Policy Learning and Policy Failure 9781447352013

Table of contents :
Front Cover
Policy Learning and Policy Failure
Copyright page
Dedication
Table of contents
List of figures and tables
List of abbreviations
Notes on contributors
Acknowledgments
1 Policy learning and policy failure: definitions, dimensions and intersections1
Introduction
Organising the literatures
Defining learning and failure
Learning and failure processes
Learning and failure products
Analytical levels of learning and failure: micro, meso and macro
Learning and failure at the micro-level and their interactions
Learning and failure at the meso-level and their interactions
Learning and failure at the macro-level and their interactions
Understanding the link between learning and failure
Conclusion
Note
References
2 Pathologies of policy learning: what are they and how do they contribute to policy failure?1
Introduction
Conceptualising bovine tuberculosis (BTB) management as a degeneration of epistemic learning
Unpacking epistemic learning
Pathologies of epistemic learning and organisational capacity in BTB management
Absorptive capacity
Administrative capacity
Analytical capacity
Communicative capacity
Conclusion
Notes
References
3 Overcoming the failure of ‘silicon somewheres’: learning in policy transfer processes
Introduction
Case selection
Policy transfer
Learning and adaptation
Success and failure in policy transfer
Silicon Valley
Silicon Somewheres
New Jersey
Texas
South Korea
Medicon Valley
Analysis
Conclusion
Acknowledgements
References
4 Between policy failure and policy success: bricolage, experimentalism and translation in policy transfer
Introduction
The diffusion of innovation and transfer of policies
Truncated transfer and ‘dud’ diffusion
The wrong lessons (and whose lessons?) for policy formulation
Negative lesson-drawing
Intermediaries and interests
Policy translation
Temporal transfers
Conclusion
Acknowledgements
Note
References
5 British Columbia’s fast ferries and Sydney’s Airport Link: partisan barriers to learning from policy failure
Introduction
Failing to learn from policy failure
Barriers to learning from policy failure: the case studies
Case 1: British Columbia’s fast ferries
Case 2: The Sydney Airport Link
References
6 Policy failures, policy learning and institutional change: the case of Australian health insurance policy change
Introduction
The concept of policy failure: who judges, when and against whose standards?
Learning from policy failure
Public political inquiry
Expert inquiry
Policy failure as a catalyst for institutional change
Australian health insurance policy reform from early 1970s to late 1990s
Conclusion
Acknowledgements
References
7 Policy myopia as a source of policy failure: adaptation and policy learning under deep uncertainty
Introduction
Modelling policy myopia
Dealing with different types of uncertainty
Learning to deal with policy myopia: promise and problems
Conclusion
Notes
References
Index
Back Cover

Citation preview

New Perspectives in Policy & Politics Edited by Sarah Ayres, Steve Martin and Felicity Matthews

Policy learning and policy failure Edited by Claire A. Dunlop

New Perspectives in Policy & Politics Edited by Sarah Ayres, Steve Martin and Felicity Matthews

POLICY LEARNING AND POLICY FAILURE Edited by Claire A. Dunlop

First published in Great Britain in 2020 by Policy Press North America office: University of Bristol Policy Press 1-9 Old Park Hill c/o The University of Chicago Press Bristol 1427 East 60th Street BS2 8BB Chicago, IL 60637, USA UK t: +1 773 702 7700 t: +44 (0)117 954 5940 f: +1 773-702-9756 [email protected] [email protected] www.policypress.co.uk www.press.uchicago.edu © Policy Press 2020 British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data A catalog record for this book has been requested ISBN 978-1-4473-5200-6 (HB) ISBN 978-1-4473-5201-3 (ePdf) ISBN 978-1-4473-5202-0 (ePub) The right of Claire A. Dunlop to be identified as editor of this work has been asserted by her in accordance with the Copyright, Designs and Patents Act 1988. All rights reserved: no part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise without the prior permission of Policy Press. The statements and opinions contained within this publication are solely those of the editor and contributors and not of the University of Bristol or Policy Press. The University of Bristol and Policy Press disclaim responsibility for any injury to persons or property resulting from any material published in this publication. Policy Press works to counter discrimination on grounds of gender, race, disability, age and sexuality. Cover design by Dave Worth Cover image credit: kindly supplied by Asif Akbar Printed and bound in Great Britain by CPI Group (UK) Ltd, Croydon, CR0 4YY Policy Press uses environmentally responsible print partners

For Kaye

Contents List of figures and tables List of abbreviations Notes on contributors Acknowledgements 1

2

3

4

5

6

7

vi vii viii x

Policy learning and policy failure: definitions, dimensions and intersections Claire A. Dunlop Pathologies of policy learning: what are they and how do they contribute to policy failure? Claire A. Dunlop Overcoming the failure of ‘silicon somewheres’: learning in policy transfer processes Sarah Giest Between policy failure and policy success: bricolage, experimentalism and translation in policy transfer Diane Stone British Columbia’s fast ferries and Sydney’s Airport Link: partisan barriers to learning from policy failure Joshua Newman and Malcolm G. Bird Policy failures, policy learning and institutional change: the case of Australian health insurance policy change Adrian Kay Policy myopia as a source of policy failure: adaptation and policy learning under deep uncertainty Sreeja Nair and Michael Howlett

Index

1

23

49

71

93

113

133

155

v

List of figures and tables Figures 2.1 2.2 3.1 7.1 7.2 7.3 7.4

Conceptualising knowledge modes as policy learning Expanding epistemic learning Learning processes during policy transfer Typology of uncertainties by options, outcomes and values Policy maker’s knowledge and comprehension matrix Different kinds of risk faced by policy makers and potential solutions Characteristics of different types of uncertainty

27 29 56 137 138 139 139

Table 2.1

Organisational capacities and epistemic learning degeneration

vi

31

List of abbreviations ACAP ADCAP ALP ANCAP BBC BC BTB CBA COMCAP DEFRA EBPM EFRAC EU IBM IMF IR ISG IST KAIS LTV MAFF MPA NASA NDP NFU NGOs OECD PFP R&D RBCT RCT SMU SVM TI UC UK US

absorptive capacity administrative capacity Australian Labor Party analytical capacity British Broadcasting Corporation British Columbia bovine tuberculosis cost-​benefit analysis communicative capacity Department for Environment, Food and Rural Affairs evidence-​based policymaking Environment, Food and Rural Affairs Select Committee European Union International Business Machines corporation International Monetary Fund international relations Independent Scientific Group Institute of Science and Technology Korea Advanced Institute of Science Ling-​Temco-​Vought corporation Ministry of Agriculture, Fisheries and Food Major Projects Authority National Aeronautics and Space Administration New Democrat Party National Farmers Union non-​governmental organisations Organisation for Economic Co-​o peration and Development pay-​for-​performance research and development Randomised Badger Culling Trial randomised controlled trial Southern Methodist University Silicon Valley Model Texas Instruments Universal Credit United Kingdom United States

vii

Notes on contributors Malcolm G.  Bird is associate professor of political science at the

University of Winnipeg, Manitoba, Canada. He is fascinated by the evolution of Canadian state-​owned enterprises (SOEs) and the decision-​making processes within Westminster Parliamentary systems. His research efforts focus on examining the modernisation of contemporary Canadian SOEs and how they have adapted their internal operations and governance regimes over the last thirty years. He marvels at the capacity of Canadian SOEs to balance diverse and often competing demands from numerous stakeholders as well as from their political masters. He holds a PhD in Public Policy and Administration from Carleton University. Claire A. Dunlop is professor of politics at the University of Exeter,

UK. A public policy and administration scholar, Claire’s main fields of interest include the politics of expertise and knowledge utilisation; epistemic communities and advisory politics; risk governance; policy learning and analysis; impact assessment; and policy narratives. Her recent co-​edited volume (with Claudio M. Radaelli and Philipp Trein) is Learning in Public Policy: Analysis, Modes and Outcomes (Palgrave, 2018). Since 2014 she has been editor of Public Policy and Administration and in 2018 became a trustee of the UK Political Studies Association. Sarah Giest is an assistant professor with the Institute of Public

Administration at Leiden University, Netherlands. Sarah specialises in public policy analysis focusing on policy instruments and capacity in the innovation, technology and sustainability realm. This includes, for example, the use of big data for public climate change efforts or the capacity of government to innovate in urban settings. Her work has been published, among others, in Energy Policy, Environmental Science & Policy, Policy Sciences, and Public Administration. She is on the Editorial Board for Policy Design and Practice and member of the Young Academy Leiden. Michael Howlett is Burnaby Mountain Professor and Canada Research

Chair (Tier 1) in the Department of Political Science at Simon Fraser University, Canada. He specialises in public policy analysis, political economy, and resource and environmental policy. Adrian Kay is an Honorary Professor at the Crawford School of Public

Policy at the Australian National University where he was previously

viii

Notes on contributors

Professor of Government. He is a past President of the Australian Political Studies Association and has also held Chairs in the UK and Asia. He was a member of the UK government’s European Fast Stream for several years and worked for the EU Commission in Brussels prior to a career in academia. His research lies at the intersection of international and comparative public policy, with a current focus on contributing to understanding the relationships between Islam and public policy making in different institutional contexts. Sreeja Nair is a research fellow at the Nanyang Environment and

Water Research Institute, Nanyang Technological University, Singapore where her research focuses on bridging socio-​cultural and political dimensions with technological aspects of water sustainability in Asia. Her research interests include policy design under uncertainty and impacts of environmental change on communities, focusing on water and agriculture sectors. Joshua Newman is a senior lecturer in the College of Business,

Government and Law at Flinders University in Adelaide, South Australia. His research interests involve the relationship between government and private sector organisations, including regulation, privatisation, and public-​private partnerships. In addition, Joshua has written about research utilisation and evidence-​based policy, policy outcomes and evaluation, and managing wicked policy problems. He is the author of Governing Public-​Private Partnerships (McGill-​Queen’s University Press, 2017). Diane Stone is Dean of the School of Public Policy at the Central

European University. In order to oversee the transition of the School from Budapest to its new home in Vienna, she has moved from her position in Australia as Centenary Professor in the Institute for Governance and Policy Analysis at the University of Canberra. Previously, she was a Professor of Politics and International Studies at Warwick University (1996–​2019), a Professor of Politics at the University of Western Australia (2010–​13), and a European Commission Marie Curie Chair (2004–​08) at Central European University. Additionally, she worked at the World Bank, in the Secretariat that launched the Global Development Network in 1999, then became a member of its Governing Body (2001–​05). Currently, Professor Stone is Consulting Editor of Policy & Politics as well as Vice President of the International Public Policy Association. Her most recent publication is Global Policy and Transnational Administration (OUP, 2019).

ix

newgenprepdf

Acknowledgements The chapters for this volume arise from a policy failure and learning workshop hosted by Lee Kuan Yew School of Public Policy (LKYSPP), National University of Singapore, 19–​21 February 2014. All the authors wish to thank the workshop organisers –​Michael Howlett, M Ramesh and Xun Wu  –​and all the other participants for their intellectual generosity and support. As editor, I want to extend my sincere thanks to all the anonymous referees whose comments inspired our authors and helped improve the chapters. The collection first appeared as an edited special issue of Policy & Politics (2017, volume 45, issue 1) and we thank the editorial team who made it such a rigorous and fun experience! Claire A. Dunlop Exeter, June 2019

x

1

Policy learning and policy failure: definitions, dimensions and intersections1 Claire A. Dunlop

Introduction The social security system in the UK has long been regarded as overly bureaucratic and too complex for either bureaucrats or claimants to entirely understand. In 2010, the Conservative-​led coalition government unveiled Universal Credit (UC) as the answer to this historic policy problem. An ambitious plan to merge six in-​and out-​ of-​work benefits, UC aims to ensure not only a simpler system but also that it is more beneficial to be in work than on the dole. Entailing a huge administrative challenge, the implementation of UC has been dogged by problems from the outset; with costs spiralling and the timetable slipping, the policy was effectively ‘reset’ in 2013 and political pressure to abandon it mounted in the months that followed. Yet, 2013 proved to be a turning point. Looking into the precipice of the failure of a flagship reform, policy makers engaged in policy learning. Along with analysing the technical problems and capacity deficits they faced, civil servants learned from previous experiences of implementing complex policies and from similar problems in social security reform in Australia. Appointing senior troubleshooters responsible for getting the programme on track and engaging a recovery team –​Major Projects Authority (MPA) –​the UC appeared to have been turned around (see Timmins, 2016 for a full account). But such optimism was premature. Since its staggered roll-​out, the Resolution Foundation think tank estimated UC will result in 3.2 million working families being worse off by an average of £48 a week. As many as 600,000 claimants are no longer entitled to any assistance at all (Finch and Gardiner, 2018). In April 2018, the Trussell Trust charity reported an average 52 per cent rise in demand for its food banks in areas where UC had been rolled out (Trussell Trust, 2018). Though reputed by government at

1

Policy Learning and Policy Failure

the time, in February 2019 the sixth Secretary of State for Work and Pensions presiding over UC –​Rt Hon Amber Rudd –​conceded that its introduction had contributed to increased food insecurity (HC Deb, 11 February 2019, c593). As we can see, though policy failures present valuable and multiple opportunities for policy learning, as the case of UC demonstrates, this potential is very difficult to exploit. In short, applying lessons in a sustained way to reset policies that are failing is very difficult. And yet, though the likelihood of policy failure is at least as high as policy success, the existing literature has focused disproportionately on the latter. Compared to the large volume of publications on ‘good practices’ and ‘best practices’, far less scholarly attention has been paid to ‘bad practices’ or ‘worst practices’ despite their widespread prevalence. As a result, public officials have failed to learn valuable lessons from these experiences. Certainly, we would not dispute that policy learning is difficult especially in situations underpinned by ambiguity (March and Olsen, 1975). But, studies of policy failure are marked by how rarely failure is averted or followed by learning (Bovens and ’t Hart, 2016: 662; Moran, 2001). The studies in this volume stand as testament to this broken link. Both policy learning and policy failure are classic topics of policy studies. The analysis of learning was central to the post-​war beginnings of policy analysis and management science (notably, Deutsch, 1966; Heclo, 1974; Lindblom, 1959; 1965; and Simon, 1947). Analysis of policy failures (or fiascos, blunders, disasters, anomalies) came a little later in part a result of the advent of policy evaluation in the 1970s –​ (for example, Bovens and ’t Hart, 1996; Dunleavy, 1995; Gray, 1996; Hall, 1981; 1993; Ingram, 1980; Moran, 2001; Wolf, 1979).This interest has been sustained. There have been four journal special issues on policy learning since 2009 –​two on learning and transfer (Evans, 2009; Pemberton, 2009), a third on learning at the organisational level (Zito and Schout, 2009) and, most recently, a collection of articles exploring learning and policy change (Moyson et al, 2017). Policy failure too has been the subject of intellectual energy with two recent special issues –​one on the persistence of policy failures (Howlett et al, 2015) and the other on foreign policy failures (Bovens and ’t Hart, 2016) –​and several book-​length treatments (Birkland, 2006; Crewe and King, 2013; Schuck, 2014). The links between the two literatures appear obvious, yet there are very few studies that address how we can (though see O’Donovan, 2017): learn from failure, learn to limit failure and fail to learn. This collection offers a rare attempt to bring these two literatures together.

2

Definitions, dimensions and intersections

We start by defining policy learning and failure before organising the main studies in these fields along the key dimensions of processes, products and analytical levels. The intention here is three-​fold, to:  provide a manageable review of the literatures, highlight key intersections between learning and failure, and give voice to some of the questions implied by their linkages. We continue with an overview of the chapters in this volume, outlining where they sit in the wider literature and how they link learning and failure. We conclude by sketching a research agenda linking policy scholars with policy practice.

Organising the literatures Defining learning and failure To understand learning and failure –​both as independent and linked phenomena –​we need to work from some agreed definitions. At first blush, this appears to be a tricky task; policy learning and failure are each conceived in very different ways by different scholars. In one view, both learning and failure have an ‘eye of the beholder’ quality –​ learning can be hard to perceive and demonstrate (Radaelli, 2009), and failure is often treated as similarly elusive, as something that is framed and made rather than existing in its own right (Bovens and ’t Hart, 1995; 2016; Edelman, 1964; Zittoun, 2015). This view, that learning and failure can never be treated neutrally, is not simply a reflection that politics matters (though it does, of course), it is also in part the result of the common currency they have in life in general; we have all experienced both learning and failure. Another perspective views both phenomena as technical and so, eminently, accessible. We can generate metrics to demonstrate changes in our understandings and beliefs (Zafonte and Sabatier, 1998) and policy outcomes (Wolman, 1981).Yet another perspective treats policy learning and failure as highly complex and difficult to analyse in any systematic way (Pressman and Wildavsky, 1973).These different perspectives reflect the multi-​disciplinary nature of these phenomena –​education studies, management sciences, psychology, sociology, political science all claim ownership of learning and failure. Policy scholars have a huge range of analytical toolkits on which to draw and, as a result, we lack any grand theories of either concept. Yet, surprisingly, this analytical eclecticism has not resulted in definitional confusion on the fundamentals. Literature reviews reveal that even those authors working in different disciplines, and with contrasting ontologies and epistemologies, alight from similar basic understandings of these

3

Policy Learning and Policy Failure

phenomena. A recent review of policy learning in the social sciences identifies a minimal definition where learning is:  ‘the updating of beliefs based on lived or witnessed experiences, analysis or social interaction’ (Dunlop and Radaelli, 2013: 599). Recent policy failure studies have achieved a similarly broad view; a policy fails ‘even if it is successful in some minimal respects, if it does not fundamentally achieve the goals that proponents set out to achieve, and opposition is great and/​o r support is virtually non-​e xistent’ (McConnell, 2015: 221). Working out from this common ground, scholars bolt on their own analytical specifications in these key dimensions: processes, products and analytical levels. Learning and failure processes The process of policy learning concerns how we identify learning and in particular addresses questions of intentionality, depth of learning, and measurement; to what extent is learning planned, how far does it go and how can we measure it? For some, learning as updating is intentional; a deliberate moment to pause and draw lessons (Hall, 1993; Rose, 1991). Such opportunities are often created by the recognition that things have not gone according to plan. Yet, other analysts treat learning as more evolutionary and organic (Heclo, 1974). Here, updates in beliefs are non-​linear and the logic is one of enlightenment as opposed to instrumental responses (Weiss, 1977). Learning can be a drawn-​out process where failure events or new evidence may need to build over time. For others still, learning can be unintended (Liberatore, 1999). Regardless of how it happens, learning is distinguished in terms of depth. Management theorists Argyris and Schön (1978) famously note the difference between ‘single-​’ and ‘double-​loop’ learning where the policy tool change of the former is more common than the deeper level learning associated with the alteration of political objectives assumed in the latter. Variations of this theme of depth pervade the learning literature (for example, Dolowitz (2009) on ‘hard’ and ‘soft’; Hall on paradigmatic change (1993); Levy (1994) on ‘simple’ and ‘complex’; and Sabatier and Jenkins-​Smith (1993) on policy-​oriented learning). The final aspect of learning processes concerns measurement. How do we know that learning has taken place? A range of methods are used to analyse policy learning, ranging from qualitative data analysis (process-​tracing and elite interviews –​for example, Dunlop and Radaelli, 2016), or quantitative data analysis (for example, Moyson, 2018), or panel studies (for example, Witting and Moyson, 2015). Yet, we know far less about the specification of what is actually being

4

Definitions, dimensions and intersections

operationalised and measured (though see Leach et al, 2014). Here, we have the curious phenomenon of measurement based on unclear and often unspecified criteria. Now we turn to failure and consider:  its origins, depth and measurement challenges. As with policy learning, we can distinguish policy failures that have been willed in some way –​most obviously, policies that are allowed to fail or are malevolent (see Newman and Bird, 2017; on state crimes and elite political corruption see deHaven-​ Smith, 2006) –​from those that are accidental and often unanticipated (Howlett, 2012).The majority of the literature focuses on the second. Turning to the depth of failure, again, Howlett pushes analysts to think clearly by examining the salience (intensity) and magnitude (extent and duration) of policy failure (2012, Table 2, 544). Judgements about intensity concern metrics of visibility and coverage  –​is the failure high profile or dramatic in its effects? Extent and duration address the range and scope of the failure –​has its impact been felt widely, by high status groups, over a long period? On measurement, policy failure has the opposite problem to that found in learning studies. What is being measured is well scoped-​out in studies but the methods are limited to qualitative accounts. For example, Bovens and ’t Hart (1996) use evaluation to distinguish between two logics of failure measurement –​ programmatic and political. The former is the world of facts where failures are declared on the basis of metrics  –​cost–​benefit analysis (CBA), the comparison of objectives and outcomes, and so on. By contrast, political measurement logics are value-​driven –​impressions, experiences, incentives and stories all come into play. Policies which fare badly by both logics are the deepest failure type (Bovens and ’t Hart, 2016, Table 1, 657). Learning and failure products Yet, the processes that underpin policy learning and failure are often left undefined, with analysts preferring to focus on the type of phenomena being produced. The literature tends to focus on products in terms of policy preferences; policy learning has been described as the: lessons drawn about policy instruments (instrumental learning), societal construction of policy problems (social learning), or feasibility of policy objectives (political learning) (May, 1992; see also Bennett and Howlett, 1992). Some theorists have taken us beyond the realm of policy design preferences, for example, into the world of institution building and social identities (Checkel, 2001). There has been a good deal of lively discussion about how to categorise policy failures (and successes)

5

Policy Learning and Policy Failure

(Bovens, 2010; Bovens and ’t Hart, 2016; Marsh and McConnell, 2010). One of the most widely used is McConnell’s (2010; 2015) three-​fold categorisation where failure is the result of: technical and substantive deficiencies that prevent goals being reached (programme failure), or an inability to negotiate the policy process and translate an idea into reality (process failure), or partisan distortion of the policy (political failure). Analytical levels of learning and failure: micro, meso and macro Finally, we have the specification of analytical levels where analysts explore the relationships between processes and products. This concerns questions of who is learning or failing, and in what political arenas. Scholars examine this in relation to the familiar micro-​, meso-​and macro-​levels, where learning or failure may be the driver (that is, independent variable) or subject of change (dependent variable). Learning and failure at the micro-​level and their interactions Micro-​level studies are concerned with the micro-​foundations of human action. Drawing on economics, behavioural sciences and social psychology learning studies at this level zoom in on models of the mind, emotions and rationality and how individuals’ learning is mediated by heuristics, persuasion, identity and evidence (see, for example, Denzau and North, 1994; Deutsch, 1966; Kahneman, 2011; Simon, 1947). While we know a good deal about what conditions individual policy actors’ learning, we know far less about its inverse –​the policy impact of an individual’s learning capacity. Micro-​level analyses are well-​represented in the failure literature. Attention is focused mainly on the impact of individuals on policy failure (less is known about the inverse). Again, scholars draw on behavioural psychology explaining policy failure in terms of human cognition –​for example, over-​reliance on analogies as evidence (Khong, 1992), personality traits (Brummer, 2016), inaccurate risk calculations (Owen, 2012), or poor leadership (Bovens and ’t Hart, 1996). These micro-​level concerns of learning and failure intersect in two ways. First comes the effects of policy failure on individuals’ ability to learn. For example, a recent study uses models of contingent learning from economics to explore how surprise failure  –​in this case the Euro crisis –​forced immediate responses from policy makers which ultimately, over time, has resulted in a learning dividend for those actors (Kamkhaji and Radaelli, 2017). Second, we consider how an

6

Definitions, dimensions and intersections

individual’s learning capacity has an impact on policy failure. Whether a better equipped learner prevents or mitigates failure in ways another cannot has received little attention so far, yet this concern dovetails with the leadership literature which could usefully explore the impact of leadership training on failure situations. Learning and failure at the meso-​level and their interactions Next comes the meso-​level: the realm of group interactions in policy making. Like most policy theories, the mid-​range is the home of the bulk of policy learning and failure analyses. Learning in group settings is addressed in two main ways. First comes learning as the causal mechanism, where updates in beliefs drive change through, for example, the policy debates of advocacy coalitions (policy-​oriented learning, Sabatier and Jenkins-​Smith, 1993) or highly specialised lessons provided by epistemic communities in situations of uncertainty or think tanks in areas of complexity (Haas, 1992; Stone, 2005). The governance literature has also rediscovered learning in recent years exploring the other side of the relationship  –​how organisational properties regulate learning and influence what lessons are adopted. For example, this concerns reform processes that lead to experiential learning (Olsen and Peters, 1996); delegation systems open enough to update themselves (Dunlop and James, 2007; Waterman and Meier, 1998); socialisation processes that drive learning in organisations (Checkel, 2005); experimental institutions (Gerstenberg and Sabel, 2002); technocratic governance regimes (Demortain, 2011); and, models of mutual adjustment (Elgström and Jönsson, 2000). Only one side of policy failure at the meso-​level is well studied. Though we know little about the impact of failure on groups and institutional processes, in contrast, much attention focuses on failure as the dependent variable where studies examine the collective processes that mediate the likelihood and form of failure. Janis’s (1972) famous groupthink study in foreign policy blazed the trail. In this case the close-​knit, secretive nature of collective interactions sowed the seeds of policy failure. In political economy, the asymmetrical power of well-​resourced groups is the ubiquitous explanation for policy (and economic) failure (Rodrik, 2014).There is also a growing literature associating failure with group activities at particularly complex stages of the policy process where the range of actors’ interpretations result in friction, contestation and unintended consequences –​most obviously, implementation (Kerr, 1976; May, 2015; Pressman and Wildavsky, 1973; Schuck, 2014; for a wider account of failure and policy stages

7

Policy Learning and Policy Failure

see Howlett et al, 2009). Finally, we have bureaucratic analyses that link policy failures to weaknesses in institutional resources and turf wars (Gabriel, 1986; Peters, 2015; Vaughan, 1996). Thinking about intersections at the meso-​level, we can first consider the impact of learning processes on failure. Here, we tap into recent work exploring the idea of dysfunctional learning (Dunlop, 2014, Table 2, 216; Dunlop and Radaelli, 2016; 2018). The proposition is that learning may not always be a good thing. In their account of policy failures in the Eurozone, Dunlop and Radaelli outline this disruptive potential and the fragility of learning between Member States and EU institutions (see 2016: 111, 117–​219). How can the failure of groups or collective processes mediate policy learning? We have established that there is a dearth of knowledge about failure as an independent variable and much of what we do know is negative. Policy failures in implementation processes and of the technical programme variety rarely trigger learning processes at all (Moran, 2001).The same can be said of fiascos linked to a party or government that are often repeated by their rivals or replacements (Howlett et  al, 2015). Even where there is policy evaluation or inquiries, such collective processes are no guarantee that lessons will follow and be applied in future. Learning and failure at the macro-​level and their interactions Finally, we have analysis that aggregates to the macro-​level. Here, we enter the world of institutional analyses that explore learning and failure dynamics embedded in policy histories, cultural identities and society-​wide memories. Learning studies at the macro-​level are less numerous that those of the mid-​range. Developing wide-​ranging state or sector narratives requires highly specialised methodological skills –​notably, the ability to marshal large amounts of (historical) data to produce precision process-​tracing. While institutional analyses are fewer in number, they pack an analytical punch. Many of the examples of state-​centred learning analysis can be considered modern classics. Most obviously, we can point to Peter Hall’s (1993) seminal work on economic policy making in Britain where social learning lies at the heart of a paradigmatic policy break from Keynesianism. Such third order changes are macro-​level earthquakes, momentous ‘strategic moments’ in state development (Hay, 2001: 202), whose reverberations are felt across society, its citizens, institutions, cultural practices and history. What is the impact of macro-​level variables on learning? Do certain types of societies have a greater or lesser disposition to learn than others?

