The study also examines the impact of for a society’ (Easton, 1953) and ‘who gets what
various types of attitudes on street-level when and how’ (Lasswell, 1936) are among the
bureaucrats’ behavior. While their abstract most famous definitions of politics, very few
and general support for the focal policy does political science studies focus on how citizens
not have much effect, their attitudes on the respond to public policies. Some would say that
policy instruments, target population, and this is the province of evaluation research.
their workload have strong impacts. In a However, evaluation is characterized by a focus
study of street-level bureaucratic coping in on methods, whereas very little theory develop-
the implementation of the same two Danish ment has occurred, especially extremely little
policies, the same types of street-level political science theory. Some law and society
bureaucrat attitudes have been found to be scholars have attempted to explain variation
very important in explaining variation in in compliance among citizens and, to lesser
coping. However, coping is also strongly degree, firms. So far, very few political scientists
affected by the number of staff that politi- and public policy researchers have tried to theo-
cians allocate for implementation (Winter, rize and test hypotheses about variation in out-
2002; Heinesen et al., 2004). Thus, the appli- come and how implementation behavior affects
cation of quantitative analysis have con- outcomes. In political science journals the con-
firmed that Lipsky’s coping concepts are very trast between many studies of citizens’ attitudes
useful but also demonstrated that individual and behavior at the input side of politics and
attitudes of street-level bureaucrats as well as very few outcome studies is striking. Yet, the
the level of resources allocated for their work study of outcomes is as much, if not more, about
are much more important for explaining the policy than are most public opinion studies that
use of coping than Lipsky expected. relate to the input side of policy.
My suggested redefinition of the dependent
variable of implementation research from
Outcome Studies goal-achievement to a behavioral performance
variable has not only the advantage of making
My suggestion of using implementation output/ it easier to explain variation in implementa-
performance as dependent variable in imple- tion outputs and easier to make generaliza-
mentation research does not imply that out- tions. The conceptualization of performance is
come/impacts are unimportant in public policy also likely to make it much easier to study the
analyses. On the contrary, implementation relation between implementation outputs and
scholars, as well as other political scientists, have outcomes (May and Winter, 1999; Winter and
paid far too little attention to explaining policy May, 2001; 2002). In such studies delivery level
outcomes and to examining the relation performance/output changes from being a
between implementation outputs and outcomes dependent variable to become an independent
(Lynn, Heinrich and Hill, 2001). As mentioned variable in outcome studies. Most likely, we
above, some implementation scholars do need different theorizing for explaining imple-
include outcome in their implementation mentation outputs and outcomes.
models or framework (Hull and Hjern, 1987; As claimed by Elmore (1982; 1985), to change
Elmore, 1982; Mazmanian and Sabatier, 1981; target groups’ problematic behavior requires an
Winter, 1990; Goggin, Bowman, Lester and understanding of the incentives that are operat-
O’Toole, 1990). It might, however, be fruitful ing on these people as well as of how street-level
Peter-3383-Chapter-08.qxd 6/8/2006 7:51 PM Page 162


bureaucrats can influence and build on these has found that street-level bureaucrats’ use of
incentives. For example, in examining Danish coping as well as various implementation
farmers’ compliance with environmental regu- styles affect the outcome in terms of labor
lations, Winter and May (2001) map the regula- market integration of refugees and immi-
tees’ action model. In multiple regression grants. However, the effects depend on the
analyses of survey data of 1,562 farmers, they context (Lynn, Heinrich and Hill, 2001; Hill
show that compliance is affected by farmers’ (1) and Hupe, 2002). In municipalities with a dif-
calculated motivations based on the costs of ficult integration task – due to the composition
complying and the perceived risk of detection of of immigrants and labor market conditions –
violations (while the risk of sanctions, as in coping and professional distance increases
most other studies, had no deterrent effect), (2) employment, while in municipalities with an
normative sense of duty to comply and (3) easy integration task formalism decreases
social motivations based on adaptation to employment (Heinesen et al., 2004). As coping
expectations from significant others. Inspectors is normally assumed to bias the implementa-
signal such expectations through their style of tion in a dysfunctional way, it is remarkable
interacting with target groups. Inspectors’ for- that it here has positive outcomes, at least in
malism increases compliance up to a point by the short run, which may be due to creaming
providing greater certainty of what is expected of the most resourceful immigrants. The find-
of regulatees, while coercive styles with threats ings from the two studies illustrate that deliv-
of sanctions backfire for regulatees who are not ery performance variables can be constructed
aware of the rules. Willingness to comply is not that are fruitful both as dependent variables in
enough if the ability to comply is not there. explaining implementation outputs and as
Thus, awareness of rules and financial capacity independent variables in explaining outcomes.
increase farmers’ compliance. As the most relevant criterion for evaluating
An understanding of the motivations and the relevance of a policy is its outcomes, some
incentives of target groups is essential for speci- implementation scholars are tempted to skip
fying causal links between implementation focusing on outputs and go directly to examining
behavior and target group responses. Further the relation between implementation processes/
research along this line has shown that inspec- structures and outcomes (Lynn, Heinrich and
tors not only affect farmers’ compliance directly Hill, 2001; Heinrich and Lynn, 2000; Meier and
through social motivation. They can also do so O’Toole, 2004). However, if implementation is
indirectly by using deterrence, because frequent important to outcomes, it is likely to work
inspections increase farmers’ perceived risk of through outputs, and we will not get a full under-
being caught if violating the rules – or they can standing of the causal links between implemen-
use information provision for increasing regula- tation and outcomes, unless we understand how
tees’ awareness of rules. Affecting their norma- implementation structures and processes shape
tive commitment to comply is much trickier. outputs, and how outputs shape outcomes. Often
Inspectors try often to do so, but are unlikely a given implementation structure – for example
to succeed because farmers do not trust them specialization of front-line staffs – affects several
enough. In contrast, advice from credible front-line practices with opposite effects on out-