8

Definitions, dimensions and intersections

In another modern classic of comparative politics, in his analysis of Italy, Robert Putnam (1993) demonstrates that cultural rules have an impact on learning in stable and continuous ways over long periods. Distrust of state institutions and attachment to alternative familial institutions in the south resulted in the reduced capacity for social learning. What of policy failure at the macro-​level? Again, our knowledge of this level is somewhat lopsided. And, again, it is failure as an independent variable where knowledge is limited; we know little about the impact of policy failures in societies as a whole. Elmore (1987) argues that failure at the system level presents the opportunity for learning. This seems self-​evident, but we need to go further and provide empirical analysis. Attention has focused on its inverse –​how societies mediate and regulate failure. One sizeable body of work concerns state corruption and the impact of cultural values and historical experiences on failure. For example, Rothberg’s (2003) wide-​ ranging study explores the inherent instability of newly independent nation states (whose numbers have nearly trebled since the end of the Second World War) and the impact of weak economic structures and social conflict on policy outcomes. In policy studies, a recent study of healthcare in India traces the failure to develop state policy capacity back to the historically rooted and sustained dominance of the private sector (Bali and Ramesh, 2015). Turning to the intersections of policy learning and failure, we first consider the impact of policy failure at the macro-​level on policy learning. While in commercial innovations, system-​level failure is famously vaunted as an opportunity for rebirth and rethinking (Schumpeter’s (1942) famous ‘creative destruction’), societies that experience fundamental failures may run from its embrace. The West’s response to the recent banking crisis provides a case in point. Here ‘too big to fail’ arguably became the cognitive order of the day. Society-​wide failure of large parts of the financial systems of major economies became reframed as a problem of public sector profligacy (Blyth, 2013). One of the many effects of this was to dilute the power of the lessons that were drawn. This example raises some fundamental issues of power and blame avoidance that recur in policy failure studies (Balla et al, 2002; Newman and Head, 2015). The risk here is that societies learn the wrong lessons from failure, lessons that then go on to affect future macro-​level failures. The circular potential of learning and failure has been brilliantly worked through in a recent analysis of European integration and the failure in the Eurozone (Jones et al, 2016). Incomplete learning in the 1990s and early 2000s resulted in a weak governance architecture which

9

Policy Learning and Policy Failure

unleashed the forces that took the Eurozone into deep crisis. This failure was met with a sequence of partial reform, again underpinned by shallow learning, followed by failure. Paradoxically this sustained cycle of incomplete reform followed by crisis has resulted in further integrationary pressures  –​leaving the EU project ‘failing forward’ (Jones et al, 2016).

Understanding the link between learning and failure Collectively, the chapters in this volume demonstrate that effective research of the intersection between policy learning and policy failure requires clarity about the nature of our conceptual choices. Empirically, they confirm the difficulty of learning in a range of policy settings and of applying lessons to ameliorate failure. Yet, despite the range of challenges, an overarching theme is that the potential to learn from policy failure remains strong, and the consequence of failing to learn makes further research in this area important. Claire Dunlop (2017) begins the volume with an exploration of how dysfunctional forms of policy learning impact policy failure at the meso-​level. Using the long-​running policy failure of the management of bovine tuberculosis (BTB) in England, analysis focuses on negative lessons generated by the interactions of an epistemic community of scientific experts and civil servants charged with balancing the competing interest actors to craft a workable policy. Two ideas are of note for the learning and failure literatures. First, by introducing the idea of dysfunctional or degenerate learning, Dunlop reminds us that learning is not always a ‘good thing’. The chapter conceptualises learning degenerations as structured into the management of scientific advisors on BTB. Offering a transferable analytical framework linking learning and organisational capacity to failure allows for more prescriptive thinking about how organisational capacity can be generated to encourage the functional learning required to avoid or correct failures. In her chapter, Sarah Giest (2017) explores the impact of different types of learning on the success and failure of the transfer of the famous Silicon Valley Model (SVM) of innovation. Working with the idea of ‘adaptive learning’, this contribution underlines the importance of understanding the learning process, and critically, the depth of learning that underpins policy transfer. Giest uses four cases to demonstrate how different learning processes generated by actors at the meso-​level, mainly networks of stakeholders and experts, mediate the extent to which policy transfer is a success or failure. Giest deepens our basic

10

Definitions, dimensions and intersections

definition of learning as updates by exploring how these updates are affected. While learning by imitation and trial and error resulted in policy failures in the two US cases, deeper level adaptive learning outside North America helped secure successful SVMs. Giest’s analysis suggests that depth of learning in these cases was influenced by spatial and temporal distance from the original SV case. Giest wisely tempers her conclusions noting the measurement problems associated with policy learning. Continuing the transfer theme, Diane Stone (2017) highlights the importance of interrogating how we define policy failure and learning. Taking us beyond our basic definitions, Stone’s sophisticated conceptual chapter highlights the contingent nature of policy failure through an examination of the dynamics that underpin policy transfer. Rather than frame a policy transfer as a failure or success, Stone argues that scholars must recognise transfer (and so failure) as a messy process involving an array of meso-​level actors. Echoing our discussion of the importance of perception and interpretation in how we conceptualise learning and failure, transfer too has multiple dimensions and failure of it is rarely outright or settled. Rather than treat partial transfer as failure, this is a re-​imagining of transfer where ‘failure’ and ‘success’ sit alongside each other, underpinned by ongoing and dynamic learning processes. Two aspects are of particular note. First, the treatment of imperfect transfer as underscored by flawed lesson-​drawing is useful as it takes us back to questions about the depth of learning. Second, Stone highlights two aspects of learning that are often overlooked in mainstream accounts: ‘negative lesson-​drawing’ and selective learning. While we think about learning in terms of updates, we should be forensic in our exploration of the basics and, specifically, what is being updated. Joshua Newman and Malcolm Bird (2017) move us beyond a focus on process to offer a much-​needed analysis of policy failure as an independent variable. Comparing two transportation cases –​fast ferries in British Columbia and Sydney’s airport link –​the chapter adds to our knowledge of the impact failure on group dynamics in general and on collective processes of learning more specifically. Both cases demonstrate the importance of defining policy failures as being attached to the goals outlined by the policy’s proponents. The failure cases coincided with changes in government and presented the opportunity for partisanship –​whereby incoming administrations that did not ‘own’ the projects could foster the conditions to stall learning that could have rescued the policies. Rather than explore the various remedies available, the new governments took the political

11

Policy Learning and Policy Failure

route linking the failures to the outgoing administrations by piling on criticism and engaging in activities –​for example, holding inquiries –​ that boosted the profile of the failures. This takes us back to how we broaden out the definition of learning and define the beliefs being updated:  are they substantive or political? It also raises the issue of learning and intentionality. While learning can be intentional, ignorance can also be wilful. Adrian Kay’s (2017) chapter addresses two gaps in the study of policy learning and failure. His study of policy change in Australian health insurance explores both how failure works as an independent variable affecting policy learning and situates analysis of learning at the system level. Kay’s study demonstrates the capacity for failure to become its own cause (to adapt Wildavsky, 1979). Here, failure delegitimised health policy institutions making them increasingly vulnerable and giving them weak learning capacity to reform in anything but a suboptimal way. The result is a cycle of failure and dysfunctional learning. Kay’s chapter demonstrates the capacity for policy regimes to become systems whose ingrained logics individual elites and policy communities cannot tame or reform. Here we have a bread-​and-​butter policy issue of political salience and significant electoral consequences; so the incentives to learn are high. This carries implications for how we define our core concepts; failure and learning can be locked together creating a perversely resilient policy failure inescapable through updates or the presence of new policy actors. Finally, Sreeja Nair and Michael Howlett (2017) go beyond the empirical and push us to think about the impact of the future on policy actors. Their chapter introduces the idea of ‘policy myopia’ as a pressing source of failure in policy making and explores the possibility of developing policies that learn to help mitigate its impacts. In their conceptualisation of myopia, Nair and Howlett note that while the problem of bounded rationality and short-​term uncertainty is widely acknowledged as the central existential condition for all policy making, the long-​term problem of an uncertain, and sometimes unknowable, future is rarely acknowledged. As uncertainty deepens, so too does the probability of policy failure. In these circumstances, actors need to design policies with flexibility and adaptation built in. Where policy solutions are robust over a range of possible scenarios and over time, we can say policies have learning capacity organised into them. Yet, in cases of radical uncertainty, learning may not be possible (or indeed preferable) at all. This reminder of the limits of learning is important. Updating our belief systems assumes that a certain amount of knowledge exists in the first place.

12

Definitions, dimensions and intersections

Conclusion No edited collection is exhaustive and this one is no exception. The collection tells us a good deal about how we might conceptualise learning and failure processes in terms of degenerations, transfer and uncertainty over the long term (Dunlop; Stone; Nair and Howlett).Yet, more work needs to be done on policy failure methods and operationalisation of learning. In terms of analytical levels, three of the studies extend our understanding of the familiar theme of learning as the cause of failure (Dunlop; Giest; Stone), and two (Kay; Newman and Bird) provide much-​needed cases studying the impacts of failure on learning. With the exception of Kay (2017) our studies are clustered around the familiar meso-​level. None address micro-​level arenas of failure and learning. Much of our discussion has focused on policy learning and failure analysis of the policy process. However, policy studies is a practice-​ oriented endeavour, and both topics have strong real-​world dimensions of fundamental concern to policy makers and practitioners as well as analysts (Taylor, 2016). Learning and failure studies are beginning to reflect this and offer analysis in and for the policy process that concentrates on the prescriptive techniques that can help on the ground. Intellectual endeavours on the design implications of learning and failure are still in their infancy but two streams of activity are making headway. For learning, analysis of international organisations makes particularly strong offerings on how governments should learn. Different instruments and methods for cross-​national learning include: benchmarking, peer review, checklists, facilitated coordination and extrapolation (Barzelay, 2007; Borrás and Radaelli, 2011; Schäfer, 2006). The prescriptive turn in failure studies is less concerned with how not to fail and more focused on its inverse –​how to succeed in policy making (Rutter, 2010; Rutter et al, 2012; Timmins, 2016). Bringing together policy learning and failure literatures demonstrates that there are many points of intersection to help us push the research agenda forward. And so, we finish by sketching out some possible topics for the future that focus on linking research with policy practice. First, we consider the impact of learning on failure. This is where most of our studies are grouped. While it seems uncontroversial to assert that learning can often be a good thing, and can help us produce better public policy, time-​and resource-​poor policy actors need to know that these lessons are going to be worth learning. Think about this at the individual level. Fast thinking is wired into human cognition, the slow reflection associated with learning is an exertion that comes with costs (Kahneman, 2011). Only when policy makers understand

13

Policy Learning and Policy Failure

that the lessons before and after failure are worth generating will they take the hit and get reflective. One key practical challenge is to make the learning updates produced after failure clearer and more applicable to the policy world. For example, this could mean working across disciplines with evaluation scholars, psychologists, historians and so on, to compare the impact of presenting information in different formats –​ numbers, stories, case studies and so on. At ease working with concepts that span the disciplines, policy scholars are well placed to do this. What about the impact of learning on failure at group level? Groups and organisations in policy making can learn much from the business world. The most successful organisations are ones that ‘structure in’ learning from failure, or develop a mindful approach (Langer, 1989; Weick, 1995; 2009). Of course, it is simpler to appreciate the benefits of learning from failure in commercial environments where goals and structures of authority are usually clearer than the public sector where policy objectives are often multiple and competing, and authority is diffuse (McConnell, 2015). Timely and fulsome analyses of even the simplest problem are easily complicated in such interaction-​dense environments (Howlett et al, 2015: 210). Uncovering and applying the lessons of failure at a macro, social level is one area where there are many blueprints. For example, moving beyond the policy literature, we can take inspiration from the work on post-​conflict societies and, specifically, the use of public reconciliation commissions and inquiries to take whole populations beyond seemingly intractable disputes (Malley-​Morrison et al, 2013; Marier, 2009). What about the impact of failure on the learning environment? We know little about policy failure as an independent variable and its impact on learning at all levels. Thinking about how policy failure affects individual policy makers’ cognition, the early policy termination literature suggests that there will be psychological discomfort from closing a policy programme or agency (deLeon, 1978), yet systematic empirical evidence is needed. Again, we can take a cue from the renewed interest in findings from behavioural psychology that demonstrate that we are risk-​averse but also poor risk calculators. For policy analysts, one challenge is to work with policy makers to understand pre-​and post-​failure mindsets and how they might encourage or inhibit learning and constructive engagement. The impact of policy failure on groups and their learning processes is another area of analytical darkness. Again, we can look to psychology where it has been long established that failure has a destabilising impact on groups (Wolman, 1960). The implications of these group dynamics on the prospects for organisational resilience or, post-​crises reforms

14

Definitions, dimensions and intersections

or, stability in post-​failure policy-​making environments are significant and require attention. Finally, we need to understand more about failure on a social scale. For example, can a society become accustomed to persistent failure (or success) in a policy area? Following Dror (2014), Bovens and ’t Hart (2016) argue that society at large may be the key to strengthening the link between learning and failure. One implication of this is that societies could collectively determine what should be framed as a failure and indeed be allowed to fail. Of course, such opening-​up of policy making, where learning and failure become public goods, requires that conditions for this reflexive type of learning are in place. Where deliberative mechanisms are merely symbolic or privileged the loudest voices, we end up with a dysfunctional learning form where framing contests create false consensus (Pellizzoni, 2001) and risk a dialogue of the deaf. Note 1 This chapter and edited collection arises from the ‘Policy Failure’ workshop hosted by Lee Kuan Yew School of Public Policy (LKYSPP), National University o f Singapore, 19–​21 February 2014. All the authors wish to thank the workshop organisers  –​Michael Howlett, M Ramesh and Xun Wu  –​and all the other participants for their intellectual generosity and support. As guest editor, I want to extend my sincere thanks to all the anonymous referees whose comments inspired our authors and helped improve the chapters. Thanks also to the team at Policy & Politics who have made this such a rigorous and fun experience! Particular thanks go to Sarah Ayres, Sarah Brown, Felicity Matthews and Rebecca Tomlinson for their patience, humour and insight. With regard to this introduction, many thanks to two anonymous referees whose feedback helped improve the manuscript. Of course, any errors or omissions in this chapter remain my responsibility.

References Argyris, C, Schön, D, 1978, Organizational learning: A theory of action perspective, Reading, MA: Addison-​Wesley Bali, A, Ramesh, M, 2015, Health care reforms in India: Getting it wrong, Public Policy and Administration 30, 3–​4, 300–​19 Balla, SJ, Lawrence, ED, Maltzman, F, Sigelman, L, 2002, Partisanship, blame avoidance, and the distribution of legislative pork, American Journal of Political Science 46, 3, 515–​25 Barzelay, M, 2007, Learning from second-​hand experience: Methodology for extrapolation-​oriented case research, Governance 20, 3, 521–​43 Bennett, CJ, Howlett, M, 1992, The lessons of learning: Reconciling theories of policy learning and policy change, Policy Sciences 25, 3,  275–​94

15

Policy Learning and Policy Failure

Birkland, T, 2006, Lessons of disaster, Washington, DC: Georgetown University Press Blyth, M, 2013, Austerity, Oxford: Oxford University Press Borrás, S, Radaelli, CM, 2011, The politics of governance architectures: Creation, change and effects of the EU Lisbon strategy, Journal of European Public Policy 18, 4, 463–​84 Bovens, M, 2010, A comment on Marsh and McConnell: Towards a framework for establishing policy success, Public Administration 88, 2,  584–​5 Bovens, M, ’t Hart, P, 1995, Frame multiplicity and policy fiascoes: Limits to explanation, Knowledge and Policy 8, 4, 61–​83 Bovens, M, ’t Hart, P, 1996, Understanding policy fiascoes, New Brunswick, NJ: Transaction Bovens, M, ’t Hart, P, 2016, Revisiting the study of policy failures, Journal of European Public Policy 23, 5, 653–​66 Brummer, K, 2016, ‘Fiasco prime ministers’:  Leaders’ beliefs and personality traits as possible causes for policy fiascos, Journal of European Public Policy 23, 5, 702–​17 Checkel, JT, 2001, Why comply? Social learning and European identity change, International Organization 55, 3, 553–​88 Checkel, JT, 2005, International institutions and socialization in Europe: Introduction and framework, International Organization 59, 4,  801–​26 Crewe, I, King, A, 2013, The blunders of our governments, London: Oneworld deHaven-​S mith, L, 2006, When political crimes are inside jobs: Detecting state crimes against democracy, Administrative Theory and Praxis 28, 3, 330–​55 deLeon P, 1978, Public policy termination: An end and a beginning, Policy Analysis 4, 3, 369–​92 Demor tain, D, 2011, Scientists and the regulation of r isk, Cheltenham: Edward Elgar Denzau, AT, North, DC, 1994, Shared mental models: Ideologies and institutions, Kyklos 47, 1, 3–​31 Deutsch, KW, 1966, The nerves of government, New York, NY: The Free Press Dolowitz, D, 2009, Learning by observing: surveying the international arena, Policy & Politics, 37, 3, 317–​34 Dror, Y, 2014, Avant-​garde politician: Leaders for a new epoch, Washington, DC: Westphalia Press Dunleavy, P, 1995, Policy disasters: Explaining the UK’s record, Public Policy and Administration 10, 2, 52–​70

16

Definitions, dimensions and intersections

Dunlop, CA, 2014, The possible experts: How epistemic communities negotiate barriers to knowledge use in ecosystems services policy, Environment and Planning C –​Government and Policy 32, 2, 208–​28 Dunlop, CA, 2017, Pathologies of epistemic learning: What are they and how do they contribute to policy failure? Policy & Politics 45, 1,  19–​37 Dunlop, CA, James, O, 2007, Principal–​a gent modelling and learning:  The European Commission, experts and agricultural hormone growth promoters, Public Policy and Administration 22, 4,  403–​22 Dunlop, CA, Radaelli, CM, 2013, Systematizing policy learning: From monolith to dimensions, Political Studies 61, 3, 599–​619 Dunlop, CA, Radaelli, CM, 2016, Policy learning in the Eurozone crisis: Modes, power and functionality, Policy Sciences 49, 2, 107–​24 Dunlop, CA, Radaelli, CM, 2018, The lessons of policy learning: Types, triggers, hindrances and pathologies. Policy & Politics 46, 2, 255–​72 Edelman, M, 1964, The symbolic uses of politics, Chicago, IL: University of Illinois Press Elgström, O, Jönsson, C, 2000, Negotiation in the European Union:  Bargaining or problem-​solving? Journal of European Public Policy 7, 5, 684–​704 Elmore, RF, 1987, Instruments and strategy in public policy, Policy Sciences 7, 1, 174–​86 Evans, M, 2009, New directions in the study of policy transfer, Policy Studies 30, 3, 237–​402 Finch, D, Gardiner, L, 2018, Back in credit? Universal Credit after Budget 2018, London: Resolution Foundation www.resolutionfoundation. org/a​ pp/u ​ ploads/2​ 018/1​ 1/B ​ ack-i​ n-C ​ redit-U ​ C-a​ fter-B ​ udget-2​ 018. pdf Gabriel, R, 1986, Military incompetence, New York, NY: Hill and Wang Gerstenberg, O, Sabel, CF, 2002, Directly-​deliberative polyarchy: An institutional ideal for Europe? in C Joerges, R Dehousse (eds) Good governance in Europe’s integrated markets, Oxford: Oxford University Press, pp 289–​341 Giest, S, 2017, Overcoming the failure of ‘silicon somewheres’: Learning in policy transfer processes, Policy & Politics 45, 1, 39–​54 Gray, P, 1996, Disastrous explanation or explanations of disaster? A reply to Dunleavy, Public Policy and Administration 11, 1, 74–​82 Haas, PM, 1992, Introduction: Epistemic communities and international policy coordination, International Organization 46, 1, 1–​35 Hall, PA, 1981, Great planning disasters, London: Penguin

17

Policy Learning and Policy Failure

Hall, PA, 1993, Policy paradigms, social learning and the state: The case of economic policymaking in Britain, Comparative Politics 25, 3,  275–​96 Hay, C, 2001, The ‘crisis’ of Keynesianism and the rise of neo-​liberalism in Britain: An ideational institutionalist approach, in JL Campbell, OK Pedersen (eds) The rise of neoliberalism and institutional analysis, Princeton, NJ: Princeton University Press, pp 193–​218 HC Deb, 11 February 2019, c593 Universal Credit:  Food Insecurity https://​hansard.parliament.uk/​Commons/​2019-​02-​11/​debates/​ 3AF052EC-​3800-​4B86-​ADDC-​8BE7EC861FAF/​UniversalCredi tFoodInsecurity Heclo, H, 1974, Modern social politics in Britain and Sweden, New Haven, CT: Yale University Press Howlett, M, 2012, The lessons of failure: Learning and blame avoidance in public policy, International Political Science Review 33, 5, 539–​55 Howlett, M, Ramesh, M, Perl, A, 2009, Studying public policy: Policy cycles and policy subsystems, Canada: Oxford University Press Howlett, M, Ramesh, M, Wu, X, 2015, Understanding the persistence of policy failures: The role of politics, governance and uncertainty, Public Policy and Administration 30, 3–​4, 209–​20 Ingram, HM, 1980, Policy failure:  An issue deserving analysis, in DE Mann (ed) Why policies succeed or fail, Beverly Hills, CA: Sage, pp  11–​32 Janis, IL, 1972, Victims of groupthink, Houghton, NY: Mifflin Jones, E, Kelemen, RD, Meunier, S, 2016, Failing forward? The Euro crisis and the incomplete nature of European integration, Comparative Political Studies, 46, 7, 1010–​34 Kahneman, D, 2011, Thinking, fast and slow, New York: Penguin Kamkhaji, J, Radaelli, CM, 2017, Crisis, learning and policy change in the European Union, Journal of European Public Policy, 24, 5,  714–​34 Kay, A, 2017, Policy failures, policy learning and institutional change: The case of Australian health insurance policy change, Policy & Politics 45, 1, 87–​101 Kerr, DH, 1976, The logic of ‘policy’ and successful policies, Policy Sciences 7, 3, 351–​63 Khong, YF, 1992, Analogies at war, Princeton, NJ:  Princeton University Press Langer, EJ, 1989, Mindfulness, Boston, MA: Addison-​Wesley

18

Definitions, dimensions and intersections

Leach, W, Weible, C, Vince, S, Sidikki, S, Calanni, J, 2014, Fostering learning through collaboration:  Knowledge acquisition and belief change in marine aquaculture partnerships, Journal of Public Administration Research and Theory 24, 3, 591–​622 Levy, JS, 1994, Learning and foreign policy: Sweeping the conceptual minefield, International Organization 48, 2, 279–​312 Liberatore, A, 1999, The management of uncertainty:  Learning from Chernobyl, Amsterdam: Gordon and Breach Publishers Lindblom, CE, 1959, The science of ‘muddling through’, Public Administration Review 19, 2, 79–​88 Lindblom, CE, 1965, The intelligence of democracy, New York NY: The Free Press Malley-​Morrison, K, Mercurio, A, Twose, G, 2013, International handbook of peace and reconciliation, New York, NY: Springer March, J, Olsen, JP, 1975, The uncertainty of the past:: Organizational learning under ambiguity, European Journal of Political Research 3, 2,  147–​71 Marier, P, 2009, The power of institutionalized learning:  The uses and practices of commissions to generate policy change, Journal of European Public Policy 16, 8, 1204–​23 Marsh, D, McConnell, A, 2010, Towards a framework for establishing policy success: A reply to Bovens, Public Administration 88, 2, 586–​7 May, PJ, 1992, Policy learning and failure, Journal of Public Policy 12, 4,  331–​54 May, PJ, 2015, Implementation failures revisited:  Policy regime perspectives, Public Policy and Administration 30, 3–​4, 277–​99 McConnell, A, 2010, Policy success, policy failure and grey areas in-​ between, Journal of Public Policy 30, 3, 345–​62 McConnell, A, 2015, What is policy failure? A primer to help navigate the maze, Public Policy and Administration 30, 3–​4, 221–​42 Moran, M, 2001, Not steering but drowning: Policy catastrophes and the regulatory state, Political Quarterly 72, 4, 414–​27 Moyson, S, 2018, Policy learning over a decade or more and the role of interests therein: The European liberalization policy process of Belgian network industries, Public Policy and Administration, 33, 1, 88–​117 Moyson, S, Scholten, P, Weible, C, 2017, Policy learning and policy change: Assessing their relation from different theoretical perspectives, Policy and Society, 36, 2, 161–​77 Nair, S, Howlett, M, 2017, Policy myopia as a source of policy failure: Adaptation and policy learning under deep uncertainty, Policy & Politics 45, 1, 103–​18

19

Policy Learning and Policy Failure

Newman, J, Head, B, 2015, Categories of failure in climate change mitigation policy in Australia, Public Policy and Administration 30, 3–​4,  342–​58 Newman, J, Bird, MG, 2017, British Columbia’s fast ferries and Sydney’s Airport Link:  Partisan barriers to learning from policy failure, Policy & Politics 45, 1, 71–​85 O’Donovan, K, 2017, Policy failure and policy learning: Examining the conditions of learning after disaster. Review of Policy Research 34, 4,  537–​58 Olsen, JP, Peters, BG (eds) 1996, Lessons from experience: Experiential learning in administrative in eight democracies, Oslo:  Scandinavian University Press Owen, D, 2012, The hubris syndrome, London: Methuen Pellizzoni, L, 2001, The myth of the best argument: Power, deliberation and reason, The British Journal of Sociology 52, 1, 59–​86 Pemberton, S, 2009, Policy transfer: A re-​assessment, Policy & Politics 37, 3, Special Issue, 313–​458 Peters, BG, 2015, State failure, governance failure and policy failure:  Exploring the linkages, Public Policy and Administration 30, 3–​4,  261–​76 Pressman, JL, Wildavsky, AB, 1973, Implementation: How great expectations in Washington are dashed in Oakland, Berkeley, CA:  University of California Press Putnam, R, 1993, Making democracy work, Princeton, NJ: Princeton University Press Radaelli, CM, 2009, Measuring policy learning: Regulatory impact assessment in Europe, Journal of European Public Policy 16, 8, 1145–​64 Rodrik, D, 2014, When ideas trump interests: Preferences, worldviews, and policy innovations, Journal of Economic Perspectives 28, 1, 189–​208 Rose, R, 1991, What is lesson-​drawing? Journal of Public Policy 11, 1,  3–​30 Rothberg, RI, 2003, When states fail: Causes and consequences, Princeton, NJ: Princeton University Press Rutter, J, 2010, Policy making:  What worked?, London:  Institute for Government, www.instituteforgovernment.org.uk/​blog/​1274/​ policy-​making-​what-​worked/​ Rutter, J, Marshall, E, Sims, S, 2012, The ‘S’ factors, London: Institute for Government, www.instituteforgovernment.org.uk/s​ ites/d​ efault/​ files/​publications/​The%20S%20Factors.pdf Sabatier, PA, Jenkins-​Smith, HC, 1993, Policy change and learning: An advocacy coalition approach, Boulder, CO: Westview Press

20

Definitions, dimensions and intersections

Schäfer, A, 2006, A new form of governance? Comparing the open method of co-​ordination to multilateral surveillance by the IMF and the OECD, Journal of European Public Policy 13, 1, 70–​88 Schuck, PH, 2014, Why government fails so often: And how it can do better, Princeton, NJ: Princeton University Press Schumpeter, JA, 1942, Capitalism, socialism and democracy, London: Routledge, 1994 Simon, H, 1947, Administrative behavior:  A study of decision-​making processes in administrative organization, New York, NY: The Free Press Stone, D, 2005, Knowledge networks and global policy, in D Stone, S Maxwell (eds) Global knowledge networks and international development, London: Routledge Stone, D, 2017, Understanding the transfer of policy failure: Bricolage, experimentalism and translation, Policy & Politics 45, 1, 55–​70 Taylor, M, 2016, Why policy fails –​and how it might succeed, Royal Society of Arts (RSA) annual lecture 12 September 2016, London: RSA, www.thersa.org/d​ iscover/p​ ublications-a​ nd-a​ rticles/m ​ atthew-t​ aylor-​ blog/​2016/​09/​why-​policy-​fails-​and-​how-​it-​might-​succeed Timmins, S, 2016, Universal credit:  From disaster to recovery? London:  Institute for Government, www.instituteforgovernment. org.uk/​ s ites/​ d efault/​ f iles/​ p ublications/ ​ 5 064%20IFG%20-​ %20Universal%20Credit%20Publication%20WEB%20AW.pdf Trussell Trust, 2018, The next stage of Universal Credit: Moving onto the new benefit system and foodbank use, Trussell Trust www.trusselltrust. org/​wp-​content/​uploads/​sites/​2/​2018/​10/​The-​next-​stage-​of-​ Universal-​Credit-​Report-​Final.pdf Vaughan, D, 1996, The challenger launch decision: Risky technology, culture and deviance at NASA, Chicago, IL: University of Chicago Press Waterman, RH, Meier, KJ, 1998, Principal–​agent models:  An expansion? Journal of Public Administration Research and Theory 8, 2, 173–​202 Weick, KE, 1995, Sensemaking in organizations, Thousand Oaks, CA: Sage Weick, KE, 2009, Making sense of the organization:  The impermanent organization, Vol 2, New York, NY: John Wiley & Sons Weiss, CH, 1977, Research for policy’s sake:  The enlightenment function of social research, Policy Analysis 3, 4, 531–​45 Wildavsky, AB, 1979, Policy as its own cause, in A Wildavsky (ed) The art and craft of policy analysis, Basingstoke: Macmillan, pp 62–​85

21

Policy Learning and Policy Failure

Witting, A, Moyson, S, 2015, Learning in post-​recession framing contests:  Changing UK road policy, in N Schiffino, L Taskin, J Raone, C Doni (eds) Organising after crisis: The challenge of learning, Brussels: Peter Lang Wolf, C, Jr, 1979, A  theory of nonmarket failure:  Framework for implementation analysis, Journal of Law and Economics 22, 1, 107–​39 Wolman, B, 1960, Impact of failure on group cohesiveness, Journal of Social Psychology 51, 2, 409–​18 Wolman, H, 1981, The determinants of program success and failure, Journal of Public Policy 1, 4, 433–​64 Zafonte, M, Sabatier, PA, 1998, Shared beliefs and imposed interdependencies as determinants of ally networks in overlapping subsystems, Journal of Theoretical Politics 10, 4, 473–​505 Zito, AR, Schout, A (eds) 2009, Learning and governance in the EU policy-​making process, Journal of European Public Policy 16, 8, Special Issue, 1103–​1253 Zittoun, P, 2015, Analysing policy failure as an argumentative strategy in the policymaking process: A pragmatist perspective, Public Policy and Administration 30, 3–​4, 243–​60

22

2

Pathologies of policy learning: what are they and how do they contribute to policy failure?1 Claire A. Dunlop

Introduction As Chapter 1 highlights, the literature on policy failure is growing (Dunlop, 2017). A good deal of attention has been focused on defining what we mean by failure. The most complete and best-​used typology comes from McConnell who identifies three main types –​process, programme and political (McConnell, 2010). The failure to link policy ideas to reality is either a failure of: process –​the management of the policy-​making process (for example, examination of policy options, managing experts and stakeholders and commanding legitimacy); programme –​the technical design and implementation of the policy; or, politics –​the distortion of policy ideas for partisan or electoral reasons. This conceptualisation is important; by uncovering the different characters and subjects of failure, scholars can now move on from definition, and concentrate efforts on explaining why policy failure happens and is often repeated. So far, three approaches dominate empirical studies (see Chapter 1): policy stages explanations (for example, implementation analysis); analyses that treat failure as a function of specific political institutions or people (for example, leadership studies); and, analyses examining organisational capacity (for example, policy tool analysis). This chapter is located in the third tradition. Most capacity-​related accounts treat the failure to translate policy ideas into reality as a function of the poor design of policy tools. Here, we take a different tack using theories of policy learning to understand policy failure; where failure is treated as a ‘degeneration’ of policy learning. Analytically, the chapter drills down on the rational ideal type of policy learning –​epistemic learning. This is the realm of evidence-​based policy making (EBPM), where experts advise

23

Policy Learning and Policy Failure

decision-​makers on issues of technical complexity (see Ingold and Monaghan, 2016 for a recent discussion). Empirically, the chapter presents a policy failure whose causes are rooted in the process of policy making. Specifically, the management of bovine tuberculosis (BTB) in England since 1997 is conceptualised as a failure of epistemic learning –​where learning processes degenerated as the result of various weaknesses in government’s management of its relationship with an epistemic community established to advise it. Drawing on documentary evidence and interviews with 54 elites (scientists, policy makers and interest group actors),2 management failures in BTB are analysed as problems of learning about different aspects of organisational capacity. Why take this approach? Why link failure with learning? Studies of policy learning have long focused on the benefits associated with this activity. But, learning may not always be a good thing. It may backfire. Linking policy learning with policy failure allows us to illuminate this darker aspect; where enlightenment becomes ‘endarkenment’ (Weiss, 1979). Why focus on policy learning about organisational capacity? We know a good deal about the ideational dimension of policy learning; indeed it is the foundational theme of classic studies exploring the link between beliefs and social learning –​for example, Deutsch, 1966; Heclo, 1974; and more recently Béland and Cox, 2011; Hall, 1993). The governance dimension –​that links learning with organisational capacity development –​has enjoyed less widespread attention in policy studies (Borrás, 2011). By making these connections, we illuminate the relationship between governance failure and learning (Peters, 2015; Schout, 2009), and specifically how capacities can be generated to avoid or correct failures. The chapter is structured as follows. The first section outlines BTB management as a case of policy failure and, more specifically, as a failure of epistemic learning in government. It then outlines a recent learning typology where the epistemic form is differentiated from three other types. The second section zooms in on epistemic learning, expanding the concept to identify four different roles epistemic communities play when advising decision-​makers and pathologies associated with each. With these possible roles established, the third section builds on the organisational literature to outline the capacity challenges faced by decision-​makers engaged in epistemic learning and the ways in which advisory relationships can go wrong and learning can degenerate. These degenerations are understood as rooted in failures in government’s organisational capacities. Empirically, our analysis of BTB policy in England finds that epistemic learning degenerated as a result of weaknesses in the government’s analytical and communicative

24

Pathologies of policy learning

capacities. We conclude with some reflections on the value of learning theories as a conceptual lens for policy failure.

Conceptualising bovine tuberculosis (BTB) management as a degeneration of epistemic learning This section presents the management of bovine tuberculosis (BTB) since 1997 as a policy failure. Specifically, it is categorised as a failure of epistemic learning. Before offering a conceptual elaboration, we first present the empirical story. In what ways has the management of BTB failed? Bovine tuberculosis (BTB) has affected cattle in large parts of England since the 1950s. Despite being declared BTB-​free in the early 1970s, the number of infected herds rose gradually until the 1980s. Since then, BTB incidence has spread geographically and increased sharply –​in 1986, 638 cattle were slaughtered; by 2004 this figure was around 22,500 (effectively doubling every four years). The costs of managing the disease are considerable. For example, in 2002/​03 the Department for  Environment, Food and Rural Affairs (DEFRA) spent £73 million –​£31 million in compensation for famers with affected herds, £29 million on BTB testing and associated veterinary costs and £13 million on disease research. Control of BTB is problematic due to the existence of a wildlife reservoir for the disease. In the UK, badgers are commonly infected with TB, and through their roaming across farmland, contribute to the spread of infection in cattle. By the 1990s research suggested that, while the evidence implicating the badger was ‘compelling’ (Krebs et al, 1997), for an effective policy response research was required to reveal the magnitude of badgers’ contribution to the disease and to explore the effectiveness of different badger culling strategies. In 1998, the Randomised Badger Culling Trial (RBCT) was initiated to provide data on the three types of culling  –​proactive, reactive and no culling –​for a cost–​benefit analysis. Given the strong public attachment to the badger and powerful animal welfare lobby in the UK, the RBCT was restricted to culling 30 per cent –​no future policy could be based on the animal’s ‘virtual elimination’ (ISG, 1998; 1999). The expectation was that the experiment would take around five years to complete, and that a proactive culling would be the best option. When the ISG finally reported back in 2007, it could not have been clearer. The trial data and cost–​benefit analysis suggested that badger culling ‘cannot meaningfully contribute towards future control of BTB’ (ISG, 2007). The RBCT uncovered new evidence about badger

25

Policy Learning and Policy Failure

ecology. As badger colonies were disrupted by culling, a ‘perturbation effect’ took place with displaced badgers that had escaped the cull moving away from their territory and re-​colonising –​taking the disease with them to other cattle herds. Given this, reactive culling had no discernible benefits, and the positive effects of proactive culling were largely offset by a negative ‘edge effect’ where incidence of cattle infection increased at the periphery of the culling zones (the net result was a saving of only 14 cattle herds over five years). For culling to be effective, it would have to be on a scale and intensity that was socially and politically unpalatable, practically unfeasible (given badger movements) and economically inefficient. Instead, the ISG argued that future policy should focus on cattle-​to-​cattle transmission. Despite having created the bedrock of a new international scientific consensus on badger ecology, the ISG’s conclusions were contested by a new group of scientists commissioned by DEFRA (King, 2007). While these arguments raged, governments’ approach to BTB has been to continue cattle testing, movement restriction and compensated slaughter scheme. In 2012 alone, the cattle slaughter figure exceeded 31,000, costing nearly £100 million. In the years since the ISG’s report, successive administrations have struggled to develop a national control policy for BTB. The latest development in 2013 was the beginning of a four-​year cull targeting ‘at least’ 70 per cent of badgers in TB hotspots. This policy has been highly controversial and, so far, unsuccessful. The first tranche achieved only 30 per cent removal rate, leading the Secretary of State to famously declare that ‘the badgers had moved the goalposts’! (BBC, 2013). How can we understand this failure? What is it a failure of? The failure to control BTB can be treated as a failure of epistemic learning. After ten years of experimentation on BTB management which yielded a clear result, decision-​makers are left with the same fundamental policy challenge (see McConnell, 2010 on failure indicators).To be clear, this is not a normative argument that decision-​makers should have followed the ISG’s conclusions. Rather, analytically, the proposition is that the policy limbo and problem escalation since the 1970s, and in particular since 1997, is in large part the consequence of pathologies in how scientific advice has been commissioned and managed by government. To explore policy failure in BTB policy making as a degeneration or pathology of epistemic learning, we first need to be clear about what we mean by policy learning. The policy learning literature is vast and reveals a variety of types. We distil this scholarship to define learning as updating of beliefs (Dunlop and Radaelli, 2013). This matches the central concern of all policy analysis –​the study of how beliefs inform

26

Pathologies of policy learning

policy debates, content, performance, institutional structures and, on occasion, change. Beliefs are predominantly updated through social interaction, appraisals of one’s experience or evidence-​based analysis. Dunlop and Radaelli (2013) develop a four-​fold learning typology, based on a review of this literature. Specifically, the proposal is that the type of policy learning decision-​makers engage in is a product of two conditions associated with decision-​making around knowledge-​ dense issues. The first of these concerns the issue’s level of problem tractability. Where technical uncertainty is extreme and issue intractable, decision-​ makers want authoritative advice –​classically in the form of epistemic communities (Haas, 1992).The second dimension concerns the certification of actors:  that is the extent to which a group of experts exists to advise policy makers on the issue at hand. These experts will hold consensual knowledge and their ‘performances and claims’ to expertise validated by the state (McAdam et al, 2001, 121).Where no single certified group exists or one has been discredited, epistemic authority can be, and expertise may be, localised or plural. Taken together, levels of issue tractability and actor certification provide the basic conditions for four modes of policy learning that dominate the social sciences literature (see Figure  2.1):  epistemic learning, Figure 2.1: Conceptualising knowledge modes as policy learning PROBLEM TRACTABILITY Low

High

Low 2. Reflexive learning

3. Learning through bargaining

1. Epistemic learning

4. Learning in the shadow of hierarchy

CERTIFICATION OF ACTORS

High

Source: adapted from Dunlop and Radaelli, 2013, Figure 1

27

Policy Learning and Policy Failure

reflexive learning, learning through bargaining and learning in the shadow hierarchy. These four have been outlined in detail elsewhere (Dunlop, 2014; Dunlop and Radaelli, 2016). Our interest is in epistemic learning and the BTB case meets the basic conditions for this –​its management has long been the exclusive domain of scientists and it is marked by low tractability. Before exploring why things ‘go wrong’ in epistemic learning, the next section unpacks this learning type into four modes and the dysfunctional forms they can take.

Unpacking epistemic learning Using a theory of adult learning, the property space of each of the four types of learning can be expanded (Dunlop and Radaelli, 2013). By differentiating between instances where the decision-​maker (that is, the learner) focuses on the contents or objectives of knowledge creation, we capture the variety of roles inhabited by expert groups in epistemic learning settings. This differentiation is essential if we are to understand the different functions they may perform over the lifespan of an issue, and the variety of ways in which learning can succeed or degenerate. We generate four roles for each cell of the basic model. Figure 2.2 contains those for epistemic communities (see Dunlop and Radaelli, 2013, 608–​10). The role of ‘teacher’ is the ideal type role for an epistemic community. Here, the issue is at its least tractable and experts their most authoritative. The challenge for decision-​makers is to absorb as much knowledge as possible to get a basic grip on the main dimensions of the problem. The literature suggests this ideal type category is time-​limited –​decision-​ makers do not want to be positioned behind a heavy veil of ignorance for too long. Moreover, the aim of the epistemic teacher is to help decision-​makers identify their preferences. Epistemic communities make effective teachers where decision-​makers are able to listen to these communities but not be captured by them. In the worst-​case scenario, where decision-​makers lose their analytical distance from, or are baffled by, expert advisers we risk ‘groupthink’ (Janis, 1972) where decision-​ makers fail to ask critical questions and are left unable to innovate. At the opposite pole is the ‘contributor’ –​the weakest role for experts generating epistemic learning. In these situations, decision-​makers have grasped the direction they want policy to go in and the evidential basis required to reach that destination. Thus, the issue is open to knowledge forms beyond those of the epistemic teacher  –​most notably from wider society. Here, the epistemic community becomes an efficiency

28

Pathologies of policy learning Figure 2.2: Expanding epistemic learning DECISION-MAKERS’ FOCUS ON LEARNING OBJECTIVES

High

DECISION-MAKERS’ FOCUS ON LEARNING CONTENT Low

High

Low

Epistemic community as CONTRIBUTOR

Epistemic community as PRODUCER OF STANDARDS

Degeneration = knowledge debased

Degeneration = production of low-quality knowledge

Epistemic community as FACILITATOR

Epistemic community as TEACHER

Degeneration = knowledge politicisation

Degeneration = groupthink

Source: adapted from Dunlop and Radaelli, 2013, Figure 4

device. At best, they will form the basis of, and make a continued contribution to, the emerging policy paradigm. At worst, they serve a perfunctory purpose –​for example, as a symbol that evidence has been taken seriously. In extreme cases, the epistemic community becomes marginalised and its knowledge debased. Where expert groups are ‘facilitators’ of epistemic learning, decision-​ makers are aware of the policy goal but require evaluation evidence to help them reach it. Here, the epistemic community’s role is to provide decision-​makers with that policy-​relevant evidence. This degenerates where the policy objective begins to distort or politicise the research process  –​for example, data analysis is driven by policy rather than epistemic concerns or evidence cherry-​picked by decision-​makers. The last quadrant is the epistemic community as a ‘producer of standards’. This concerns situations where decision-​makers want to facilitate the production of robust evidence. Through its authoritative knowledge, the epistemic community determines what standards are appropriate in a policy area. Where this degenerates, decision-​ makers fail to provide the resources for the production of high-​quality knowledge and robust standards. It is important that we are clear about the status of decision-​maker/​ epistemic community relationships. For, no matter what role the epistemic community inhabits at a particular moment in an issue’s lifecycle, the advisory process is a form of informal governance (Christiansen and Piattoni, 2003). When decision-​makers call upon

29

Policy Learning and Policy Failure

experts, they are not making formal delegations of authority. They are working with knowledge producers whose professional training, institutional identities and values derive from their disciplines or specialist organisations. Thus, for epistemic learning to be successful, decision-​makers must learn how to manage their experts. The next section explores the specific organisational capacity challenges associated with each epistemic mode and examines how they played out in the BTB case.

Pathologies of epistemic learning and organisational capacity in BTB management So, decision-​makers and epistemic communities are involved in complex relationships where they inhabit multiple roles across an issue’s lifespan. The management of those relationships is critical to the success of epistemic learning and resultant policy. To keep learning pathologies at bay, governments require various types of organisational capacity (Peters, 2015). Organisational capacity addresses the central concern of governance –​the ability to develop the skills and tools required to achieve policy goals (Howlett et al, 2015a). In technically complex policy dilemmas marked by uncertainty, epistemic communities represent the key policy partner. Yet, we must be realistic about what such partnerships can yield. Certainly, the epistemic communities’ literature is replete with examples of the limited, or partial, influence of expertise on policy action and less still on actual outcomes (see Dunlop, 2013 for a review). What we lack, however, is a systematic way of identifying and analysing the aspects of governance that frustrate epistemic learning and threaten policy failure. Focusing on the link between organisational capacity and epistemic learning offers one promising analytical strategy. Policy learning around an issue takes place in organisations marked by different types of capacity. Empirical studies on advisory relationships suggest that breakdowns, or weaknesses, in these capacities can create a breeding ground for pathogens that undermine learning. Drawing on accounts of policy learning and capacity in public administration (notably, Bennett and Howlett, 1992; Borrás, 2011; Howlett et  al, 2015a) here we propose four types of capacity that support epistemic learning; explore how weaknesses in them may contribute to the degeneration of it, and how they relate to the variety of roles inhabited by governments and their epistemic advisers (see Table  2.1 for a summary). To be clear, this typology does not aim to offer hard and fast prescriptions for action. Rather, we offer a set of capacity types

30

Pathologies of policy learning Table 2.1: Organisational capacities and epistemic learning degeneration Epistemic Organisational communities capacity as… approached as…

Learning about…

Government focuses on…

Learning degeneration as…

Teacher

Absorption (ACAP)

Listening

Understanding expert advice

Groupthink

Producer of standards

Administrative (ADCAP)

Management tools

Supporting robust evidence creation

Failure to produce robust knowledge

Facilitator

Analytical (ANCAP)

Analytical tools

Defining policy relevance

Politicisation of research process

Contributor

Communicative (COMCAP)

Dialogic tools

Achieving stable paradigms

Debasing evidence-​ based policy paradigm

Source: author

based on public administration concepts, elaborated with reference to empirical studies, that authors can use to critically examine advisory relationships. Further empirical testing over time will lead to a clearer view of how useful and accurate these conceptualisations are. But, for now, they provide us with a much-​needed analytical starting point. Absorptive capacity Absorptive capacity concerns the ability of decision-​makers to acquire knowledge from experts to help reduce uncertainty. Specifically, we are dealing with the ability to listen to, understand and, if necessary, challenge, epistemic communities. The challenge for decision-​makers is to avoid uncritical engagement that characterises ‘groupthink’ (Janis, 1972) –​where fear of being thought inferior or conflictual results in decision-​makers not asking for clarification from their advisors or ignoring warning signs that what experts are saying may not meet expectations. As shown by Hirschman (1970), paradigms provide their own blinkers and hindrances to learning, at a time when thinking outside the box is vital. For decision-​makers to think outside the box they must first understand what is being said in it. The idea of absorptive capacity stems from the management literature. Three skills recur in that literature (see Zaha and George, 2002).The first involves listening through regular information gathering. Decision-​ makers suffer from information overload, and struggle to know which

31

Policy Learning and Policy Failure

messages to focus on (Jones and Baumgartner, 2005). Giving epistemic communities advisory positions inside the bureaucracy is a key part of ensuring key information is highlighted (Haas, 1992). The second dimension of knowledge absorption involves the source of information. Decision-​makers working with one set of expert advisors must develop their peripheral vision to ensure that they are alive to any significant criticisms. Generating ways of capturing misgivings on the part of other experts or rival epistemic communities is central to the avoidance of blinkered thinking. Yet, being able to hear the voices of experts and their critics is just the start. Decision-​makers must be able to understand as well as listen. In short, decision-​makers in governments (like managers in a business) must react and, where necessary, adapt the original management (policy) strategy in the face of the information updates they receive (Argyris and Schön, 1978). This third skill of gauging the speed and intensity of reaction is incredibly difficult. Indeed, the knowledge utilisation literature is replete with examples where decision-​makers do not change policy course in the face of evidence that their favoured policy solutions may not match the lessons being taught by experts (for a recent review see Turnpenny et al, 2014). Understanding how knowledge may ‘play out’ in policy terms and adapting accordingly is politically risky in all issues. But, in technically complex issues it is a high wire act indeed since decision-​makers are continually uncertain about how complete the knowledge is to which they are reacting (Dunlop, 2010). What was the absorptive capacity in BTB’s advisory governance? Throughout the research process, the government demonstrated considerable listening skills. DEFRA provided the ISG with a secretariat which played a key role in keeping open lines of communication between the scientists and civil servants. Indeed, one of the conditions of the scientists’ work was that they did not meet to discuss the experiment without their assigned civil servants (interviews with two ISG scientists and one civil servant). The ISG chair also ensured that their message reached the key decision-​maker. Following the advice of a senior government scientist, Professor John Bourne insisted that any formal reports or urgent findings be communicated by him to the Minister direct. This insistence was motivated, in part, by the concern that politically sensitive messages could become lost ‘in the system’ and the group become a scapegoat in a highly politicised issue (interviews with two ISG members and one government Minister). And, there is evidence of success in this regard –​in 2003 when the ISG shared alarming data with officials that the first round of reactive culling had

32

Pathologies of policy learning

led to an increase in the risk of BTB infections (27 per cent), the Minister acted swiftly to halt that intervention. This was something the scientists regretted (interview with three ISG members), but it does demonstrate a Ministry in ‘listening mode’. DEFRA also ensured their epistemic community’s work was open to scientific scrutiny. Each of the ISG’s annual reports were audited by an independent statistician and given a clean bill of health. Scrutiny also went beyond this routine. With the trial taking longer than expected and reactive culling having been suspended, in 2003 the Agriculture Minister ordered an interim review to be conducted by a group of veterinary scientists chaired by Professor Charles Godfray (Godfray et al, 2004).The work was also peer-​reviewed by the wider scientific community –​the ISG being free to establish the credibility of their work by publishing results as they came in international scientific journals (including Nature and The American Proceedings of the National Academy of Sciences). As outlined here, absorptive capacity requires more than information gathering and scrutiny. To truly absorb the information being given by their expert advisors, decision-​makers must critically engage. In policy learning terms, this is demonstrated by the adaptation of policy design. In the BTB case, this did not happen. At the outset, in 1997, no one (scientists included) considered that small-​scale badger culling would be rejected as a viable strategy. Rather the issue was one of what type of culling worked best. By 2004, both the ISG and Godfray review group warned decision-​makers that badger culling was unlikely to provide a viable policy solution and that cattle control and, in the long term, a vaccine should be explored. Yet, no alternative policy plans were developed at this point. As noted earlier, a key part of the absorption challenge concerns gauging the speed and intensity of reactions to information. As the BTB case demonstrates, where policy disputes are politicised and costs sunk around one policy solution, decision-​makers’ responses may be slow and uncertain, leaving policies in limbo. Administrative capacity Administrative capacity concerns establishing and managing resources to ensure the development of robust evidence. Three sets of operational tools are central to ensuring epistemic communities produce standards for an issue. First, and at the most basic level, administrative capacity concerns bureaucrats’ freedom to act. Decision-​makers’ legal obligations, and historic policy legacies, shape the room for manoeuvre and ability for epistemic learning to reach its full conclusion. Such institutional

33

Policy Learning and Policy Failure

hierarchies may work in favour of epistemic learning where regulations require that policy conforms to particular scientific standards (Weiss et al, 2005). Moreover, where power has been ceded to another level of government, or is shared with another government department or agency, bureaucrats’ ability to manage advisory relationships and act on expert advice may be complicated or compromised. Second, robust advisory science requires financial capacity; most obviously the ability to calculate and make available the requisite resources to fund research to its end point. Of course, the outlay depends on the type of knowledge required (Davies et  al, 2000). Evidence reviews, or small-​scale studies to plug empirical gaps, are relatively cheap when compared with the production of original data (a requirement most associated with situations of radical uncertainty). Moreover, decision-​makers must appreciate that their financial commitments to an epistemic community charged with creating original data may change as that research progresses. Third, there is the matter of expert recruitment and management. To secure authoritative knowledge, epistemic advisors must be able to produce the science. Moreover, while experts must understand the policy context, the management challenge for bureaucrats is to ensure that they are able to conduct their research away from the political fray. Securing and insulating this expertise is harder than it may sound. In advisory relationships, research methods function as indicators of impartiality (Stoker, 2010). For example, bureaucrats are increasingly turning to large-​scale quantitative research and randomised control trials (RCTs) because they offer statistical checks of validity and produce evidence that can be falsified by the wider expert community (Haynes et al, 2012). Allied to this, additional external experts may be recruited to perform a challenge function to check research quality. In BTB, government supported evidence creation to identify the role of the badger in the disease and identify a suitable culling strategy. DEFRA had a high degree of freedom to organise its advisory relationship. Though the European Union (EU) requires that member states deal with animal disease, how they do this is a domestic matter. Moreover, coordination problems are minimal within government; BTB involves no other government department. DEFRA used this capacity to take a bold step and commissioned the first ever randomised controlled field trial to be funded by the UK government (interview one civil servant). The role of badgers in BTB has been the subject of government-​funded science since the late 1970s. But, these were observational studies and, with infection rates continuing to rise in the 1990s, decision-​makers were convinced that only an RCT that

34

Pathologies of policy learning

compared different culling strategies would produce the robust evidence needed to guide a future cull (Krebs et al, 1997). The selection of an RCT was not simply a reflection that decision-​ makers believed previous methods had produced partial knowledge. It also reflected the evidence-​based policy-​making approach being championed by the incoming New Labour administration whose Agriculture Minister emphasised the need for policy underpinned by ‘rigorous and apolitical’ science (interview with Minister).This belief that RCTs have the explanatory power to depoliticise an issue and cut through value-​based arguments is central to the increasing appeal of this method in government in general and in BTB in particular (Dunlop, 2016). And so, it was financially well supported. That this would be the largest ever field trial conducted by government was known from the outset, and scientists furnished with the requisite resources (interviews with five ISG members). The scale of financial commitment was huge, and a two-​year delay due to foot and mouth disease plus the need for additional training of field operatives in culling techniques all contributed to an escalation of costs. An estimated £7 million a year was spent on the experiment for the ten years it ran; yet there was no suggestion that at any point the government were unwilling to commit resources. Similarly, there were few administrative capacity problems in terms of the selection of the scientists themselves. The ISG was assembled through recommendations made in 1996 by then government Chief Scientific Officer Professor Sir John Krebs and other scientific officers working inside government. Most of the groups’ seven members had previously advised DEFRA (and its predecessor Ministry of Agriculture, Fisheries and Food (MAFF)) and three had been part of the Krebs group. All were research-​active, full-​time academic scientists, and one of them counted a Nobel among their professional honours. While open-​minded about the results, these scientists shared the expectation of policy makers and Krebs that a culling solution would result from the trial even on its small scale, with proactive culling expected to be the best solution. What was not expected was the ‘perturbation effect’ which provided negative results from the scale of culling conducted. Until the publication of the final report, the ISG scientists were considered independent by the opposing sides in the debate (interviews with NFU and Badger Trust officials).While they did not enter the policy debate raging around them, the ISG did conduct public meetings to explain the trial design and its rationale. The scientists needed the cooperation of the affected communities where the culling pilots were taking place. But, the research process was not compromised

35

Policy Learning and Policy Failure

by this  –​indeed without such engagement it is unlikely that the scientists would have been granted the access to land required for the experiment to work. Moreover, as was noted earlier (the subsection on absorptive capacity), the scientists published their work widely and were scrutinised by a DEFRA appointed auditor. Moreover, though they were critical of the policy relevance of the findings, neither the Godfray nor the King reviews criticised the ISG’s science. Indeed, both sets of scientists were at pains to emphasise how robust it was (EFRAC, 2008). Thus, this dimension of epistemic learning did not degenerate. On the contrary, the operational arrangements put in place by DEFRA supported the production of a new international scientific consensus on badger ‘perturbation’. That localised badger culling continues and evaluations of that policy support the ISG’s argument, underlines that success in one area of epistemic learning is not enough to prevent policy failure. Analytical capacity Analytical capacity concerns ensuring the relevance of the knowledge being produced for policy –​that is, identifying what works. In highly technical issues where epistemic communities dominate, in-​house civil service analysts are not required to produce knowledge. Rather, their task is to work with epistemic communities to define the set of problems about which they want to learn something, and the set of solutions that are viable. The challenge is to translate this to highly specialised experts. Experts can only facilitate policy if they understand the parameters of what is policy-​relevant. This is where decision-​makers come in. If they want to receive knowledge which can contribute constructively to policy design, they must advise their experts on what policy relevance actually means in the case at hand and update them when this changes (Lindblom and Cohen, 1979; for a recent empirical example see Dunlop, 2010 on the role of the Chief Scientific Adviser in UK departments). Government is in a potentially vulnerable position in its relationships with epistemic advisers. While issue uncertainty means its grip on a policy is weaker than it would like, ultimately government still remains responsible for the production of policy-​relevant knowledge. As scientific experts possess an informational advantage, procuring scientific knowledge in some respects creates even more uncertainty for decision-​makers. In such circumstances, analytical capacity is a function of decision-​makers’ ability to determine rules to oversee

36

Pathologies of policy learning

their epistemic advisors. It is through these rules, that decision-​makers keep knowledge production relevant. Drawing on principal–​agent modelling, three particular challenges must be met (see Dunlop and James, 2007 for an empirical application of these). First, decision-​makers must recognise that, when they engage in epistemic learning they are involved in potentially non-​hierarchical relationships. To maximise the chances of getting policy-​relevant analysis, bureaucrats must set clear parameters for experts’ work (Guston, 1996). Most obviously, this can be done by setting an exact mandate and research question to be addressed, and advising epistemic actors on their precise role in policy making –​for example, are policy options to be analysed or suggested by these experts? Second, decision-​makers should select their advisers carefully. While we previously discussed the administrative matter of selecting scientists capable of conducting the research, the analytical challenge for bureaucrats is to ensure their experts’ normative or policy views do not clash with those of government (Verdun, 1999).To procure analysis that is appropriate for policy, bureaucrats must be aware of an epistemic community’s normative beliefs and the institutional affiliations of its members before insinuating them into the policy process. This is a fine line to tread, however. Where selection is adverse the risk is advisory knowledge which is not relevant to the policy goal, and also, where the experts’ normative views do not clash with decision-​makers’ preferences, of charges of bias. In both instances, the research process becomes politicised and ideas produced to facilitate policy action become the source of further policy instability. The final challenge of analytical capacity is to develop management techniques that guard against unwanted behaviour by the epistemic community (Guston, 1996). Such ‘moral hazards’ are avoided through the use of monitoring and reporting agreements (other common management tools like incentives and penalties can have little role in knowledge procurement). Where monitoring is systematised, it is commonly handled by a dedicated (and ideally stable) secretariat that services the expert group; the art here is to keep track in a way that makes intervention from the policy side possible but allows experts to engage in technical discussions freely. Striking the right balance here is critical to maintaining the functional inter-​personal relationships necessary for effective communication on policy relevance. On reporting, government’s need for control over the content of an expert group’s findings is obvious. While decision-​makers cannot write experts’ reports for them, healthy analytical capacity requires that bureaucrats are able to question and comment on early drafts

37

Policy Learning and Policy Failure

and contribute to the iterative development of conclusions. Such involvement in the evolution of an epistemic community’s findings represents a critical learning mechanism  –​as bureaucrats are able to explore how the evidence and different policy options might fit together (Dunlop and James, 2007). Clarity on how the epistemic knowledge will play out in policy terms may also require dialogue with the wider policy network and stakeholders –​where early findings of an expert group are shared. At its inception the ISG’s policy literacy was strong. It was given autonomy to execute the RBCT, but the policy context in which it was conducted was made plain –​any future policy could involve only 30 per cent of badgers being killed and that must be done humanely. So, DEFRA set the key parameter for how they answered it –​trial areas limited to 100 km2 and culling performed by trapping with closed seasons to ensure that no cubs could die underground. The selection of experts also augured well. The ISG scientists were uncontroversial; these were academic scientists with no discernible conflicts of interest. But, experts and their views do not come from ‘nowhere’. As Haas notes they have normative and policy ideas (1992). Beyond a normative commitment to positivism in knowledge creation, some members were willing to disclose their strong desire to, and assumption that they would, find a culling solution for the farming communities of which they were a part (interviews with three ISG members). And so, there was nothing out of tune between the scientists and the decision-​makers’ policy commitments and beliefs. The policy dimensions of the ISG’s work were also clear. In addition to the political parameters on the scale of culling that would be acceptable, policy makers also requested that the ISG assess the cost-​effectiveness of the culling options and explore the epidemiology of cattle-​to-​cattle transmission. As, however, the findings on the ‘perturbation effect’ began to accumulate from 2005, decision-​makers struggled to exercise control over the experiment’s findings. Specifically, the absence of any well-​ developed ‘plan B’ (see the subsection on absorptive capacity) left the ISG without any guidance on the relevance of this unexpected evidence. Trust began to break down between the scientists and DEFRA (interviews with three ISG members and two Ministers). These strains were crystallised in the 2005 public consultation which presented localised culling options to the public despite the ISG’s counsel that they would not work. The ISG’s unequivocal conclusions of 2007  –​that badger culling on the scale envisaged as politically acceptable could not ‘meaningfully contribute towards future control of BTB’ (ISG, 2007) –​put DEFRA’s preferred policy under threat.

38

Pathologies of policy learning

Rather than accept the ISG’s conclusions, a further group of scientists were commissioned to review the study. Led by Government Chief Scientist  –​Professor Sir David King  –​the King group argued that though the ISG’s work was robust, in crucial respects it was irrelevant in policy terms; its small scale made its findings too niche. Citing the success of large-​scale badger culling in Ireland, King suggested that positive results would be achieved on trial sites of 300 km2 (King, 2007). While acknowledging that implementing badger culling areas of 300 km2 would mean culling around 70–​80 per cent of the badger population (King, in EFRAC, 2008, Ev90), the group did not explore the practicalities of that –​in either political or logistical terms (EFRAC, 2008, Ev 91–​2). Rather, they emphasised that ‘these are issues that can be addressed by officials’ (King in EFRAC, 2008, Ev88). This sub-​text –​that the ISG and results had become politicised and left to determine what policy was relevant in this case –​had been anticipated by Godfray back in 2004: the ISG have borne too heavy a responsibility for the running of these projects, and links between policy formulation by DEFRA and the scientific input from the ISG have not been as seamless as would be desirable. In designing future projects of this size we would recommend that the essential independent scientific group has a less direct management role. (Godfray et al, 2004, 6) While we could argue that the ISG’s research was policy-​relevant in one respect –​it demonstrated that small-​scale culling on DEFRA’s terms could be counter-​productive –​fundamentally it left decision-​ makers without a culling policy option backed by science. This is not only the failure to absorb the ISG’s evidence and create a plan B. It is also suggests weakness of analytical capacity and purposeful politicisation. Though DEFRA was aware in 2005 of the mounting evidence against culling, it did not intervene in the research or discuss alternatives with the ISG (interviews with DEFRA officials and ISG scientists). Nor did decision-​makers work with its key stakeholder –​ the NFU –​to explore the implications of the findings in advance of their publication. Indeed, such was the NFU’s surprise at the ISG’s conclusions that it suggested that its final report had ‘been slanted at the last moment to…give a definite steer against culling’ (NFU Policy Director, Martin Haworth, in EFRAC, 2008, Ev29–​30, 33, emphasis added). This led the NFU to threaten to withdraw cooperation on a range of other agricultural issues until culling was put back on the

39

Policy Learning and Policy Failure

policy agenda. The result was the creation of a counter-​epistemic community that re-​imagined the policy parameters and rendered the ISG’s conclusions irrelevant. Indeed, this re-​imagining drives the current localised culling policy. Failure to redefine policy relevance in the face of unanticipated scientific findings is not a new phenomenon (see Dunlop, 2010 on biofuels, for example). The consequences of not developing and communicating a ‘plan B’ to the scientists was politicisation of the advisory process. The ISG appeared out of touch; a group producing epistemic lessons without any real-​world application (something anticipated by Godfray in 2004). The gap between policy (expectations) and the evolving evidence was filled by additional scientists and stakeholders; that is, moving away from epistemic learning. We cannot prove our counterfactual of course, but much of the testimony of civil servants, scientists and interest groups at the BTB inquiry (EFRAC, 2008) suggests that reconfiguring the policy options in light of the epistemic lessons on perturbation in 2004–​05 would have boosted the chances that the gold standard science was also useable, and narrowed the space that was filled by contestation. Communicative capacity Communicative capacity concerns the capacity to develop and maintain a policy paradigm that commands approval and legitimacy in wider society (for comprehensive review see Bartels, 2015). This goes beyond the key policy stakeholders to relations at the macro-​level. Communicative capacity in epistemic learning involves developing a social consensus around what an evidence-​based approach in a given sector entails; in short, agreeing what evidence is legitimate. What citizens define as evidence, and the extent to which that evidence should determine policy, is often contested. Where such a paradigm cannot be constructed and communicative capacity is weak, epistemic learning will also be weak and policy decisions based on evidence liable to instability. Thus, the stakes are high, and government highly motivated to achieve broad agreement. The default technology used to boost communicative capacity and involve citizens in the advisory process is the consultation (LSE GV314 Group, 2012). Yet, these exercises are often criticised as being insensitive to local differences –​ where particular populations’ voices are over-​or under-​represented. Moreover, consultations present citizens with pre-​selected policy options. For some, more capacity is generated when engagement is developed ‘upstream’ in the policy process where citizens, scientists and

40

Pathologies of policy learning

policy makers come together to discuss the policy problem before any solutions have been formulated (Wilsdon and Willis, 2004). BTB was not marked by any radical ‘upstream’ policy making involving citizens. Rather, the ISG and its work became important symbols of the rational evidence-​based policy-​making (EBPM) paradigm which came to prominence in the late 1990s. While a scientific approach was not new in BTB policy making, the nature of the ISG and its research did mark a shift. Though it was contentious among stakeholders from different sides for different reasons, the creation of the ISG embodied the first New Labour administration’s aspirations for the modernisation of policy making in the UK (Sanderson, 2006). Specifically, the decision to make the group independent of DEFRA, and the choice of an RCT, were emblematic that the best of best practice would be used to find the ‘definitive answer’ (Krebs et al, 1997) to contentious policy problems like BTB. The commitment to a high scientific approach to BTB was sincere, and citizens’ approval assumed. When the experiment was commissioned, policy makers in DEFRA and their Ministers believed this would yield the silver bullet that had eluded their predecessors for nearly two decades (interview with two Ministers and one NFU official). Commissioning such a large-​scale RCT was seen by some in the policy community in the European Commission and Ireland as an over-​rationalisation of the problem (interview with DG Agriculture and DAFM officials). But, throughout the RBCT, policy makers defended the research. Notably, they were willing to highlight the difference between its evidence-​based approach with the ‘farmer-​ driven approach’ of the Republic of Ireland (interview with RoI scientist and policy maker). Given the unexpected length of the study, the Minister was keen to explore whether the social parameters on badger culling, which had been based on various public opinion surveys and presented in 1997 as a ‘red line’ that could not be crossed, were still as restrictive. The public consultation in 2005 yielded over 35,000 submissions and, unsurprisingly, no consensus. Despite the ISG being clear that reactive and proactive culling would be unworkable, the different culling options were presented to the public. The ensuing public row between DEFRA and the ISG laid bare the inherent nebulosity of EBPM (Sanderson, 2006); what counts as evidence is politically and socially contested. Any hope that EBPM could represent a stable paradigm for policy making in controversial issues like BTB is undermined by DEFRA’s response to the ISG’s final report. By mobilising a counter-​epistemic

41

Policy Learning and Policy Failure

community –​the King group –​the government undermined the idea that a consensus could be reached on what knowledge matters for BTB. The debate and policy action since 2007 reflect this instability with the current Conservative-​led coalition defending localised badger culls as evidence-​based against scientists and animal welfare lobbyists that cite the work of the ISG. While this is presented as a degeneration of epistemic learning, it is pertinent to ask whether EBPM is a paradigm that can actually be debased. It appears to be a ‘vehicular’ idea that is broad enough to have elements of legitimacy and illegitimacy for everyone. Such nebulosity placates wider society and closes off debate (McLennan, 2004) –​ as society engages in a dialogue of the deaf –​but it does not yield stable policy paradigms. Given this, perhaps no government could muster the communicative capacity strong enough to weaken its elasticity.

Conclusion How does the policy learning perspective perform against the approaches that dominate policy failure studies? Thinking about the stages-​based approaches, learning analysis is not confined to a single moment in policy making (for example, decision-​making or implementation). Rather, it treats all forms of policy failure –​processual, programmatic or political –​in their entirety. As well as avoiding the artificiality of stagist analyses, the learning framework helps us escape the atheoretical determinism of accounts that treat failure as a function of institutional characteristics or individual characters. By focusing on learning, we have a causal mechanism that trains attention on how decision-​ makers’ beliefs are updated (or not). Finally, the learning account helps systematise governance capacity accounts of policy failure. While tool-​ based explanations identify causal mechanisms, the accounts remain stand alone. By linking a recent policy learning framework to the capacity literature, we have a single model for the analysis of a variety of policy-​making settings. Of course, there are trade-​offs with this kind of approach. First, we may simply not ‘buy’ policy making as the realm of learning. There are certainly plenty of recurring policy failures that appear to give substance to that argument. Adopting a learning approach requires that we move beyond treating policy learning as something which is either present or absent, to accepting that learning is omnipresent –​although it is sometimes dysfunctional. Second, conceptual lenses famously obscure as much as they illuminate. If we are looking for learning are we

42

Pathologies of policy learning

missing power? We would answer ‘no’. Policy learning may be about the science of BTB and capacity building around the management of science. But, as our case demonstrates, it may also be a way to define and redefine power relations –​pausing epistemic learning and opening up the issue to additional stakeholders, for example. What have we learned about the policy failure in the epistemic mode? Two key contributions can be highlighted. First, exploring policy failure using theories of policy learning and organisational capacity illuminates both the multi-​faceted nature of learning and its complex relationship with policy performance and governance (Borrás, 2011; Dunlop, 2015; Zito and Schout, 2009). While the capacity to learn from experts may be strong in one respect, weaknesses in other areas may be enough to frustrate policy action based on the advice being given. In the BTB case, despite strong administrative capacity, weaknesses in certain aspects of absorptive, analytical and communicative capacity interacted to create the conditions for failure. Specifically, the inability to absorb the epistemic community’s message deeply enough to create a policy ‘plan B’ left decision-​makers isolated from reality. The lack of control over how the findings were linked to policy resulted in the politicisation of the research process; and the absence of social consensus around the research and method left evidence-​based policy making on BTB debased and ineffective. Second, conceptualising policy failure as the degeneration of learning about organisational capacity underlines the fundamentally dynamic character of policy failure. In particular, this case demonstrates the temporal and political dimensions of failure (Howlett et  al, 2015b; McConnell, 2010). Taking time first, decision-​makers did not alter or strengthen their absorptive and analytical capacities as the experiment results began to come in. This case demonstrates that such blocks to capacity building can be highly political  –​from 2004 decision-​ makers struggled to create a non-​culling policy alternative in the face of pressure from the NFU, for example. Then after the ISG’s report was released in 2007 learning degeneration that had been unintended then became orchestrated; with a counter-​epistemic community (King group) created to open up the science and issue. Finally, this leads us to prescriptive implications, and specifically how capacity can be re-​built or re-​designed. Identification of the two central dimensions of learning reveals what is needed for functional policy learning (Dunlop and Radaelli, 2016). When learning is incomplete or becoming dysfunctional, the framework can be used to generate alternative learning strategies.

43

Policy Learning and Policy Failure

Notes 1 The article arises out of original research funded by the British Academy (BA SG-​ 50865). Previous versions were presented at ‘Policy Failure Workshop’, Lee Kuan Yew School of Public Policy (LKYSPP) National University of Singapore, 19–​21 February 2014. The article has taken various forms and has benefited hugely from colleagues’ insightful comments. Particular thanks are extended to Sarah Ayres, Michael Howlett, Adrian Kay, Allan McConnell, Peter May, B Guy Peters, M Ramesh and Diane Stone, and all the participants of the Singapore workshop. Any errors or omissions remain my responsibility. 2 These semi-​structured interviews were conducted between December 2009 and April 2010. Respondents were elites central to the advisory governance of BTB and were identified through reports (official, NGO and scientific) and established professional contacts. Interviews were conducted under ‘Chatham House Rules’ and lasted between 45 minutes to 6 hours! They were digitally recorded, transcribed by the author and manually coded to elicit findings.

References Argyris, C, Schön, D, 1978, Organizational learning: A theory of action perspective, Reading, MA: Addison-​Wesley Bartels, KPR, 2015, Communicative capacity:  Public encounters in participatory practice and theory, Bristol: Policy Press BBC, 2013, The badgers have moved the goalposts, BBC Spotlight South West, 9 October, www.youtube.com/w ​ atch?v=UVNAdvry2dk Béland, D, Cox, RH, 2011, Ideas and politics in social science research, New York: Oxford University Press Bennett, CJ, Howlett, M, 1992, The lessons of learning: Reconciling theories of policy learning and policy change, Policy Sciences 25, 3,  275–​94 Borrás, S, 2011, Policy learning and organizational capacities in innovation policies, Science and Public Policy 38, 9, 725–​34 Christiansen, T, Piattoni, S (eds) 2003, Informal governance in the European Union, Cheltenham: Edward Elgar Davies, HJO, Nutley, SM, Smith, PC (eds) 2000, What works? Evidence-​ based policy and practice in public services, Bristol: Policy Press Deutsch, KW, 1966, The nerves of government, New York, NY: The Free Press Dunlop, CA, 2010, The temporal dimension of knowledge and the limits of policy appraisal: Biofuels policy in the UK, Policy Sciences 43, 4, 343–​63 Dunlop, CA, 2013, Epistemic communities, in M Howlett, S Fritzen, X Wu, E Araral (eds) Routledge handbook of public policy, Abingdon: Routledge, pp 229–​43

44

Pathologies of policy learning

Dunlop, CA, 2014, The possible experts: How epistemic communities negotiate barriers to knowledge use in ecosystems services policy, Environment and Planning C –​Government and Policy 32, 2, 208–​28 Dunlop, CA, 2015, Organizational political capacity as learning, Policy and Society 34, 3, 259–​70 Dunlop, CA, 2016, Contestation and contingency in advisory governance, in M Bevir and RAW Rhodes (eds) Rethinking governance: Rules, rationalities and resistance, Abingdon: Routledge Dunlop CA, 2017, Policy learning and policy failure:  Definitions, dimensions and intersections, Policy & Politics 45, 1, 3–​18 Dunlop, CA, James, O, 2007, Principal–​a gent modelling and learning:  The European Commission, experts and agricultural hormone growth promoters, Public Policy and Administration 22, 4,  403–​22 Dunlop, CA, Radaelli, CM, 2013, Systematizing policy learning: From monolith to dimensions, Political Studies 61, 3, 599–​619 Dunlop, CA, Radaelli, CM, 2016, Policy learning in the Eurozone crisis: Modes, power and functionality, Policy Sciences 49, 2, 107–​24 EFRAC (Environment, Food and Rural Affairs Committee), 2008, Badgers and cattle TB: Final report of the independent scientific group on cattle TB, Vol II Oral and written evidence, HC 130-​II, Incorporating HC 725, London: Stationary Office Godfray, HCJ, Curnow, RN, Dye, C, Pfeiffer, D, Sutherland, WJ, Woolhouse, MEJ, 2004, Independent scientific review of the randomised badger culling trial and associated epidemiological research, London: DEFRA Guston, DH, 1996, Principal–​agent theory and the structure of science policy, Science and Public Policy, 23, 4, 229–​40 Haas, PM, 1992, Introduction:  Epistemic communities and international policy co-​ordination, International Organization 46, 1,  1–​35 Hall, PA, 1993, Policy paradigms, social learning and the state: The case of economic policymaking in Britain, Comparative Politics 25, 3,  275–​96 Haynes, L, Service, O, Goldacre, B, Torgerson, D, 2012, Test, learn, adapt, London: Cabinet Office Heclo, H, 1974, Modern social politics in Britain and Sweden, New Haven, CT: Yale University Press Hirschman, AO, 1970, The search for paradigms as a hindrance to understanding, World Politics 22, 3, 329–​43 Howlett, M, Ramesh, M, Wu, X, 2015a, The dynamics of policy capacity, Policy and Society 34, 3–​4, 165–​284

45

Policy Learning and Policy Failure

Howlett, M, Ramesh, M, Wu, X, 2015b, Understanding the persistence of policy failures: The role of politics, governance and uncertainty, Public Policy and Administration 30, 3–​4, 209–​20 Ingold, J, Monaghan, M, 2016, Evidence translation: An exploration of policymakers’ use of evidence, Policy & Politics 44, 2, 171–​90 ISG (Independent Scientific Group) on Cattle TB, 1998, Towards a sustainable policy to control TB in cattle (first annual report of the independent scientific group on cattle TB), London: MAFF ISG (Independent Scientific Group) on Cattle TB, 1999, An epidemiological investigation into bovine tuberculosis (second annual report of the independent scientific group on cattle TB), London: MAFF ISG (Independent Scientific Group) on Cattle TB, 2007, Bovine TB: The scientific evidence. A science base for a sustainable policy to control TB in cattle: An epidemiological investigation into bovine tuberculosis. final report, London: DEFRA Janis, IL, 1972, Victims of groupthink, Houghton, NY: Mifflin Jones, BD, Baumgartner, FR, 2005, The politics of attention:  How government prioritizes problems, Chicago: University of Chicago Press King, D, 2007, Bovine tuberculosis in cattle and badgers, London: DEFRA Krebs, J, Anderson, R, Clutton-​Brock, T, Morrison, I, Young, D, Donnelly, C, 1997, Bovine tuberculosis in cattle and badgers, Report to The Rt Hon Dr Jack Cunningham MP, London: MAFF Lindblom, C, Cohen, D, 1979, Useable knowledge, New Haven, CT: Yale University Press LSE GV314 Group, 2012, Groups and the limited pluralism of the set-​piece consultation, The British Journal of Politics and International Relations 14, 1, 175–​86 McAdam, D, Tarrow, S, Tilly, C, 2001, Dynamics of contention, Cambridge: Cambridge University Press McConnell, A, 2010, Policy success, policy failure and grey areas in-​ between, Journal of Public Policy 30, 3, 345–​62 McLennan, G, 2004, Travelling with vehicular ideas: The case of the third way, Economy and Society 33, 4, 484–​99 Peters, BG, 2015, State failure, governance failure and policy failure:  Exploring the linkages, Public Policy and Administration 30, 3–​4,  261–​76 Sanderson, I, 2006, Complexity, ‘practical rationality’ and evidence-​ based policy making, Policy & Politics 34, 1, 115–​32 Schout, A, 2009, Organizational learning in the EU’s multi-​level governance system, Journal of European Public Policy 16, 8, 1124–​44 Stoker, G, 2010, Translating experiments into policy, Annals of the American Academy of Political and Social Science 628, 47–​58

46

Pathologies of policy learning

Turnpenny, J, Russel, D, Jordan, A, 2014, The challenge of embedding an ecosystem services approach:  Patterns of knowledge utilisation in public policy appraisal, Environment and Planning C –​Politics and Space 32, 2, 247–​62 Verdun, A, 1999, The role of the Delors committee and EMU, Journal of European Public Policy 6, 2, 308–​29 Weiss, CH, 1979, The many meanings of research utilization, Public Administration Review 39, 5 (Sept/​Oct), 426–​31 Weiss, CH, Murphy-​Brown, E, Birkeland, S, 2005, An alternative route to policy influence, American Journal of Evaluation 26, 1, 12–​30 Wilsdon, J, Willis, R, 2004, See-​through science: Why public engagement needs to move upstream, London: Demos Zaha, SA, George, G, 2002, Absorptive capacity:  A review, reconceptualization and extension, Academy of Management Review 27, 2, 185–​203 Zito, AR, Schout, A, 2009, Learning theory reconsidered:  EU integration theories and learning, Journal of European Public Policy 16, 8, 1103–​23

47

3

Overcoming the failure of ‘silicon somewheres’: learning in policy transfer processes Sarah Giest

Introduction Silicon Valley has received a great deal of attention from scholars and decision-​makers, due to its unparalleled success. It serves as a prominent exemplar for the success of high-​technology clusters and is often a role model for ‘Silicon Somewheres’ around the world. Silicon Valley refers to an area of the Santa Clara Valley that begins at San Carlos, about 20 miles south of San Francisco, and extends along the San Andreas Fault south to San Jose. During the 1980s and 1990s, policy makers worldwide tried to imitate the Californian success story and something similar is happening in other high-​technology fields today, where Silicon Valley remains the blueprint for much of the initiatives that are being undertaken (Cooke, 2004; Jaruzelski, 2014). There are a variety of theories surrounding the evolution of this famous cluster and the mechanisms that kept it successful. Some emphasise the idiosyncrasies of its establishment, while others construct a historical path-​dependent argument stressing ‘inevitability’ where current success arises from forces that trace back as far as the Gold Rush (Moore and Davis, 2004). Analyses find various factors including its entrepreneurs, the technologies that they commercialised and the firms they created as crucial elements for the Silicon Valley success (Kenney and Patton, 2006).And, while much attention has been paid to the success factors of Silicon Valley, there is scarce information on whether transferring this model has worked and what current transfer cases learn from earlier adopters (Hassink and Lagendijk, 2001; Karch et al, 2016). Addressing this gap, the chapter raises the question of how the transfer of the Silicon Valley Model (SVM) plays out in various settings. The goal is to explore successful and failed Silicon Valley imitators in order

49

Policy Learning and Policy Failure

to identify learning and adaptation mechanisms of the Valley-​approach to different settings over time. This is relevant to the growing policy transfer literature, as transfer has witnessed an upsurge in the last couple of years; however, there is limited research on how interventions change over time (Flanagan and Uyarra, 2016). Policy transfer is ‘a process in which knowledge about policies, administrative arrangements, institutions and ideas in one political setting (past or present) is used in the development of policies, administrative arrangements, institutions and ideas in another political setting’ (Dolowitz and Marsh, 2000, 5). Thereby, knowledge exchange is highly dependent on the setting it occurs in as well as on the individuals involved in the process. Policy transfer is also not an all-​or-​nothing procedure. As Dolowitz and Marsh (2000) point out, there are different degrees of transfer: copying, emulation, combinations and inspiration. These categories move from direct and complete transfer to searching for inspiration to create policy change. They are also fluid, as different levels of government might look for varying degrees of transfer. Case selection The cases presented all operate under the assumption that the SVM offers a solution to facilitate cluster initiatives and eventually even support economic growth in the region. The analysis includes two cases in the US, where transfer was being done in a similar environment and the same industry: New Jersey’s ‘Silicon Valley East’, and Texas. Beyond that, the research includes two more cases that are located outside of the US: in South Korea and Europe. The South Korean case is set in the same industry, but faces a different geographical context, while Medicon Valley (Copenhagen and Scania) is not only in a different part of the world, but also uses the SVM for the life science industry. In short, all cases portray some form of policy transfer by referring directly to Silicon Valley as well as establishing connection with the cluster through experts, individuals or organisations while varying in time and location. The analysis of the cases is based on the examination of publicly available documentation of the policy transfer. For New Jersey, the research primarily relies on the extensive study by Leslie and Kargon (1996), whereas Texas, South Korea and Medicon Valley include official reports by Ministries, municipalities and companies. The documents span publications from the mid-​1990s up until 2015. This limits the case analysis in developing concrete evidence of political learning, since these documents do not typically provide detailed accounts of different

50

Learning in policy transfer processes

learning strategies. The chapter further presents positive cases of transfer, also called ‘borrowing scenarios’ (Dolowitz and Marsh, 2000) in addition to giving a short description of the development of Silicon Valley itself. The geographic dispersion accounts for distant transfer that occurs beyond the US context of Silicon Valley as well as transfer in the same system. In all cases, the efforts of imitation are government-​driven and largely top-​down. Whereas long-​established networks, such as the one in Silicon Valley, evolve spontaneously and organically, imitators are largely coordinated efforts by government to facilitate an industry in a specific area (Casper, 2007; Ingstrup and Damgaard, 2013; Park et al, 2016; Su and Hung, 2009). In these efforts, it is difficult to locate a single actor steering this policy process, the analysis thus includes government as well as private stakeholders and consultants. The four cases serve as examples of policy transfer but are not exhaustive for regions that have used the SVM. Especially unsuccessful attempts are hard to find, as policies do not reach a formal stage. The focus is thereby on the adaptation process of the SV model to the cases, so-​called ‘adaptive learning mechanisms’. The examples show that elements of policy transfer fail in one way or another, but the question is whether, or not, the adaptive learning mechanisms put in place can prevent or overcome failure of the transfer. In cases of failure, learning processes are oftentimes cut off or incomplete leading to a generic and stylised model of local agglomeration, which tends to be largely unconcerned with the actual detailed interplay between government, industry and academia (Iammarino and McCann, 2006). Based on this assumption, the chapter puts forward the following reasoning for potentially disappointing scenarios:  those being unsuccessful in replicating Silicon Valley have transferred policy on the ground of the desirability of goals without regard to the feasibility and that there are variations of learning that go into these policy transfer scenarios. The following section reviews the literature on learning mechanisms and policy transfer, looking specifically at ‘soft’ transfer, the degree of transfer and learning in networks. The third section outlines the characteristics of Silicon Valley itself, while section four looks at its imitators in North America, Asia and Europe. The final two sections look at the connection between learning and the success or failure of policy transfer.

Policy transfer The creation of Silicon Valley-​like clusters starts with policy transfer. Policy transfer is defined as a process by which knowledge about

51

Policy Learning and Policy Failure

policies, administrative arrangements, institutions and ideas in one political setting (past or present) is transferred to another political setting (Dolowitz and Marsh, 2000). Rose (1993) and Dolowitz and Marsh (2000) further make a distinction between different degrees of transfer. Both accounts differ slightly, but largely describe the same continuum from direct copying to regions being inspired by another policy or model. Rose (1993) identifies the degrees of copying, adaptation, creating a hybrid, synthesis and inspiration while Dolowitz and Marsh (2000) distinguish between copying, emulation, combinations and inspiration. Thereby, copying refers to transferring a more or less intact programme that is running in another jurisdiction. This mostly happens within the same country, as it does not account for major differences in political system, language or culture. It is also likely to fail, since the borrower is not familiar with the ways in which the policy works in the lender’s context (Park et al, 2016). Adaptation or emulation adjusts for contextual differences by imitating the ideas behind the policy or programme. When a jurisdiction uses combinations of another programme or creates a hybrid that means that elements of the original example are newly compiled in a different region. Synthesis refers to ‘combining familiar elements from programs in a number of different places to create a new program’ (Rose, 1993, 30). Finally, transferring based on inspiration describes an intellectual stimulus that is then developed into a new programme (Dolowitz and Marsh, 2000; Rose, 1993). Stone (2004) frames the lower degrees of transfer in a slightly different way, by talking about selective borrowing. Selective borrowing means that policies and procedures are changed and adapted according to local needs and conditions. This opens the door to a more pessimistic way of portraying transfer by highlighting ‘incomplete’ or ‘uninformed’ transfers, which are bound to be unsuccessful. Another aspect of policy transfer is that it is not always voluntary. Inside and outside forces of nations and regions can lead to the adoption of a new policy or procedure (Dolowitz and Marsh, 1996). In the field of innovation and cluster policy, uncertainty and competition often push policy makers to adopt or copy other regions’ examples. Also, the fast pace of technologies can force governments into varying degrees of policy transfer. ‘Governments, not knowing how to deal with the issues of technological advances, turn to each other for precedents and ideas’ (Dolowitz and Marsh, 1996, 349). In the case of Silicon Valley especially, copying is encouraged by those hoping for another success story and is also easier to justify due to its unparalleled achievement. In this context of pushed or forced transfer, Rose (1993) warns that

52

Learning in policy transfer processes

‘if politicians adopt lessons solely on the ground of the desirability of goals without regard to the feasibility of means, they run the risk of backing a shipwrecked program’ (1993, 46), which refers to policies that have high political desirability, but low practicality. Because some programmes are context-​dependent, they are not capable of being generalised. Stone (2004) further highlights the possibility of soft forms of transfer that include the structures and practices in which non-​state actors play a more prominent role. This has been a dominant form of transfer in ‘Silicon Somewheres’ where experts and consultants advise industry and government stakeholders on possible ways to replicate high-​technological success. Soft transfer is also typical for network settings where local actors collaborate and extend their expertise by reaching out to stakeholders that hold knowledge on Silicon Valley or are situated in the cluster. The concept of policy transfer is thereby inherently vague, because at times it cannot ‘determine with precision the phenomenon it is trying to explain’ (Evans and Davies, 1999). For example, the effects of policy transfer on ‘success’ and ‘failure’ are less clear. James and Lodge (2003) therefore suggest that no distinct measure can be derived from policy transfer, because the current understanding prevents it from being disentangled from other processes of policy making. More recent approaches have tackled these shortcomings by considering the role of learning, in which policy transfer becomes the explanatory variable for possible processes of change (Capello and Lenzi, 2016; Dolowitz and Marsh, 2000; Radaelli, 2000). Stone (1999) builds on this idea by defining transfer processes as ideas or programmes underpinned by a deeper and prior process of learning. Learning and adaptation Unpacking some of the ideas that have been raised in the context of policy transfer and learning, there are different understandings of who learns what and when. Cohen and Levinthal highlight that ‘learning capabilities involve the development of the capacity to assimilate existing knowledge’ (1990, 130). The mechanism behind this is that the more stakeholders connect to other stakeholders and the more knowledge is gained, the better everyone can identify missing information and who can offer it. This process encompasses three components: exploratory learning, transformative learning and exploitative learning. The three elements are part of one process, as exploratory learning describes the confrontation with new knowledge, transformative learning changes the way in which this knowledge is

53

Policy Learning and Policy Failure

assimilated and combined with prior knowledge, while exploitative learning describes how new knowledge is translated into action, which will ideally benefit the organisation or network (Buenstorf and Murmann, 2005; Harvey et  al, 2010; Lane et  al, 2006).These elements can be summarised under the label of ‘adaptive learning’, which defines ‘the careful study of the functioning and the purposes of institutions in other regions and their adaptation to their own regional conditions’ (Hassink and Lagendijk, 2001, 68). Learning within the context of policy transfer further falls into the category of serial learning (sequential) from exogenous sources (Newig et al, 2016). This includes adapting policies based on external sources or, more informally, initiating processes of ‘trial-​and-​error’ (Newig et al, 2016). Strictly speaking, learning only begins when implementation problems are overcome and external policy models are adjusted to local contexts (Rose, 1991; May, 1992). The concept of learning regions describes institutional networks that develop and implement a regional innovation strategy, while regional learning encompasses cooperation between actors in a region through which they learn (Boekema et  al, 2000; Capello and Lenzi, 2016; Hassink, 2005). Hassink (2005) develops these ideas further suggesting a learning cluster concept, which bridges the gap between regional learning and the learning region by tackling the problem of lock-​ins in regional economies. In all these theories, the region or cluster ‘is not seen as a container, in which attractive location factors may or may not happen to exist, but rather as a milieu for collective learning through intense interaction between a broadly composed set of actors’ (Maskell and Malmberg, 1999, 174). Part of this learning process is also ‘un-​learning’. This idea stems from a path-​dependent conception of regions, which assumes that competences acquired over the years are constrained around a limited number of activities (Capello and Lenzi, 2016; Mathews, 1997).This implies that stakeholders have to unlearn certain routines and set-​ups that were established over many years in order to be open to change. Sometimes it might even be necessary to dismantle and remove institutions which hinder the implementation of a new policy (Maskell and Malmberg, 1999). In short, regions and clusters have learning trajectories that are more or less developed, but often follow a long-​ established path, which takes effort to unlearn. They also have different levels of learning capabilities. This is similar to what Rose (1993) calls ‘self-​correcting’, where stakeholders in general and policy makers in particular have to react to changes in the environment that might affect the programme or the goals of the programme in the context

54

Learning in policy transfer processes

of policy transfer (Smith, 2004). Since learning is difficult to capture, changes in the transferred policy account for learning and adaptation processes (Evans and Davies, 1999). Success and failure in policy transfer The relationship between policy transfer and the success or failure of such an endeavour is challenging, because constituting what is identified as a success or failure is often complex (James and Lodge, 2003). Dolowitz and Marsh therefore suggest ‘concentrating upon the extent to which policy transfer achieves the aims set by a government when they engaged in transfer, or is perceived as a success by the key actors involved in the policy area’ (2000, 17).The reasons for achieving those goals vary. Factors easing the way for successful transfer, according to Stone (2004), include the relationship between the perceived problem and the solution, and the amount of information available on the original policy programme. Dolowitz and Marsh (2000) suggest that failure, on the other hand, is due to uninformed, incomplete or inappropriate transfer. Addressing success and failure from a learning perspective highlights conditions favourable for learning, which include the existing (institutional) set-​up as well as the capacity of actors involved to turn the knowledge into a lesson for their own region (Stone, 1999). A case of policy transfer without or limited learning can then lead to a less coherent transfer of policy and ultimately an ad hoc or piecemeal solution. Based on this, the idea of learning during and after policy transfer can be further unpacked as even regions who have potentially learned can fail in setting up a regional cluster. In short, there are different types of learning connected to policy transfer. Learning can be the adoption of solutions which were successful in other regions (lesson-​drawing (Rose, 1991); policy transfer (Dolowitz and Marsh, 1996), or adjusting imported solutions to the specific problem of a region (Benz and Fürst, 2009). Figure 3.1 visualises the three possible scenarios. It highlights that once policy transfer is undertaken there is the possibility of imitation where no learning takes place. Policies are simply implemented by copying the original. Another option is trial-​and-​error learning, which describes gradual changes over time to policies, and temporary solutions (Knoepfel and Kissling-​Näf, 1998). This, as Toens and Landwehr (2009) point out, is sometimes at odds with the time pressure high-​technology industries face and according to Newig et al (2016) is often outright rejected due to opposition by stakeholders.

55

Policy Learning and Policy Failure Figure 3.1: Learning processes during policy transfer Policy transfer

No learning imitation

Trial-and-error learning

Adaptive learning

Source: author

Another possibility is adaptive learning, where policies are analysed and then adjusted before being implemented in the local setting. This requires reflection on the fit of the transferred policy and continuous adjustments in light of economic uncertainties and path dependencies (Flanagan and Uyarra, 2016). The following section summarises the Silicon Valley development and further aims to identify the learning and adaptation mechanism during transfer in four Silicon Somewheres.

Silicon Valley Silicon Valley’s pioneers started out with the goal of replicating the developments in Boston’s technology complex. However, lacking close ties with government agencies and the financial support from established electronic producers, Stanford’s leaders facilitated the formation of new technology enterprise and forums for cooperation with local industry. These start-​ups then forged a model of semiconductor production in close relationship with the region’s social and technical networks (Saxenian, 1994). The existing electronic industry, companies such as Hewlett-​Packard and Varian, informal investors and a close cooperation of university and business then fostered an entrepreneurial support system and the invention developed into a high-​technology cluster. Thereby, the semiconductor itself played an important role. It was used in workstations, personal computers and computer networking, and thus helped create new industries. The semiconductor industry further encouraged the establishment of an entrepreneurial culture in connection with investment opportunities for venture capital. It also drew attention to the region for new firms and scientists. The year 1993 marks a new stage in Silicon Valley’s evolution with the commercial exploitation of the internet and the creation of the World Wide Web. Since then

56

Learning in policy transfer processes

Silicon Valley has continuously grown, with smaller networks or cooperations being combined to form one big and powerful network (Fleming and Frenken, 2006; Kenney and Patton, 2006). Today, the cluster is regarded as an incubator region which is host not only to an array of entrepreneurs, but also to institutions that nurture the firms they create. This system was formed over time in tandem with the industries in the region (Kenney and Patton, 2006). Based on these developments, scholars draw different conclusions as to the key dynamics at play between a supporting (institutional) system and entrepreneurs, the exploitation or commercialisation of knowledge and low entry barriers for entrepreneurship  –​not only financially, but also in social and psychological terms. For example, Kenney and Patton (2006) conclude that the evolution of a strong network not only quickly attracted venture capital, but also strengthened the relationship among individuals and the institutional level. Moore and Davis (2004) point to the role of government as the driving force behind sharing findings on semiconductors (antitrust settlement) and investing in research and development (R&D). Pre-​ dating the semiconductor industry, Moore and Davis (2004) also see defence expenditure as important for the development of transistors and helpful to early firms’ bottom lines. Another aspect of the Silicon Valley phenomena that received attention in the literature is the role of Stanford University. Many hypothesised that the driving factor of its success is the role of the university as a connector for larger and smaller companies. Professor Frederick Terman’s role in Stanford’s transformation and rise to prominence has been particularly emphasised. As a Provost and Dean of engineering, he increased ties among business and academic circles and encouraged faculty to be entrepreneurs (Moore and Davis, 2004). There are conflicting statements in the literature about which development stage would be most informative. Many analysts make conclusions based on the status quo, without including the history of the cluster. Leslie and Kargon (1996), for example, point out that today’s Silicon Valley –​or at least the one in the 1990s –​‘may now offer a more appropriate development guide for globally integrated high technology regions’ than it did in the beginnings of its development (Leslie and Kargon, 1996, 472). In contrast, Bresnahan and Gambardella (2004) point out that ‘looking at Silicon Valley in its mature phase cannot tell us much about the preconditions or causal mechanisms’ (Bresnahan and Gambardella, 2004, 2). In fact, Bresnahan, Gambardella and Saxenian (2001) state that the Silicon Valley of 40 years ago is closer to today’s clusters than any of them is to the Silicon Valley of today.

57

Policy Learning and Policy Failure

Silicon Somewheres New Jersey New Jersey made the attempt of copying Silicon Valley in the mid-​ 1960s. The US state recruited the ‘father of Silicon Valley’, Fred Terman, who was retiring from Stanford. During that time New Jersey was already a leading high-​tech centre with 725 companies –​most prominently the inventor of the transistor ‘Bell Labs’  –​and home to a large science and engineering workforce of 50,000 researchers (Wadwha, 2013). New Jersey further ranked fourth among all US states in R&D expenditures with about 10 per cent of the national total. However, the efforts by Terman and others to create a successful cluster failed. The establishment of an applied, graduate institute modelling Stanford University turned out to be difficult, as companies were reluctant to invest or relocate to the region. The connection to Princeton was also complicated, due to the university’s emphasis on theory. Consequently, ‘while New Jersey’s high technology companies continued to prosper –​by 1991 the state’s 700 laboratories and 100,000 scientists and engineers represented a $15 billion annual investment in R&D –​they did so individually rather than collectively, and certainly not on the Silicon Valley model’ (Leslie and Kargon, 1996, 451). One of the reasons for this failed attempt was the strong focus on the role of the university facilitated by the lack of skilled labour. As Leslie and Kargon (1996) point out, Terman ‘tended to overestimate the importance of a particular kind of educational institution as the catalyst for regional development while underestimating the difficulty of convincing competing firms to cooperate’ (Leslie and Kargon, 1996, 438). This focus was further propelled by a 1962 Bell Labs study suggesting that the company would be facing a severe shortfall of scientists and engineers in the future and that it should consider establishing some sort of university partnership (Leslie and Kargon, 1996). This prognosis fed into the hype around an Institute of Science and Technology (IST) and led to an educational approach evolving only around solid state physics, materials science and molecular biology. Also, there was a limited role for small firms within IST or the New Jersey cluster in general, even though the generation of start-​ups was one of the trademarks of Silicon Valley (Almeida and Kogut, 1999). In short, the case of New Jersey shows that much emphasis was put on the role of Stanford in Silicon Valley while some of the other dynamics that propelled the Valley into its position, such as the support of entrepreneurs and start-​ups and the collaboration and networking

58

Learning in policy transfer processes

among stakeholders, were neglected. It seems decision-​makers were unable to balance the immediate need for skilled labour with some of the long-​term efforts. Texas Around the same time, Texas was also looking to become the ‘next Silicon Valley’ (Miller and Côté, 1985; Rogers and Larsen, 1984). Despite being a one-​crop economy, targeting cattle, cotton and finally oil, post-​war entrepreneurs were looking to benefit from federal defence contracts. This led to established companies like General Dynamics or Lockheed setting up large facilities in the area and attracting smaller companies in the semiconductor industry, often connected to Bell Labs (Leslie and Kargon, 1996).Texas also hired Fred Terman to set up higher education institutions similar to Stanford. These efforts were propelled by the fact that during the 1980s, Dallas belonged to the fastest growing metropolitan region, together with Phoenix and San Diego, and experienced high growth. And for a while, Dallas was the nation’s third-​largest technology centre with the growth of companies such as Ling-​Temco-​Vought (LTV Corporation) and Texas Instruments (Mayer, 2011; Pollard and Storper, 1996). Texas Instruments (TI) was further able to find a niche in the electronic market as the first company to develop production techniques for the recently invented transistor. Gordon Teal from Bell Labs was hired to lead the development team in 1952 and shortly after, the company produced the first portable radio and became a key vendor for growing IBM (Price, 2013). However, TI’s failure to break into the booming microcomputer market in the early 1980s cut its development short. At the height of TI’s success, the company and the metropolitan region in general were mainly concerned with, similar to New Jersey, the availability of scientific resources and skilled labour. In comparison to California, Texas produced 550 less PhDs and had no first-​class research university of its own. In consultation with Fred Terman a new Graduate Research Centre was developed. The institution was modelled after New Jersey’s IST, with graduate students only and an interdisciplinary research profile. It recruited people from industry to provide a link between research and the market. The Graduate Research Centre hired Thomas Martin, one of Terman’s former doctoral students at Stanford and later Dean of engineering at Arizona and Florida, to join Southern Methodist University (SMU) in Dallas. The project, however, suffered a range of setbacks due to decline in the semiconductor industry, which led to funding cuts from

59

Policy Learning and Policy Failure

companies and government. And finally, ‘like their counterparts in New Jersey, the leaders of high technology industry in Texas discovered that educational institutions could not, by themselves, integrate a regional economy’ (Leslie and Kargon, 1996, 456). In short, Texas borrowed from both Silicon Valley and New Jersey to establish the cluster, emulating some of the strategies implemented in the Valley. It, however, was a latecomer to the semiconductor market and struggled to learn from the mistakes made in New Jersey connected to Stanford-​like structures and the role of small companies. South Korea Despite the failures of ‘Silicon Valley East’ and Texas, a similar attempt was made in South Korea –​again with Fred Terman’s consultation. And the South Korean case was a surprising success story. The leading educational institution, the Korea Advanced Institute of Science (KAIS) was (again) modelled on IST, but adjustments were made that facilitated horizontal integration. KAIS had support from the Ministry of Science and Technology, which allowed it to ‘foster, if not compel, horizontal integration and collective learning even within a high technology community of large, and often fiercely competitive, vertically-​integrated enterprises such as Samsung, Goldstar and Hyundai’ (Leslie and Kargon, 1996, 458). Parallel to developing the Institute, major changes in economic policy supported the growing industry. South Korea and its newly elected leader, Park Chung Hee, made plans in the early 1960s to invest more in R&D and gain access to foreign know-​how and capital (Kwon, 2009). Part of that plan included negotiations with the US. This led to the establishment of the Korea Institute of Science and Technology (KIST). KIST was not only the ‘bridge between domestic industry and advanced technologies of foreign countries’, it also encouraged Korean firms to invest more in their own research and development efforts (Leslie and Kargon, 1996, 460). It was further modelled after the Battelle Institute, meaning that its findings could be immediately applicable to the business sector (Woojin, 2010). In a phase where manufacturing was still South Korea’s main focus, the education system was expanded and by the 1980s, the Korean government turned to fostering high-​technology industries such as semiconductors. This transitioned in the 1990s into more knowledge-​ intensive industries. During these years, government made significant investments into technology and science parks and aimed at helping

60

Learning in policy transfer processes

private companies to develop into high-​tech ones (Arslanhan and Kurtsal, 2010; Mazzarol, 2013). In collaboration with US experts, the Korean Institute was ultimately able to manage the Silicon Valley efforts and succeeded. Important lessons from the case include the close collaboration between government, industry and research institutes (Mazzarol, 2013) as well as specific R&D policies targeting high-​tech sectors. Korean officials developed linkages with international high-​tech networks in the US early on, which gave them the opportunity to learn from Silicon Valley experiences and expert knowledge. In fact, while Silicon Valley is still the largest and most enduring region for technological innovation, Seoul is now its closest rival. Much of this was made possible by public investment. The Science Ministry, for example, recently announced a $1.5 billion initiative to upgrade Korea’s mobile infrastructure (Wortham, 2015). In short, elements transferred from the SVM were localised and combined in a way that was conducive to the South Korean case. The example also highlights a long-​term strategy connected to the attempts of emulating some of the Silicon Valley features, reaching beyond the actual transfer. Similar developments can be found in the example of Medicon Valley. Medicon Valley One of the more recent efforts to imitate Silicon Valley is Medicon Valley. The cluster focuses on life sciences and is located in the bi-​national Øresund region, which spans greater Copenhagen in Denmark and Scania in southern Sweden (Asheim and Moodysson, 2008). First steps to actively create the Scandinavian cluster were taken during an Interreg project in 1997. The project was backed by European funds and a pre-​study on the potential of the area to become a cluster. Both parts of Denmark and Sweden have a history of life science research and the settlement of large pharmaceutical companies like Novo Nordisk or Lundbeck. The Swedish area of Scania developed a basis for life sciences through the early presence of Astra (now AstraZeneca) and Pharmacia (now owned by Pfizer), which relocated parts of their research activities to Lund. In Denmark, similar developments took place, which led to the decision in 1994 to actively stimulate bi-​national regional development in life sciences (Moodysson et al, 2010). This resulted in the branding of the region as ‘Medicon Valley’ and government incorporating strategies from Silicon Valley (Lyck, 2006).

61

Policy Learning and Policy Failure

Most notably, Denmark changed its strategy from using national policies to support cluster development to focusing on framework conditions for entrepreneurs and to strengthen the regional development and cooperation around new business development and innovative networks (OECD, 2007). Denmark further offered sequence incentives for small firms in order to counter the risk of uncertain returns for those companies. The programme was modelled after the US Small Business Innovation Research Program (Rosenfeld, 2001), an initiative that was used by the Defence Advanced Research Projects Agency and NASA to act as strategic investors in Silicon Valley (Mazzucato, 2015). These changes were not without challenges. While the vision of building a biotech-​version of Silicon Valley facilitated cooperation among Danish and Swedish government stakeholders, over time, the drivers of the process disappeared and market forces struggled to cope with high transaction costs and inappropriate price policy concerning the bridge connecting Denmark and Sweden (Lyck and Boye, 2009).To improve upon current policy linked to firm creation and labour market flexibility, the Danish Ministry of Higher Education and Science has established the ‘Innovation Centre Denmark’ located in Silicon Valley with the purpose of strengthening Danish research, education, innovation and policy development at home (Ministry of Higher Education and Science, 2015). Overall, the Medicon Valley developments highlight the adaptation of the SVM, partially motivated by the difference in industry and the constant adjustment of policies in response to changing requirements in the region.

Analysis The cases reveal the complex situation that policy makers face when aiming to establish a Silicon Valley-​like cluster. Instead of following a static ‘recipe’ approach (Moore and Davis, 2004), the policies need to be tailored towards pre-​conditions in the region. Comparing the four Silicon Valley imitators shows that reflection on the Silicon Valley model by policy makers seems to grow as time and geographical distance increase. While New Jersey and Texas, both located in the US, fall into the copying and emulation category of imitators, South Korea and Medicon Valley were able to adapt the model in their favour and apply a mixture of different policies (South Korea) or use the cluster as an inspiration for policy changes, branding and regional agglomeration (Medicon Valley).

62

Learning in policy transfer processes

In New Jersey, policy makers focused primarily on the role of the university in the SVM, resulting in a partial transfer. Due to the fast decline of the industry and the coupled slow success of the university, there was no time to adjust the initiatives. Texas also used an educational institute as the integrative element of the cluster and mostly adjusted ad hoc, such as recruiting industry stakeholders to the newly established Graduate Research Centre. The trial-​and-​error procedure, however, could not counter all setbacks and ultimately failed. South Korea adapted the model before implementing, especially the Advanced Institute of Science, which played the key role in advancing integration. South Korea also developed a long-​term plan in which major changes in economic policy were undertaken to shape the region along the Silicon Valley idea without directly imitating it. And finally, Medicon Valley applied the Silicon Valley idea to the life science industry and developed a long-​term plan, which involved adjusting tax policies and start-​up initiatives. Obstacles to sustaining the SVM thus include limited time to try out policies or update the ones already implemented as well as a focus on the educational aspects of Silicon Valley. The adaptation of the model before implementing, also described as adaptive learning (Hassink and Lagendijk, 2001), led to long-​term strategies and altering fundamental policies reaching beyond the region and the sector. The cases present little evidence for the active ‘un-​learning’ of certain patterns and routines (Hassink, 2005; Maskell and Malmberg, 1999), but this dynamic could be part of the adaptation and updating of policies over time. New Jersey, Texas, South Korea and Medicon Valley highlight different processes at play during policy transfer –​partial imitation, trial-​and-​error and adaptation or adaptive learning. And whereas trial-​ and-​error happens during the transfer and requires a longer breath to address various obstacles, the adaptation process happens before Silicon Valley-​like changes are made with a long-​term vision for the local cluster. More evidence is, however, needed to make a sharper distinction between these categories. Some researchers point towards the pre-​existence of a working innovation system (Feldman, 2008; Hospers, 2004). The ecosystem for innovation and the previously highlighted learning mechanisms, however, go hand in hand. For the Medicon Valley efforts, for example, Denmark and Sweden did not work together systematically, but are now close collaborators through adapting existing policies and adding novel ones. These aspects point towards a partly path-​dependent argument, where stakeholders use previously established structures

63

Policy Learning and Policy Failure

while also implementing new initiatives. There appears to be a great variation in the ability of regions to adapt existing policies to the policy transfer dynamics. This opens up possibilities for incorporating new institutions and ideas. Other factors include continued collaboration and knowledge transfer, identifying the cluster’s competitive advantage over other hubs and establishing a strong network. Defining a competitive advantage in the market is crucial, because any cluster following Silicon Valley –​at least the ones in a similar industry –​has lost first-​ mover advantage and will be compared to its success. This implies a long-​term strategy, such as the one seen in South Korea, which can accommodate changes beyond the immediate surge of an industry and focuses on underlying factors, such as a constant in-​flow of skilled labour. A  policy strategy can offer an assessment of collaborative structures and possible gaps in the network as well as give an indication of investment plans and opportunities. All in all, as pre-​conditions and the degrees of transfer differ, it comes down to the ongoing learning and adaptation capacities of both individuals and organisations in order to engage successfully in establishing high-​tech clusters. An important feature is also the flexibility towards change in existing structures. As especially the South Korean example shows, changing dynamics of restructuring, adaptation and the establishment of new institutions challenge policy transfer efforts.

Conclusion Policy transfer occurrences have seen a surge in recent years due to technological advances in communication and economic pressure to follow ‘best practices’. Their application has led to many successful, but also failed, examples, with limited attention being paid to underlying learning processes. This research tested the assumption that those being unsuccessful in replicating programmes, have transferred policy on the ground of the desirability of goals without regard to the feasibility and that there are variations of learning that go into these policy transfer scenarios. Based on transfer cases of the Silicon Valley Model in Asia, Europe and the US, the findings suggest that various challenges arise in policy transfer and that the type of learning contributes to its overall success or failure. The learning mechanisms varied across the cases of Medicon Valley, New Jersey, South Korea and Texas. There was imitation, trial-​and-​error and adaptive learning. Trial-​and-​error

64

Learning in policy transfer processes

describes scenarios in which obstacles are largely tackled gradually and over time and solutions are often temporary. Adaptive learning includes elements of planning and adjusting the larger structures to accommodate the transferred policies. This requires reflection on the fit of the transferred policy before its implementation. The three learning processes presented here do not prevent failure, but seen as a continuum, the more policy makers move towards adaptive learning before and during transfer, the more likely it is that they can cope with obstacles occurring during the process. There is far more that can be said about the policy transfer in ‘Silicon Somewheres’, as policy transfer itself might take on more or less prominence in various settings. While the study does not offer a conclusive answer on the connection between the type of learning and the success or failure of transfer, it does contribute to the discussion around learning patterns during transfer processes. More cases are thus needed to support the argument that the three types of learning mechanisms affect the outcome of policy transfer differently. Acknowledgements The chapter has benefited greatly from the insightful comments by the volume’s editor, Claire Dunlop, and two anonymous reviewers. The fruitful discussion with participants at the International Workshop on ‘Policy Design and Governance Failures’ at the Lee Kuan Yew School of Public Policy in Singapore also helped to shape the chapter. References Almeida, P, Kogut, B, 1999, Localization of knowledge and the mobility of engineers in regional networks, Management Science 45, 7,  905–​17 Arslanhan, S, Kurtsal, J, 2010, To what South Korea owes success in innovations? Implications for Turkey, TEPAV Policy Note, September Asheim, B, Moodysson, J, 2008, The Öresund region:  A dynamic region in Europe due to inter-​regional collaboration? Working Papers OnLine (WPOL), Institut Universitari d’Estudis Europeus, pp. 1–​17 Benz, A, Fürst, D, 2009, Policy learning in regional networks, European Urban and Regional Studies 9, 1, 22–​35 Boekema, F, Morgan, K, Bakkers, S, Rutten, R, 2000, Introduction to learning regions:  A new issue for analysis, in F Boekema, K Morgan, S Bakkers, R Rutten (eds) Knowledge, innovation and economic growth: The theory and practice of learning regions, Cheltenham/​ Northampton: Edward Elgar, pp 3–​16

65

Policy Learning and Policy Failure

Bresnahan, T, Gambardella, A, 2004, Old-​economy inputs for new-​ economy outcomes:  What have we learned? in T Bresnahan, A Gambardella (eds) Building high-​tech clusters, Silicon Valley and beyond, Cambridge: Cambridge University Press, pp 331–​59 Bresnahan, T, Gambardella, A, Saxenian, A, 2001, ‘Old economy’ inputs for ‘new economy’ outcomes: Cluster formation in the new Silicon Valleys, Industrial and Corporate Change 10, 4 Buenstorf, G, Mur mann, JP, 2005, Er nst Abbe’s scientific management: Theoretical insights from nineteenth-​century dynamic capabilities approach, Industrial and Corporate Change 14, 4, 543–​78 Capello, R, Lenzi, C, 2016, Persistence in regional learning paradigms and trajectories: Consequences for innovation policy design, European Planning Studies, 24, 9, 1587–​604 Casper, S, 2007, Creating Silicon Valley in Europe: Public policy towards new technology industries, Oxford: Oxford University Press Cohen, W, Levinthal, D, 1990, Absorptive capacity: A new perspective on learning and innovation, Administrative Science Quarterly 35, 1,  128–​52 Cooke, P, 2004, Special issue:  Globalisation of biotechnology, introduction, European Planning Studies 12, 7, 915–​20 Dolowitz, D, Marsh, D, 1996, Who learns what from whom: A review of the policy transfer literature, Political Studies 44, 2, 343–​57 Dolowitz, D, Marsh, D, 2000, Learning from abroad: The role of policy transfer in contemporary policy making, Governance: An International Journal of Policy and Administration 13, 1, 5–​23 Evans, M, Davies, J, 1999, Understanding policy transfer: A multi-​level, multi-​disciplinary perspective, Public Administration 77, 2, 361–​85 Feldman, M, 2008, The entrepreneurial event revisited: In a regional context, in C Karlsson (ed) Handbook of research on innovation and clusters, cases and policies, Cheltenham: Edward Elgar, pp 318–​42 Flanagan, K, Uyarra, E, 2016, Four dangers in innovation policy studies –​and how to avoid them, Industry and Innovation 23, 2, 177–​88 Fleming, L, Frenken, K, 2006, The evolution of inventor networks in the Silicon Valley and Boston regions, Advances in Complex Systems 10, 1, 53–​71 Harvey, CJ, Bartz, KK, Davies, JR, Francis, TB, Good, TP, Guerry, AD et al, 2010, A mass-​balance model for evaluating food web structure and community-​scale indicators in the central basin of Puget Sound, National Oceanic and Atmospheric Administration, United States Department of Commerce, Technical Memorandum NMFS-​NWFSC-​106, Seattle, Washington

66

Learning in policy transfer processes

Hassink, R, 2005, How to unlock regional economies from path dependency? From learning region to learning cluster, European Planning Studies 13, 4, 521–​35 Hassink, R, Lagendijk, A, 2001, The dilemmas of interregional institutional learning, Environment and Planning C: Politics and Space 19, 1, 65–​84 Hospers, G-​J, 2004, Innovation by imitation? Territorial benchmarking in EU regional innovation policy, Centre for European Studies working paper no 07/​04, The Netherlands: University of Twente Iammarino, S, McCann, P, 2006, The structure and evolution of industrial clusters: Transactions, technology and knowledge spillovers, Research Policy 35, 7, 1018–​36 Ingstrup, M, Damgaard, T, 2013, Cluster development through public sector facilitation, in K Brown, J Burgess, M Festing & S Royer (eds), Resources and Competitive Advantage in Clusters. 1st edn, Rainer Hampp Verlag: Mering, pp 46–​65 James, O, Lodge, M, 2003, The limitations of ‘policy transfer’ and ‘lesson drawing’ for public policy research, Political Studies Review 1,  179–​93 Jaruzelski, B, 2014, Why Silicon Valley’s success is so hard to replicate, Scientific American, 14 March 2014, www.scientificamerican.com/​ article/​why-​silicon-​valleys-​success-​is-​so-​hard-​to-​replicate/​ Karch, A, Nicholson-​Crotty, S, Woods, N, Bowman, A, 2016, Policy diffusion and the pro-​innovation bias, Political Research Quarterly 69, 1,  83–​95 Kenney, M, Patton, D, 2006, The co-​evolution of technologies and institutions: Silicon Valley as the iconic high-​technology cluster, in P Braunerhjelm, M Feldman (eds) Cluster genesis, technology-​based industrial development, Oxford: Oxford University Press Knoepfel, P, Kissling-​Näf, I, 1998, Social learning in policy networks, Policy & Politics 26, 3, 343–​67 Kwon, H, 2009, Policy learning and transfer: The experience of the developmental state in East Asia, Policy & Politics 37, 3, 409–​21 Lane, P, Koka, B, Pathak, S, 2006, The reification of absorptive capacity: A critical review and rejuvenation of the construct, Academy of Management Review 31, 4, 833–​63 Leslie, S, Kargon, R, 1996, Selling Silicon Valley: Frederick Terman’s model for regional advantage, Business History Review 70, 4, 435–​72 Lyck, L, 2006, Øresundsregionalisering:  Medicon Valley, Center for Tourism and Cultural Management, Copenhagen Business School Report, Copenhagen

67

Policy Learning and Policy Failure

Lyck, L, Boye, P, 2009, Cross border regions as forerunners in EU integration  –​ an institutional perspective on structural change, Paper presented at the Annual SNEE European Integration Conference, The Swedish Network for European Studies in Economics and Business (SNEE), Mölle, Sweden, 26–​29 May Maskell, P, Malmberg, A, 1999, Localised learning and industrial competitiveness, Cambridge Journal of Economics 23, 2, 167–​85 Mathews, J, 1997, A  Silicon Valley of the East:  Creating Taiwan’s semiconductor industry, California Management Review 39, 4, 26–​54 May, P, 1992, Policy learning and failure, Journal of Public Policy 12, 4,  331–​54 Mayer, H, 2011, Entrepreneurship and innovation in second tier regions, Cheltenham: Edward Elgar Mazzarol, T, 2013, Building a national innovation system: What can we learn from Korea? The Conversation, https://​theconversation. com/​building-​a-​national-​innovation-​system-​what-​can-​we-​learn-​ from-​korea-​9449 Mazzucato, M, 2015, The entrepreneurial state: Debunking private vs. public sector myths, Anthem Press: London Miller, R, Côté, M, 1985, Growing the next Silicon Valley, Harvard Business Review (July/​August),  114–​23 Ministr y of Higher Education and Science (Denmark), 2015, Innovation Centre Denmark, Silicon Valley, http:// ​ u fm. dk/​ e n/​ research- ​ a nd- ​ i nnovation/ ​ international- ​ c ooperation/​ g l o b a l - ​ c o o p e r a t i o n / ​ i n n ova t i o n - ​ c e n t re s - ​ a n d - ​ a t t a c h e s /​ innovation-​center-​denmark-​silicon-​valley Moodysson, J, Coenen, L, Asheim, B, 2010, Two sides of the same coin? Local and global knowledge flows in Medicon Valley, in F Belussi, A Sammarra (eds) Business networks in clusters and industrial districts: The governance of the global value chain, Abingdon: Routledge, pp  356–​76 Moore, G, Davis, K, 2004, Learning the Silicon Valley way, in T Bresnahan, A Gambardella (eds) Building high-​tech clusters, Silicon Valley and beyond, Cambridge:  Cambridge University Press, pp  7–​40 Newig, J, Kochskämper, E, Challies, E, Jager, N, 2016, Exploring governance learning:  How policymakers draw on evidence, experience and intuition in designing participatory flood risk planning, Environmental Science and Policy 55, 353–​60

68

Learning in policy transfer processes

OECD (Organisation for Economic Co-​operation and Development), 2007, Competitive Regional Clusters, National Policy Approaches, OECD Reviews of Regional Innovation, Paris: OECD Park, C, Lee, J, Wilding, M, 2016, Distorted policy transfer? South Korea’s adaptation of UK social enterprise policy, Policy Studies 38, 1,  39–​58 Pollard, J, Storper, M, 1996, A  tale of twelve cities:  Metropolitan employment change in dynamic industries in the 1980s, Economic Geography 72, 1, 1–​22 Price, G, 2013, Texas Instruments, Handbook of Texas Online, Texas State Historical Association, www.tshaonline.org/h ​ andbook/o ​ nline/​ articles/​dnt02 Radaelli, C, 2000, Policy transfer in the European Union: Institutional isomorphism as a source of legitimacy, Governance: An International Journal of Policy and Administration 13, 1, 25–​43 Rogers, E, Larsen, J, 1984, Silicon Valley fever: Growth of high-​technology culture, New York, NY: Basic Books Rose, R, 1991, What is lesson-​drawing? Journal of Public Policy, 11, 1,  3–​30 Rose, R, 1993, Lesson-​d rawing in public policy, Chatham, NJ: Chatham House Rosenfeld, S, 2001, Networks and clusters:  The yin and yang of rural development, in the conference proceedings exploring policy options for a new rural America, Kansas City:  Federal Bank Reserve Bank of Kansas City,  102–​20 Saxenian, A, 1994, Lessons from Silicon Valley, Technology Review July 97, 5, 42–​51 Smith, A, 2004, Policy transfer in the development of UK climate policy, Policy & Politics 32, 1, 79–​93 Stone, D, 1999, Learning lessons and transferring policy across time, space and disciplines, Politics 19, 1, 51–​9 Stone, D, 2004, Transfer agents and global networks in the ‘transnationalization’ of policy, Journal of European Public Policy 11, 3,  545–​66 Su, Y-​S, Hung, L, 2009, Spontaneous vs policy-​driven: The origin and evolution of the biotechnology cluster, Technological Forecasting and Social Change 76, 5, 608–​19 Toens, K, Landwehr, C, 2009, The uncertain potential of policy-​ learning: A comparative assessment of three varieties, Policy Studies 30, 3, 347–​63

69

Policy Learning and Policy Failure

Wadwha, V, 2013, Silicon Valley can’t be copied, MIT Technology Review, 3 July Woojin, J, 2010, Korea as a successful recipient country, Korea International Cooperation Agency, http://​tinyurl.com/​mu4wr2n Wortham, J, 2015, What Silicon Valley can learn from Seoul, The New York Times, 2 June

70

4

Between policy failure and policy success: bricolage, experimentalism and translation in policy transfer Diane Stone

Introduction This chapter re-​assesses some of the literature on policy transfer (see for example, Benson and Jordan, 2011; Dolowitz and Marsh, 2000; Evans and Davies, 1999) and policy diffusion (see for example, Dobbin et  al, 2007; Meseguer, 2005; Shipan and Volden, 2012) in light of ideas as to what constitutes failure, partial failure or limited success. It is often presumed that when policy transfer occurs internationally, ‘best practices’ or superior standards are being transferred to recipient jurisdictions. Or, in other words, the policy success of one country is exported to another. In a more or less rational process of decision-​making, importing governments recognise policy failures or shortcomings within their borders and through processes of evaluation and learning, as well as peer review, seek solutions and adopt reforms based on successful experience elsewhere. The discussion moves away from ‘orthodox’ policy transfer studies –​where there is often assumed to be a motivated importer and a willing exporter country –​ abandoning the linear perspectives of country ‘A’ sending policy to ‘B’ and shifting to a stronger analytical focus on the messy processes of hybrid policies emerging from multiple exemplars, and the messy interpretative processes where importing countries translate and amend transferred policies. Rather than focusing on policy failure as a starting point for analysis and a prompt for reforms and policy transfer, this chapter first looks at imperfect or uninformed transfer processes as one locus of policy failure. Second, little discussed in the literature is the concept of ‘negative lesson-​drawing’ which amounts to learning what not to do. Yet, widespread consensus on what to do, or not do, is often absent when stakeholders have different and divergent perspectives. The various

71

Policy Learning and Policy Failure

informants intersecting with policy transfer processes at different stages also complicate a neat linear transmission of an intact policy approach or tool. Third, whether ‘transfer’ can be said to have been successfully accomplished is qualified by the extensiveness of hybridity, synthesis, tinkering with models, adaptation and ‘localisation’ (see for example, Mukhtarov, 2014; Stone, 2012).This can result in a transferred policy tool or institution from Country A looking completely different in Country B and operationalised in substantively different fashion than originally conceived. Something is either lost, or learnt, in translation. Rather than outright ‘failure’, policy transfer has multiple dimensions, often succeeding in some respects but not in others, according to local circumstance and actors, and upon perception and interpretation. Policy transfer studies can be regarded as being both about past successes and failures, and learning from them, as well as about transitioning to potential futures and new policy development. Each iteration of transfer is a unique concoction. Political and policy community pressures for policy transfer are often a response to perceived policy failures and crises to be ameliorated by emulating or learning from the successes of policy innovation elsewhere. In other words, policy failure and policy success are directly linked in the rationalist thinking of policy transfer. However, policy transfer is rarely a perfect process of transmission. The diffusion of knowledge and transfer of policies across countries can involve a large number of ‘proponents’ and, as will be discussed later, the intermediaries in these processes reflect different interests, sometimes with discordant views on what may amount to success. Instead of regarding incomplete or de-​railed policy transfer in negative terms as failure, the chapter will link dynamics of assemblage (Lendvai and Stubbs, 2009), bricolage (de Jong, 2013) and localisation as modalities of translation in order to argue that the binary distinction between ‘success’ and ‘failure’ is inappropriate. As in art, ‘assemblage’ and ‘bricolage’ entails creation from a diverse range of available things. The notion of ‘localisation’ concerns the local adaption, indigenisation and modification of policy into new formats. Localisation is one ending or outcome of policy mobilities (McCann and Ward, 2012); that is, transfer and/​or diffusion is never an unmediated action, for the processes of transmission itself involve (mis)interpretation, mutation and revision en route. In sum, policy translation can be understood as multiple and variable processes incorporating (i)  diffusion/​transfer; (ii) assemblage/​bricolage; (iii) mobilities/​mutation; (iv) interpretation/​localisation; and (v)  trial and error.

72

Between policy failure and policy success

One matter of definition: ‘Failure is the mirror image of success: A policy fails if it does not achieve the goals that proponents set out to achieve and opposition is great and/​or support is virtually non-​existent’ (McConnell, 2010, 356). The chapter proceeds in four remaining sections. The next section provides a short overview of general developments in the policy diffusion and transfer literatures. The third section discusses an area that has captured analytical attention –​ the view that policy transfer often does not work or goes awry. The fourth section discusses negative lesson-​drawing, and the roles of intermediaries. The fifth section addresses the current fashion that messes the hard distinction of an uncontested or unmediated bilateral diffusion from A  to B or an unchanged policy idea or instrument X with ideas of policy assemblages, mobilities and the multi-​scalar dimensions of transfer, especially the tendency of policies to mutate as they travel. The sixth concluding section returns to the dualism encapsulated in the study of both policy failure/​success and in policy diffusion/​transfer studies as attempting to impose a rational order over the reality of chaotic and messy policy processes.

The diffusion of innovation and transfer of policies Over the past four decades a substantial body of literature has emerged concerning the related concepts of policy transfer and policy diffusion. What is important, however, in the past decade plus, is that there has been some convergence between the diffusion and transfer literatures with the borrowing of concepts and the conflation of terms. The older concept of policy diffusion grew out of American political science (especially Walker, 1969) and, in particular, analysis of how various American states would copy or emulate innovations and policy developments in neighbouring states (see for example, Berry and Berry, 1999). Diffusion has been defined as ‘the process by which an innovation is communicated through certain channels over time among members of a social system’ (Berry and Berry, 1999, 171). Diffusion describes a pattern of sequential adoption of a practice, policy or programme. The ‘diffusion’ literature suggests that policy disperses into the decision-​making atmosphere or diffuses into the ‘climate of opinion’ where ideas are picked up. In other words, specific policy approaches or instruments are rather more contagious than they are consciously chosen. According to Berry and Berry, four forces may create diffusion patterns:

73

Policy Learning and Policy Failure

• a national communication network among state officials; • states are influenced by geographically proximate neighbouring states; • leader states pioneer the adoption of a policy that ‘laggard’ states subsequently follow; and • national government is a vertical influence for prompting emulation. (Berry and Berry, 1999,  172–​8) International relations (IR) scholars have since expanded upon the American-​centric origins of the approach to discuss the diffusion of ideas internationally (Dobbin et al, 2007). IR and policy scholars also introduce a further force behind diffusion  –​competition between countries (and companies) –​propelling the adoption of international norms (such as democratic or human rights norms) and ‘best practices’ (such as via OECD peer review processes: see De Francesco, 2014; McNutt and Pal, 2011) in order to remain competitive or to meet international standards (Sharman, 2008). The diffusion perspective tended to posit incremental changes in policy and to regard ‘the travel of ideas as a function of structural forces, such as industrialisation, globalisation and regionalisation, rather than the work of free agents’ (Mukhtarov, 2014, 73). However, the approach has little to say about how policies or practices are altered during processes of adoption. By identifying patterns of policy adoption, diffusion approaches are more concerned with the conditions for the spread of knowledge or norms rather than the substantive content of new policies. As such, there is less concern within this school as to whether policy has been successful or not; the pattern of policy diffusion is expected to be of a variable character. In the main (but not exclusively), policy transfer studies are a European literature that came to prominence in the early 1990s (for example, Evans and Davies, 1999; Page, 2000; Rose, 1993; Stone, 1999; 2004). There has been considerable assessment of policy convergence within the European Union (see for example, Meyer-​Sahling, 2011; Parrado, 2008) and increasingly beyond European Union (EU) borders (Casier, 2011; Tews, 2009; Vögtle and Martens, 2014). The strength of the policy transfer literature has been to focus on decision-​making dynamics internal to political systems and to address the role of agency, and processes of learning, in transfer processes. The emphasis has been upon the scope for choice in selection of policy ideas that are to be transferred, the interpretation of circumstances or environment, and (bounded) rationality in imitation, copying and modification by decision-​makers.

74

Between policy failure and policy success

Transfers can be either voluntary or coercive, or combinations thereof (Dolowitz and Marsh, 2000, 13–​17). Descriptive labels such as ‘lesson-​ drawing’ (Rose, 1993), ‘systematically pinching ideas’ (Schneider and Ingram, 1988) or ‘donor’, ‘lending’ and ‘borrower countries’ (Stead et al, 2008) have cast transfer as a voluntary activity involving choice and deliberation. Other terms, somewhat older and suggestive of the then prominence of aid conditionality and structural adjustment policies of the 1980s and 1990s, emphasise compulsory conformity, that is: ‘penetration’ (Bennett, 1991) and ‘external inducement’ (Stone, 1999; also Clifton and Díaz-​Fuentes, 2014). In this latter depiction, ‘policy transfer’ is directly connected with the contested politics of who gets what policy. Policy transfer can involve a number of processes. First, the objects of transfer can include (i) policies, (ii) institutions, (iii) ideologies or justifications, (iv) attitudes and ideas, and (v) negative lessons (Dolowitz and Marsh, 1996). Second, transfer can take place across time, within countries across policy sectors as well as across countries. The public policy literature has tended to concentrate on the different modalities of transfer, as well as specific policy instruments and/​or professional practices, whereas the IR literature has been stronger on the diffusion of norms that can promote learning and the building of policy consensus. IR thinking has influenced policy scholars now developing concepts of ‘transgovernmentalism’ (Legrand, 2015) and ‘transnational policy coordination’ (Vögtle and Martens, 2014) which stress experimentalism. Third (and an important point, given the theme of this edited volume), the early literature recognised different degrees of transfer: straightforward copying of policy, legislation or techniques as well as various forms of emulation, synthesis and hybridisation and inspiration (Dolowitz and Marsh, 1996, 351).This recognition of a spectrum of the intensity and extensity of transfer processes automatically qualifies and constrains assumptions of easily identifying instances of failed transfer. This is especially the case when negative lessons are drawn from experience elsewhere and contribute to divergent outcomes. The policy transfer literature also allows us to see the possibilities for a general convergence around broad policy objectives and principles but at the same time, differentiation and divergence with regard to the instruments adopted, type of legislation enacted or specific institutional modes of policy control/​delivery. Recognition of these multi-​layered dynamics attests to some of the conceptual difficulties in evaluating and determining instances of policy transfer failure or otherwise. The next three sections delve further into how policy transfer failure(s) might be interpreted.

75

Policy Learning and Policy Failure

In general, the policy transfer literature has matured and consolidated into a sizeable body of analysis. Indeed, due to the dynamics of globalisation and increasingly integrated political-​economies, it is often suggested that policy transfer is ‘on the rise’ as an empirical phenomenon (Davis, 2009). This has generated a number of reviews and assessments concerning ‘future directions’ in the study of policy transfer (see for example, Benson and Jordan, 2011; Marsh and Evans, 2012; Shipan and Volden, 2012). One such assessment suggests that future research could explore cognate ideas of ‘policy success’ (Dussauge-​Laguna, 2012b; Mukhtarov, 2014). Indeed, there are a number of commentators in this field who suggest that there has been an oversight or gap in the literature with a relative lack of analysis linking policy transfer processes with outcomes (Fawcett and Marsh, 2012, 163). However, in linking policy transfer to policy failure the study of policy transfer becomes ‘the object of debate rather than facilitating analyses of the social processes that constitute policy transfer’ (McCann and Ward, 2012, 327). Instead of treating policy transfer as the dependent variable –​a process that needs to be understood –​if treated as the independent variable, the focus is on the relationship between policy transfer/​diffusion and policy outcomes (Marsh and Evans, 2012, 589).

Truncated transfer and ‘dud’ diffusion Dolowitz and Marsh (2000, 17) argue that policy failure is more likely if the transfer is uninformed and/​or incomplete and/​or inappropriate. Uninformed transfer happens when policies are transferred with insufficient knowledge about the extent to which, and why, it works in the jurisdiction from which it is being transferred. Incomplete transfer results when some features of a policy are transferred, but others are not, and the success in the original jurisdiction depended at least in part on the feature(s) not transferred. Finally, inappropriate transfer occurs when the contextual factors –​cultural, political economic –​are very different, which leads to differences in policy outcomes in the two countries concerned (see Fawcett and Marsh, 2012 for an application). The ‘one size fits all’ era of structural adjustment policies of the international financial institutions immediately comes to mind. The financial conditionality of the International Monetary Fund (IMF) and the World Bank practised in heavy-​handed fashion, in the latter half of last century, represented modes of direct and indirect coercion of client countries to conform to the precepts of the Washington Consensus. Of the three dynamics, the criticism most frequently levelled at the

76

Between policy failure and policy success

development experts and economists of the IMF and the World Bank is that they sponsored a combination of ‘incomplete’ and ‘inappropriate transfers’. For example, of the three key tenets of the Washington Consensus –​deregulation, liberalisation and privatisation –​achieving the (numerous) desired outcomes of privatisation policies is very much hinged to the ‘rule of law’, especially strong property rights, and a sophisticated legal architecture that was very often not transferred nor given sufficient time to grow in developing countries or transition states (Lendvai and Stubbs, 2009; Stiglitz, 2000). Instead, rampant corruption and other unintended outcomes emerged for many countries. In this example, inappropriate or uninformed transfer is the result of erroneous thinking among policy communities in the sending organisation, or the exemplar nation. This is in distinction to the idea developed in the next section that suggests problems also emerge within the processes of transmission and in importation. Policy transfer is inevitably limited or ‘truncated’ due to the dynamics of ‘bounded learning’ (Meseguer, 2005). The cognitive processes of engaging in lesson-​drawing may well be flawed. That is, in terms of what is ‘psychologically proximate’ (Rose, 1993) or relevant information that is ‘near-​to-​hand (in geographical, cultural or historical terms)’ (Meseguer, 2005, 72). British agents of lesson-​drawing are more inclined to look towards North America, the EU and certain parts of the Commonwealth for their lessons. This arises for reasons such as habits of mind, the ‘special relationship’, the historical legacy of empire, or the ease of looking towards other English-​speaking countries (Legrand, 2015). There are in-​built cultural prejudices towards certain jurisdictions that can lead to the most appropriate lessons being overlooked or dismissed (Stone, 1999, 57). These would be cases of ‘missed lessons’ (Moynihan, 2006). Related to ‘inappropriate’ and ‘uninformed’ transference is the practice or prospect that ‘unsuccessful policies’ are successfully transferred. For example, a recent study highlights how ‘pay-​for-​performance’ (PFP) –​a popular management approach that came out of the business sector and was adopted as a centre piece of the 1978 United States (US) Civil Service Reform Act –​has been assessed extensively as largely unsuccessful in the federal government (with problems in PFP also recognised in the private sector). Yet, PFP continues to be adopted by governments in Europe, the United States and Australia (Park and Berry, 2014). Acting at an opportune moment of widely perceived problems in performance appraisal and reward within government, policy entrepreneurs advocated their innovative policy solution in a ‘garbage-​can’ decision-​making dynamic;1 that is, pre-​existing solutions were attached to a ‘new’ problem or issue.

77

Policy Learning and Policy Failure

While this diffusion from private to public sector, thence to other countries can be represented as a successful process of transfer, ‘the problems of private sector PFP were diffused to the public sector’. In short, PFP diffusion exemplifies the lack of evidence-​based decision-​ making with ‘policy adoption based on myth rather than fact’ (Park and Berry, 2014, 763). Finally, the idea that policy transfer ‘fails’ or is ‘unsuccessful’ is wedded to a rationalist epistemology of certainty that gives little credence to, or valorisation of, ‘error’ (Little, 2012). Much of the transfer literature ‘has at its core a conceptualisation of transfer agents as optimising, rational actors, who know what they are after and scan ‘the market’ for possible solutions, making decisions and trade-​offs over which policy products to adopt, albeit on the basis of imperfect knowledge’ (McCann and Ward, 2012, 327). Relaxing such assumptions allows for greater appreciation of experimentation and ‘trial-​and-​error’ in policy development.

The wrong lessons (and whose lessons?) for policy formulation The section has two parts. First, it argues that ‘negative lesson-​ drawing’ has been inadequately addressed. Second, there are many lessons advocated by various groups and stakeholders, yet only some or one are selected. Considering the dynamics of both ‘selection’ and ‘non-​selection’ brings into analytical sight to a much greater degree, the importance of power and politics, interests and intermediaries, in determining which lessons are adopted or not. Lesson-​drawing is not a politically neutral exercise. The value of lessons lies in their power to bias policy choices. This focus can also highlight an unseen dimension of ‘failure’ –​related to the forms of ‘inappropriate/​uninformed transfer’ discussed earlier –​ of a process of deliberate ‘dysfunctional’ transfer occurring as a result of rational responses or explainable circumstances. That is, ‘normative mimicry, or market pressures, whereby over-​committed policy makers have responded to complexity and crisis by unreflectively cutting and pasting from foreign models’ (Sharman, 2010, 623; also Moynihan, 2006). This is a form of ‘satisficing’ where lessons are a ‘symbolic act whereby politicians seek to enhance their status, credibility or “modernity”’. Compared to learning, or even ‘bounded learning’, this mimicry or emulation is ‘blind’ as ‘it does not entail enhanced reflection’ (Meseguer, 2005, 79). Emulation is ad hoc and piecemeal, reflective of the transfer of rhetoric and ideology. By contrast, policy

78

Between policy failure and policy success

learning may result in a more coherent transfer of ideas, policies and practices (Stone, 1999, 56). This phenomenon is in distinction to the dynamics outlined in the previous section, which emphasised processes of outward diffusion and policy export. This section shifts the focus of where the sources and causes of failings are identified; that is, resting within the importing jurisdiction. Negative lesson-​drawing The concept of ‘negative lesson-​drawing’ has been highlighted as an important reflexive dynamic in the literature. However, there has been relatively little analysis of how, when and why it happens (see Leiber et al, 2015). Applying this idea is methodologically difficult. There is rarely a policy output connected directly with a decision not to emulate –​nevertheless it occurs. Rather than policy failure per se, it is a quest to avoid policy failure where policy learning is not synonymous with policy adoption. Instead policy lessons can help crystallise what ideas and policy paths decision-​makers do not wish to follow. Many policies are ‘simply not “transferable” since they have grown out of the legal, educational and social systems of their host state’ (Hulme, 2005, 423) and are neither ideologically nor culturally proximate. Yet, there is extremely limited empirical investigation of negative lesson-​drawing. Some exceptions include one study of Britain drawing negative lessons concerning the US Freedom of Information Act (Bennett, 1991); or Canadian policy actors observing 30 years of US heavy regulatory controls on endangered species to deviate towards legislation favouring voluntary stewardship (Illical and Harrison, 2007). As noted in the latter study, ‘positive lesson-​drawing will tend to dominate as an issue reaches the political agenda, but that activists will compete by employing negative and positive lessons at the policy adoption stage’ (Illical and Harrison, 2007, 372). Negative lessons can have symbolic value and power in derailing the proposals of opponents. Elsewhere it has been suggested that negative lesson-​drawing is more prone to take place with regard to normative policy matters (such as gender mainstreaming) rather than areas related to technical issues or instrumentation (such as best practice in sewerage treatment plants) (Stone, 2004). Another perspective is that there is a tendency in many political systems to ‘overlook negative experiential learning that contradicts the policy doctrine’ (Moynihan, 2006, 1029). Alternatively, ‘negative lesson-​drawing’ may be too benign as a term for circumstances characterised by asymmetric interdependence, in which the tactics and

79

Policy Learning and Policy Failure

strategies of policy resistance by ‘subordinate’ recipient actors are glossed over (Bache and Taylor, 2003). Coercion also suggests the absence of learning and the greater prospect of inappropriate transfer. Indeed, states and organisations have learnt the hard way: over the past two decades (since the beginning of the Wolfensohn presidency) the World Bank has engaged in a policy correction to the ‘one size fits all’ approach to structural adjustment policies based on a Western neoliberal model of economic development. Increasingly appreciated with policy communities of development specialists is that local elites need to be ‘in the driving seat’ of development. That is, ‘active appropriation’ by national policy elites to purchase foreign ideas is required for international ‘best practice’ or overseas models to be effectively transplanted and take root (Stiglitz, 2000, 33). Instead of overt pressure through conditionality on loans (which still occurs), the World Bank and the other multilateral development banks have engaged in persuasion: training, conferences, secondments and other forms of knowledge transmission to inculcate local policy elites into dominant development norms. This is the ‘soft-​ side’ of policy transfer that builds common understanding and local bases of support for the transferral of policy reforms. For ‘norm brokers’ to be effective there must also be ‘norm takers’ (Acharya, 2004; see next sub-​section). In other words, for policy transfer and ideational influence to eventuate, specific institutional mechanisms for learning or persuasion need to be developed. Moreover, these policy ideas are problematically dependent on a receptive environment. This recognition represents a shift in analytical vantage point from whatever is policy transferred  –​the idea, organisation, instrument or policy tool –​as the main source of explanation, one inevitably propelling change, to an explanatory position that highlights the inherent uncertainty and politicking in the acceptance of transferred policy as more relevant and more persuasive in accounting for policy adoption and change. A virtue of focusing on ‘policy failure’ is not only a search for explanation or understanding of what did not go to plan. Policy failure also prompts questions about key actors and interests who were not incorporated into decision-​making or implementation processes. That is, the politics of exclusion. Intermediaries and interests In its most simple understanding, policy transfer is assumed to be an official process of relatively unmediated transmission of a policy

80

Between policy failure and policy success

approach between exporting and importing jurisdictions. Politicians, policy makers, bureaucrats and government appointed experts ‘do’ the transfers. Indeed, they are crucial to ‘hard’ transfer; that is, the legislative adoption and enforcement of a policy approach developed elsewhere. However, the agents of lesson-​drawing and policy transfer are a much broader category of individuals, networks and organisations. Key actors in the mechanics of policy transfer are non-​state actors such as interest groups and NGOs, philanthropic bodies, think tanks, consultancy companies, law firms and professional associations. They are engaged in ‘soft’ transfer  –​the spread of ideas and diffusion of knowledge, which is essential for providing the norms, evidence and (social) scientific understandings as to ‘why’ it makes bureaucratic and political sense to transfer policy. In particular, a number of studies have highlighted the role of ‘epistemic communities’ operating through professional associations and think tanks, seeking advisory positions in government and international organisations, to provide the ‘cause and effect’ rationales and ‘consensual knowledge’ behind their active diffusion of policy or standards (see for example, Ladi, 2005). The epistemic community approach has a specific representation of the role of knowledge or science as being based on facts and empirically discernible realities (Haas, 1992). Consensual knowledge takes the form of concrete knowledge of the physical world, objectively beholden by an epistemically privileged Cartesian observer (and collectively the epistemic community) who then turns into a dispassionate advisor to the powerful. It is a rationalist, technocratic approach to decision-​making and implies policy linearity with experts editing or re-​shaping knowledge in uni-​directional movements from basic to applied science, from problem to solution, from theorists to ‘enlightened’ policy makers. Analysis from the epistemic community perspective co-​joined with policy transfer or diffusion studies consider that solutions to problems can be found by utilising the correct knowledge and evidence. ‘Truth speaks to power’: so goes the famous phrase coined by Aaron Wildavsky (1987). Rarely is knowledge or expertise regarded as so pure and uncontested:  experts, professional analysts and their organisations are better regarded as engaged in a contest to define the truth. Not only epistemic communities, but other kinds of expert groups and interpretative communities require political patronage in order for ideas or science to become policy relevant. Although often a powerful force, (social) science is not inherently or automatically persuasive in policy debates. Indeed, the experts and evidence base that gain prominence in (transnational) policy communities may be the result of political

81

Policy Learning and Policy Failure

rather than scientific protocols where knowledge actors are used for purposes of ex post facto justification of predetermined courses of policy action. Instead of truth speaking to power, power decides what is true. Governments, particularly from stronger and more policy autonomous OECD states, decide what policy lessons are appropriate or not (Clifton and Díaz-​Fuentes, 2014; De Francesco, 2014). Burgeoning studies of transfer or diffusion to developing states or transition countries are notable for the way in which they query and contest assumptions of undiluted dichotomous diffusion or unmediated ‘import’ of transferred ideas (see for example, Grugel, 2007; Lendvai and Stubbs, 2007; 2009; Mukhtarov, 2014; Tews, 2009). A small body of literature is emerging on the ‘folly’ of EU policy transfer (Gorton et al, 2009).The projection of the EU model to the rest of the world has entailed an unreservedly Eurocentric conception of policy transfer as an outward diffusion of norms, practices and institutions (see for example, Börzel and Risse, 2009; and the 2012 special edition of West European Politics). However, the EU’s promotion of norms of democratisation has been critiqued for assuming a ‘positive identity relationship’ between European and Latin American political and policy elites (Grugel, 2007). As noted earlier, ideas and policies are only likely to be transferred successfully if there are ‘norm-​takers’ who adopt and implement them. The local context and dynamics within the importing jurisdiction is crucial in deciding which, if any, ideas are adopted (Lendvai and Stubbs, 2007). Yet, ‘intersubjectivity in communication’ means that ‘even when senders attempt to spread truthful, high quality information, the receiver may interpret the message differently from how the sender intended’ (Park et al, 2014, 399). The EU has also sought to export its model of regionalisation as the most developed and sophisticated model available as a template for other emergent regions. However, the degree and type of ‘policy transfer’ and institutional emulation from Europe that has actually taken place within East Asia, for instance, has been remarkably limited. Instead, it is characterised by an instrumental selective uptake of instruments and deviation to an ‘Asian Way’ or a model of regionalisation that is prefigured by East Asia’s distinctive political, economic and strategic history (Beeson and Stone, 2013). There has been a tendency in the analytic focus of much of the European scholarly discussion to evaluate the success and effectiveness (or not, in many assessments) of Asian regional integration in terms of ‘hard’ institutional transfer that requires a mirroring of the de jure legalistic processes characteristic of the EU model. This leads to a short-​sightedness of other policy transfer dynamics. On the one hand

82

Between policy failure and policy success

were ‘soft’ transfers of European norms and ideas that were subject to selective and negotiated uptake in transformative processes of translation. And on the other hand, instead of an unmediated and mechanical bilateral exchange from A to B –​Europe to Asia –​policy transfer processes also involved learning from many exemplars to take away a multiplicity of lessons. Notwithstanding the increasingly sophisticated comprehension of the vagaries of power games, or policy trajectories dealt by path dependencies or the various modalities of social learning now identified in many policy diffusion and transfer studies, the majority of such studies still reflect a realist ontology. That is, ‘policy’ exists as a kind of ‘package’ ready and able to be transplanted or transferred from one setting to another. The policy transfer literature… tends to work through binary oppositions:  either policy is institutionalised in another place or resisted; it either ‘fits’ or it does not fit; it is picked up by institutions or actors or it is blocked by veto players and/​or at institutional veto points. Crucially,…the literature is still dominated by a rather linear, institutionalist, perspective. (Lendvai and Stubbs, 2009, 677) As discussed in the next section, the concepts of ‘policy translation’ and ‘norm localisation’ provide a more nuanced account of the social relations of ‘assemblage’, ‘experimentalism’ and ‘bricolage’ in the mobility of policy (de Jong, 2013).

Policy translation Whether ‘transfer’ can be said to have been successfully accomplished is qualified by the reality and extensiveness of hybridity, synthesis, tinkering with models and adaptation that takes place when policies are moved from one place to the next. The use of words such as ‘transfer’, ‘transmission’ and ‘export’ become highly questionable when a transferred policy tool or institution from Country A looks completely different in Country B (and then again in Countries C and D) and when it is operationalised in substantively different fashion than originally conceived. The more frequent patterns of divergence and hybridisation, adaption and mutation are giving greater credence to the idea of policy ‘translation’ (Prince, 2010, 173) and ‘variation’ (Newburn, 2010) than to the more restricted idea of policy transfer. Taking this perspective is disruptive of linear thinking for it,

83

Policy Learning and Policy Failure

upsets the often implicit assumption that policies emerge fully formed in one particular place and then sometimes move, whole and unchanged, across space. They do not. It also troubles the idea that policies are internally coherent, stable ‘things’. They are not. An assemblage is always in the process of coming together and being territorialised just as it is always potentially pulling apart and being de-​territorialised. (McCann and Ward, 2012, 328) This stance not only takes attention away from binary assessments of ‘success’ or ‘failure’ in policy transfer, it also departs from the notion that transfers are ‘inappropriate’, ‘incomplete’ or ‘uninformed’. Rather, in complex societies the occurrence of unintended consequences, ineffective intentions, and misinterpretations of the message provide equally valuable insights into policy development, because they are part of the continuous metamorphoses that policies encounter (Little, 2012, 9). This ‘morphological understanding’ of policy transfer is not one of ‘failure’ but one of ‘trial and error’: ‘it is through error that we learn about the phenomena we are addressing and it is in the recognition that we have erred that we create space to conceive differing ways of making sense of the issues we address’ (Little, 2012, 11). It is one starting point for innovation (Parrado, 2008). This is part of the process of policy ‘translation’. Adopting a morphological stance of policy ideas in transmission allows us conceptual space for a valorisation of translation and interpretation. Translation is ‘a series of interesting, and sometimes even surprising, disturbances that can occur in the spaces between the “creation”, the “transmission” and the “interpretation” or “reception” of policy meanings’ (Lendvai and Stubbs, 2007, 175). Such approaches are critical of the rationalist underpinnings of early transfer approaches and instead stress the complexity of context (see for example, McCann and Ward, 2012; Newburn, 2010) and the need for interpretation in the assemblage of policy (Prince, 2010). Policy translation represents a ‘move away from thinking of knowledge transfer as a form of technology transfer or dissemination, rejecting if only by implication its mechanistic assumptions and its model of linear messaging from A  to B’ (Freeman, 2009, 429). Within the IR literature, this dynamic is more frequently labelled ‘norm localisation’. That is, local actors ‘do not remain passive targets and learners as transnational agents, acting out of a universal moral script…[instead] local agents also promote norm diffusion by actively borrowing and modifying transnational norms in accordance with

84

Between policy failure and policy success

their preconstructed normative beliefs and practices’ (Acharya, 2004, 269, author’s insertion). Multiple sources of lessons, combined with endogenous policy learning, also alter norm brokerage and policy transfer aspirations into a multi-​faceted translation dynamic. For example, ‘gradualism’ and ‘eclecticism’ is said to distinguish the Chinese practice of transferring policy ideas and institutions from examples observed elsewhere in the world. Unlike other (post) Communist countries, policy makers in the Chinese political and socio-​ economic systems did not face regime collapse, and the institutions of the past were not effaced but persisted. Instead, like the Japanese in the previous century (see Page, 2000), the Chinese have considered foreign policy lessons via a ‘tradition of cobbling together various foreign and domestic policy ideas in modular fashion’. In a cautious and selective process, ‘these were then reassembled onto existing institutional frameworks’ and are reflective of ‘a more generic Chinese tradition of institutional bricolage’ (de Jong, 2013, 89). This is by no means exceptional to Asian policy contexts nor even confined to a sovereign state’s actors. Intermediaries such as scholars, policy thinkers and opinion leaders ‘mutate’ policy ideas from elsewhere in the professional spaces and policy communities between exporting and importing jurisdictions. A key element of what they do is piecing scraps of knowledge together. That is, ‘assemblage’ is policy assembly through interpretation of different items of information and experience, often creating something new, that is, hybrids from what they have acquired second-​hand (de Jong, 2013). The places for these processes of synthesis and adaptation are varied but can take place in the conferences, journals and professional engagements of scientific communities and policy analysts (Mukhtarov, 2014, 77). Policies are not merely transferred over space, but their formats and their effects are also transformed by their journey through professional communities, and through time. Temporal transfers Policy borrowing evolves over time in disjunctures dependent on the legacies of the past, or ‘policy windows’ thrown open by electoral cycles or other events, or in the sequencing of adoption and implementation (Dussauge-​Laguna, 2012a). Notwithstanding the comments in the previous section concerning the EU, there have been many achievements along the path of what is a long-​term inter-​ generational project. The EU has oft been described as a ‘laboratory for policy transfer’ for increasing, albeit gradual, convergence among

85

Policy Learning and Policy Failure

member states. EU processes of regionalisation propel patterns of increasing similarity in economic, social and political organisation between countries. Member and candidate states converge around harmonising policies: structural funds, cohesion funds and the acquis communautaire. The European Commission is a top-​down influence for compliance through directives and regulations as well as joint progress on policy through the Open Method of Coordination. This process of ‘EU-​isation’ (rather than the broader social and cultural process of ‘Europeanisation’) is a combination of coercive measures and voluntary harmonisation. Back-​sliding in the European Union among new member states bring into high relief the temporal considerations (see the essays in Batory et al, 2018). At the time of accession of new member states in 2005, compliance to the acquis communautaire was relatively high and appeared to foretell a trajectory of greater convergence. Today in the wake of the global financial crisis and specific difficulties faced in the EU regarding the Euro and BREXIT, compliance has declined dramatically. The ‘durability’ of the EU’s civil service reforms is one area where patterns of deviation are now observed: the post accession behaviour of some states has seen reversals of reforms or new reorientations in the absence of sanctions available to the Commission (Meyer-​Sahling, 2011). This lack of ‘stickiness’ over time might suggest policy transfer failure as achievements decay. However, the concept of ‘translation’ problematises any quick assessment that regression is occurring. Instead, it reorients the vantage to one where displacement, dislocation, transformation and negotiation is the normal and constant state of play (Lendvai and Stubbs, 2009, 676). Even if there are cases of straightforward transmission of policy from one jurisdiction to another, the transfer does not create a cryogenically preserved policy forever more. At some point, the policy transfer process ends and endogenous forces of mutation take over. Local ownership becomes more pronounced and the ‘indigenisation’ of policy results. Logics of appropriateness entail a gradual adjustment and modifications that lead to different outcomes than may have originally been envisaged. Existing policy processes and socio-​cultural conditions alter imported ideas. What once may have been a foreign idea becomes local practice (Batory et al, 2018).

Conclusion Just as ‘policy failure’ and ‘policy success’ are often portrayed as polar opposites in a binary distinction, so too policy transfer and diffusion

86

Between policy failure and policy success

approaches can suffer from similar binary distinctions. That is, the movement of policy idea, instrument or practice from jurisdiction A to B or from one innovating organisation or political community to the next. Accordingly, rather than working in a framework where ‘success and failure are bound inexorably with each other’ (McConnell, 2010, 346), the concern in this chapter has been to escape from the idea of these dual tendencies to evoke the metaphor of policy translation as an experimental process in constant policy motion turning between innovation and reaction, compliance and invention. In sum, asking if policy transfer fails (or is ‘inappropriate’ or a poor ‘fit’) is in many respects the wrong question for the phenomenon. Instead, divergence is expected:  policy translation  –​ characterised by fluid multi-​actor processes of interpretation, mutation and assemblage –​is the constant reality. To conclude, it is worth revisiting Aristotle’s three approaches to knowledge and how it was adapted to policy studies by Flyvberg (2001). Policy translation is not always the result of a directed process of policy learning instigated by policy makers. Indeed, translation can be an analytically rational or ‘epistemic’ process of learning driven by reform-​minded bureaucrats, experts and politicians. Yet, policy translation can also be a more haphazard dynamic. Policy translation as ‘bricolage’ involves tinkering with existing local as well as borrowed policy practices to construct new or hybrid policy formations. This is a creative process that is learning of a quite different nature –​one that is the art and craft or ‘techne’ of policy. Finally, policy translation is also a form of learning that concerns prudence or ‘phroenesis’; that is, pragmatic, variable, context dependent and based on a practical value-​rationality of localisation or negative lesson-​drawing by making judgements about what is desirable policy from elsewhere. Viewing policy translation as a combination of art, episteme and judgement entails a different set of reflections upon policy diffusion and transfer: it means that we will never see some form of perfect ‘cloning’ of a policy between different places. Instead, policy translation embraces deviation and difference. If policy transfer is to be understood properly, it is as an open-​ended process. This stance consigns suggestions that we can talk normatively about ‘failed policy transfer’ to the dustbin. Acknowledgements This work has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 693799 as part of the ‘European Leadership in Cultural, Science and

87

Policy Learning and Policy Failure

Innovation Diplomacy’ (EL-​CSID) project. It does not necessarily reflect the opinions of the EU. Note 1 In the garbage-​can model, decision-​making is portrayed as a highly unpredictable and ambiguous process. Actors define goals and choose means as they go along. Organisa tions such as national ministries and executives do not have goals in the rational sense, but define them in the process of attaching problems to pre-​existing solutions which may not be the best solutions (see Park and Berry, 2014, 775; also Dussauge-​Laguna, 2012a).

References Acharya, A, 2004, How ideas spread: Whose norms matter? Norm localization and institutional change in Asian regionalism, International Organization 58, 239–​75 Bache, I, Taylor, A, 2003, The politics of policy resistance: Reconstructing higher education in Kosovo, Journal of Public Policy 23, 3, 279–​300 Batory, A, Cartwright, A, Stone, D (eds) 2018, Policy experiments, failures and innovations:  After accession in Central and Eastern Europe, Cheltenham: Edward Elgar Beeson, M, Stone, D, 2013, The European Union model’s influence in Asia after the global financial crisis, European Journal of East Asian Studies 12, 2, 167–​90 Bennett, CJ, 1991, Review article: What is policy convergence and what causes it? British Journal of Political Science 21, 2, 215–​33 Benson, D, Jordan, A, 2011, What have we learned from policy transfer research? Dolowitz and Marsh revisted, Political Studies Review 9, 3,  366–​78 Berry, FS, Berry, WD, 1999, Innovation and diffusion models in policy research, in PA Sabatier (ed) Theories of the policy process, Boulder, CO: Westview Press Börzel, T, Risse, T, 2009, The diffusion of (inter-​)regionalism: The EU as a model of regional integration, KFG Working Papers, Berlin: Freie Universität Berlin Casier, T, 2011, To adopt or not to adopt: Explaining selective rule transfer under the European neighbourhood policy, Journal of European Integration 33, 1, 37–​53 Clifton, J, Díaz-​Fuentes, D, 2014, The OECD and ‘the rest’: Analyzing the limits of policy transfer, Journal of Comparative Policy Analysis: Research and Practice 16, 3, 249–​65 Davis, A, 2009, False and frustrated transfer:  Spanish immigration policy and Catalonia, Policy & Politics 37, 3, 423–​38

88

Between policy failure and policy success

De Francesco, F, 2014, Transfer agents, knowledge authority and indices of regulatory quality: A comparative analysis of the World Bank and the Organisation for Economic Co-​operation and Development, Journal of Comparative Policy Analysis, 18, 4, 350–​65 de Jong, M, 2013, China’s art of institutional bricolage: Selectiveness and gradualism in the policy transfer style of a nation, Policy and Society 32, 2, 89–​101 Dobbin, F, Simmons, B, Garrett, G, 2007, The global diffusion of public policies:  Social construction, coercion, competition, or learning? Annual Review of Sociology 33, 449–​72 Dolowitz, D, Marsh, D, 1996, Who learns from whom: A review of the policy transfer literature, Political Studies 44, 2, 343–​57 Dolowitz, D, Marsh, D, 2000, Learning from abroad: The role of policy transfer in contemporary policy making, Governance: An International Journal of Policy and Administration 13, 1, 5–​24 Dussauge-​Laguna, M, 2012a, The neglected dimension:  Bringing time back into cross-​national policy transfer studies, Policy Studies 33, 6, 567–​85 Dussauge-​Laguna, M, 2012b, On the past and future of policy transfer research:  Benson and Jordan revisited, Political Studies Review 10, 3,  313–​24 Evans, M, Davies, J, 1999, Understanding policy transfer: A multi-​level, multi-​disciplinary perspective, Public Administration 77, 2, 361–​85 Fawcett, P, Marsh, D, 2012, Policy transfer and policy success: The case of the Gateway Review Process (2001–​10), Government and Opposition 47, 2, 162–​85 Flyvberg, B, 2001, Making social science matter: Why social inquiry fails and how it can succeed again, Cambridge: Cambridge University Press Freeman, R, 2009, What is translation? Evidence and Policy: A Journal of Research, Debate and Practice 5, 4, 429–​47 Gorton, M, Hubbard, C, Hubbard, L, 2009, The folly of European Union policy transfer:  Why the Common Agricultural Policy (CAP) does not fit Central and Eastern Europe, Regional Studies 43, 10, 1305–​17 Grugel, J, 2007, Democratization and ideational diffusion:  Europe, Mercosur and social citizenship, Journal of Common Market Studies 45, 1, 43–​68 Haas, PM, 1992, Introduction:  Epistemic communities and international policy co-​ordination, International Organization 46, 1,  1–​35 Hulme, R, 2005, Policy transfer and the internationalisation of social policy, Social Policy and Society 4, 4, 417–​25

89

Policy Learning and Policy Failure

Illical, M, Harrison, K, 2007, Protecting endangered species in the US and Canada: The role of negative lesson drawing, Canadian Journal of Political Science 40, 2, 367–​94 Ladi, S, 2005, Globalisation, policy transfer and policy research institutes, Cheltenham: Edward Elgar Legrand, T, 2015, Transgovernmental policy networks in the anglosphere, Public Administration 93, 4, 973–​91 Leiber, S, Greß, S, Heinemann, S, 2015, Explaining different paths in social health insurance countries:  Health system change and cross-​border lesson-​drawing between Germany, Austria and the Netherlands, Social Policy and Administration 49, 1, 88–​108 Lendvai, N, Stubbs, P, 2007, Policies as translation:  Situating transnational social policies, in S Hodgson, Z Irving (eds) Policy reconsidered: Meanings, politics and practices, Bristol: Policy Press Lendvai, N, Stubbs, P, 2009, Assemblages, translation and intermediaries in South East Europe, European Societies 11, 5, 673–​95 Little, A, 2012, Political action, error and failure: The epistemological limits of complexity, Political Studies 60, 1, 3–​19 Marsh, D, Evans, M, 2012, Conclusion: Special issue on policy transfer, Policy Studies 33, 6, 587–​91 McCann, E, Ward, K, 2012, Policy assemblages, mobilities and mutations: Toward a multidisciplinary conversation, Political Studies Review 10, 3, 325–​32 McConnell, A, 2010, Policy success, policy failure and grey areas in-​ between, Journal of Public Policy 30, 3, 345–​62 McNutt, K, Pal, LA, 2011, ‘Modernizing government’:  Mapping global public policy networks, Governance 24, 3, 439–​67 Meseguer, C, 2005, Policy learning, policy diffusion and the making of a new order, The ANNALS of the American Academy of Political and Social Science 598, 1, 67–​82 Meyer-​Sahling, J-​H, 2011, The durability of EU civil service policy in Central and Eastern Europe after accession, Governance:  An International Journal of Policy, Administration, and Institutions 24, 2,  231–​60 Moynihan, D, 2006, Ambiguity in policy lessons: The Agencification experience, Public Administration 84, 4, 1029–​50 Mukhtarov, F, 2014, Rethinking the travel of ideas: Policy translation in the water sector, Policy & Politics 42, 1, 71–​88 Newburn, T, 2010, Diffusion, differentiation and resistance in comparative penality, Criminology and Criminal Justice 10, 4, 341–​53

90

Between policy failure and policy success

Page, E, 2000, Future governance and the literature on policy transfer and lesson drawing, Paper presented to the ESRC Future Governance Programme, Britannia House, London, 28 January Park, S, Berry, F, 2014, Successful diffusion of a failed policy: The case of pay-​for-​performance in the US federal government, Public Management Review 16, 6, 763–​81 Park, C, Wilding, M, Chung, C, 2014, The importance of feedback: Policy transfer, translation and the role of communication, Policy Studies 35, 4, 397–​412 Parrado, S, 2008, Failed policies but institutional innovation through ‘layering’ and ‘diffusion’ in Spanish central administration, International Journal of Public Sector Management 21, 2, 230–​52 Prince, R, 2010, Policy transfer as policy assemblage: Making policy for the creative industries in New Zealand, Environment and Planning A: Economy and Space 42, 1, 169–​86 Rose, R, 1993, Lesson-​drawing in public policy: A guide to learning across time and space, Chatham, NJ: Chatham House Schneider, A, Ingram, H, 1988, Systematically pinching ideas:  A comparative approach to policy design, Journal of Public Policy 8, 1,  61–​80 Sharman, JC, 2008, Power and discourse in policy diffusion: Anti-​ money laundering in developing states, International Studies Quarterly 52, 3, 635–​56 Sharman, JC, 2010, Dysfunctional policy transfer in national tax blacklists, Governance 23, 4, 623–​39 Shipan, CR, Volden, C, 2012, Policy diffusion:  Seven lessons for scholars and practitioners, Public Administration Review 72, 6, 788–​96 Stead, D, de Jong, M, Reinholde, I, 2008, Urban transport policy transfer in Central and Eastern Europe, The Planning Review 44, 172,  62–​73 Stiglitz, J, 2000, Scan globally, reinvent locally:  Knowledge infrastructure and the localisation of knowledge, in D Stone (ed) Banking on knowledge:  The genesis of the global development network, London: Routledge Stone, D, 1999, Learning lessons and transferring policy across time, space and disciplines, Politics 19, 1, 51–​9 Stone, D, 2004, Transfer agents and global networks in the ‘transnationalisation’ of policy, Journal of European Public Policy 11, 3,  545–​66 Stone, D, 2012, Transfer and translation of policy, Policy Studies 33, 6,  483–​99

91

Policy Learning and Policy Failure

Tews, K, 2009, From law-​taking to policy making: The environmental dimension of the EU accession process, Environmental Policy and Governance 19, 2, 130–​9 Vögtle, EM, Martens, K, 2014, The Bologna Process as a template for transnational policy coordination, Policy Studies 35, 3, 246–​63 Walker, JL, 1969, The diffusion of innovations among the American states, The American Political Science Review 63, 3, 880–​99 Wildavsky, A, 1987, Speaking truth to power: The art and craft of policy analysis, Boston, Toronto: Little, Brown and Company

92

5

British Columbia’s fast ferries and Sydney’s Airport Link: partisan barriers to learning from policy failure Joshua Newman and Malcolm G. Bird

Introduction Policy learning, where experiences from other jurisdictions and time periods inform decision-​making, has been suggested as a way to improve policy outcomes  –​or at the very least, to improve a government’s ability to predict the outcomes of its own policy decisions (Mossberger and Wolman, 2003, 430). Policy failures might therefore seem to have an especially prominent place in the learning process, as examples of instruments and ideas to avoid. Nonetheless, episodes in which failure did not lead to lessons learned or to improved public policy are abundant. The implication of this non-​learning from failure is that there are situations in which the consequences of failure may not be a strong enough deterrent to prevent failure from re-​occurring. In this chapter, we explore one such situation, in which the incentives of partisanship can encourage a government to actively seek to exacerbate an existing policy failure rather than to repair it. Under these circumstances, the certain benefits of shaming the political opposition outweigh any potential rewards of improving specific policy outcomes. Using the cases of British Columbia’s fast ferries and the Sydney Airport Rail Link, we develop a scenario in which policy failure leads not to policy learning but rather to deliberately increased failure. While democratic governments have long been thought to endeavour to improve social outcomes, at least for particular groups or individuals (Downs, 1962), in some cases incentives can exist for governments to do more harm than good. To this end, we will examine two cases of policy failure in the late 1990s in the transportation sector. The first case explores an effort

93

Policy Learning and Policy Failure

by the British Columbia Ferry Corporation (BC Ferries), a public provider of marine transportation on Canada’s west coast, to introduce a fleet of high-​speed aluminium catamaran ferries (the ‘fast ferries’), and the second investigates a public–​private partnership scheme to build and operate an urban rail link between the central business district and the airport in Sydney, Australia (the Sydney Airport Link). While the concept of success and failure in public policy is a subject of considerable debate (see McConnell, 2010; Newman, 2014, for example), both cases presented here can be considered policy failures in that they were unable to reach the stated initial goals of the project, they cost substantially more than anticipated and they were viewed by the public and by those in the political sphere as being policy failures. In both cases, policy options were presented that had the potential to mitigate financial losses and to redirect the project back toward the achievement of stated policy objectives –​in short, there was the potential to learn from policy failure –​but these options were rejected by decision-​makers in favour of actions that did nothing for the success of the project but that did deliver some short-​term political and electoral rewards. In both cases, decisions made after significant obstacles to policy success were presented only served to increase the magnitude of the policy failure and to cement the project’s outcome. The projects were based on sound technical justifications and were defined by meaningful and achievable objectives. The fast ferries project was intended to address capacity constraints within the BC Ferries system, using ships that were smaller and faster than those that had previously made up the BC Ferries fleet. Likewise, the Sydney Airport Link was supported by engineering and planning studies (for example, Kinhill Engineers, 1994) as a way to increase capacity on the rail network, reduce road traffic along a major commuting corridor and revitalise urban development in a series of under-​served low density suburbs. However, in spite of a sound motivation, once the projects encountered difficulties, public perception of deficiency was manipulated in the political sphere and by media outlets as a justification of subsequent policy decisions. In both cases, newly elected governments categorically refused to address these policy failures in what could be interpreted as a deliberate effort to assign blame for failure to the government that had initiated the project. In British Columbia, technical problems were exaggerated by the incoming government in a successful attempt to portray the opposition as unable to deliver promised infrastructure. In Sydney, technical inadequacies could have been corrected (admittedly, at some significant additional cost), but the government opted instead to refrain from corrective

94

Partisan barriers to learning from policy failure

strategies so that the Airport Link could be used as an effective partisan wedge issue. In both cases, structural and political factors encouraged governments to make sub-​optimal decisions and prevented them from learning from failure.

Failing to learn from policy failure Examples of governments learning from policy failures in modern democracies are plentiful. Public health crises, for example, can motivate governments to improve policy –​such as the regulation of pharmaceuticals in the United States through the creation of the Food and Drug Administration (Carpenter, 2010). However, government action does not, of course, guarantee improvement:  in the quest for energy efficiency in the European transport sector, for instance, decades of government intervention have not solved persistent policy problems, such as, in this case, alleviating dependence on foreign oil supplies or allowing the EU to meet its air pollution and greenhouse gas emission targets (Steenberghen and López, 2008). In many instances, major action is only taken after people die or are seriously injured in high-​profile events, as was seen in the Walkerton disaster in Ontario, Canada, in which seven people died and over 2,000 people fell ill from drinking water contaminated by E coli bacteria (Holme, 2003). Sometimes policy failures inspire action to improve policy, sometimes they do not, and the thresholds for when the correction of policy failures will occur can appear ambiguous on first inspection. Ambiguous thresholds and seemingly capricious changes in policy direction should not, however, deter analysts from attempting to uncover underlying patterns in how governments act. According to one point of view, one of the basic tenets of democracy is that citizens expect governments to enact policies that will improve society (Hart, 1984). As democracy is ostensibly a desirable form of government (Dryzek and Berejikian, 1993) and as it is clearly possible for democratic governments to improve policy, explaining the factors that enable improvements to policy is of considerable importance. Policy analysis can thereby ‘contribute to the improvement of conditions in the real world’ (Dryzek, 1982, 312). Effective policy that leads to improvements in citizens’ lives, furthermore, serves a critical legitimising function for a government solidifying its right to political power (Zittoun, 2014). Policy learning, simply put, is when governments take action on domestic policies based on insights derived from either their own jurisdictions or from other policy actors (Bennett and Howlett, 1992; Radaelli, 2009; Stone, 1999). The term ‘learning’ implies that these

95

Policy Learning and Policy Failure

actions are deliberate and voluntary, rather than coerced by exogenous forces (Dolowitz and Marsh, 1996, 344), and that they seek to correct or enhance policy decisions based on ‘improved understanding’ (May, 1992, 333). Presumably, policy failures should present some of the best examples from which governments can learn strategies to improve policy decisions (Gray, 1996, 78), and fortunately, analyses of policy failures are in no short supply: in addition to in-​depth studies on failure (Borins, 1986), there has been attention paid to fiascos (Bovens and ’t Hart, 1996), disasters (Dunleavy, 1995) and catastrophes (Moran, 2001).There would seem to be a wealth of international data on policies gone wrong, from which governments can draw upon to inform their domestic decision-​making. Just because governments are capable of learning from failure does not, however, mean that they will necessarily do so. According to one compelling line of thought (Dur, 2001; Howlett, 2012), governments should never admit their own failures, as this would be interpreted as a sign of weakness by voters. Instead, rational governments should seek to avoid blame at all costs, perhaps even more strenuously than they should seek to claim credit for perceived successes (Weaver, 1986), to avoid signalling their incompetence to potential voters. In Canada, for example, when Prime Minister Paul Martin publicly acknowledged an embezzlement scheme (later referred to as the ‘sponsorship scandal’) that defrauded the federal government of over $300  million, his government suffered a defeat in the ensuing election in which the scandal was a major contributing factor  –​even though the Martin government was found not responsible for the scandal (Wanna, 2006). One important corollary of this argument is that if governments are averse to admitting (and in turn correcting) their own failures, then citizens can be stuck with a government’s bad decisions for as long as that government remains in power. Many theories of policy change assert that major policy transformation is associated with regime change (Sabatier and Weible, 2007; True et al, 2007, 157), as is best illustrated by the election of Margaret Thatcher in the UK (Hall, 1993). However, a change in government is not, of course, any kind of guarantee that a given policy failure will be corrected; in the United States, for example, 35 years of energy policies enacted by numerous governments failed to produce a reliable and sustainable source of alternative energy (Grossman, 2009). However, if it is at all possible to learn from policy failures, the implication from the argument previously discussed is that learning is more likely to occur after a change in government. In summary, then, democratic governments do have the capacity to change policies, and therefore they have the ability to learn from policy

96

Partisan barriers to learning from policy failure

failures. However, because of the imperative to avoid blame in order to retain the confidence of voters, it is unlikely that a government will admit that a given policy has failed. Without the acknowledgement of failure, recovery from failure cannot occur. Therefore, one would expect learning from policy failure only to be attempted after the election of a new government. Nonetheless, a change in government is no guarantee of policy change, and policy failures can linger in some cases for decades or more. In fact, reverse incentives may exist that inhibit incoming governments from learning from their predecessors’ failures. Presumably, a high priority for a newly elected government would be to avoid attracting blame for a previous government’s failures. Since governments can be policy risk-​averse (Hood, 2007), they may be disinclined to engage in risky ventures to address policy failures that may result in garnering blame for something that they did not initiate. An alternative strategy might therefore be to ignore a previous government’s policy failures –​even those that are still unresolved –​ in order to avoid being blamed by the electorate for that failure. In this way, policy failures can persist throughout the tenures of several consecutive governments. A more extreme strategy might be for a new government to deliberately exacerbate a policy failure that was inherited from a previous government, in order to shame the political actors who may now be in opposition. While this may be antithetical to the democratic doctrine alluded to earlier (that is, that governments should strive to improve the welfare of society), in some circumstances it may be politically expedient to draw attention to failure, and the rewards of doing so may outweigh any expected rewards from attempting to rectify the problematic condition. As many have noted, the punishment accorded to failure can often be disproportionately larger than the rewards accrued to success (Hood, 2007; Weaver, 1986). Because governments can be averse to risk, some conditions would need to be obtained in order for governments to engage in activity that might deliberately exacerbate a policy failure. First, the policy would have to be integrally attached to a previous government or a current opposition party. This would need to be cemented in the public’s perception of the policy and not liable to transfer over to the current government or to other parties. Second, the failure would have to be ongoing or persistent, or else it would run the risk of disappearing from public view, in which case drawing attention to the failure might backfire or might prove to be a waste of effort. Third, there must be some potential for the failure to be made worse.

97

Policy Learning and Policy Failure

Major infrastructure projects, sometimes called ‘megaprojects’ (Flyvbjerg et al, 2003), are good candidates for exacerbation of policy failure. They are often seen as flagship projects intended to be lasting legacies for ambitious governments; when they fail, they can malign the reputation of parties and politicians for decades. In addition, infrastructure is not something that can be easily removed, so failed megaprojects are likely to be persistent (and highly visible) problems for long periods of time. The enormous costs associated with large infrastructure such as dams, bridges, tunnels, water treatment facilities, power plants and stadiums ensure that they remain in the public eye for lengthy periods of time, as the costs are often divided over multiple fiscal years. Their large-​scale and inherent complexity requires the joint efforts of a large number of institutional actors from various state bureaus and private sector organisations. The multitude of actors, the complexity of these projects and their long durations mean that consequences can be locked in over time and can further convolute the lines of accountability and, to some degree, culpability as well. And lastly, when they fail, infrastructure megaprojects always have the potential for failure to be amplified, since without remedial measures costs can spiral out of control. In what follows, two examples of infrastructure megaprojects that experienced failure are described. In both cases, governments inherited these megaprojects, and the failures associated with them, from a previous government that had initiated the project. In both cases, even when remedies existed that would have allowed some partial recovery from failure, the inheritor government opted to amplify failure in order to reap political rewards by shaming the opposition. And in both of these cases, partisan rewards acted as a barrier to learning and so prevented recovery from policy failure.

Barriers to learning from policy failure: the case studies Case 1: British Columbia’s fast ferries There is little doubt that, from a technical perspective, British Columbia’s fast ferries project was a disaster: the ships were late in delivery, well over budget, did not meet many of their promised functional expectations and were ultimately sold for a small fraction of their total production cost. From an electoral perspective, the project was also an utter disaster: The provincial New Democrat Party (NDP), which initiated and oversaw the fast ferries project to completion, won only two seats in the 2001 election and public acrimony over

98

Partisan barriers to learning from policy failure

the project was likely a contributing factor in their defeat (Phillips, 2010, 118). A host of external sources maintains a highly critical view of this project, detailing numerous deficiencies and their root causes (notably, British Columbia, 1999; Select Standing Committee on Public Accounts, 2000; Wright, 2001). Even BC Ferries, as recently as ten years after the project was completed, has been critical of the endeavour (BC Ferries, 2011, 3). It is worth noting, however, that while all observers would agree that this project was deeply troubled, the degree and source of ‘failure’ is somewhat contested, and there are significant differences between the various popular perceptions surrounding this project. Nonetheless, the fast ferries were deeply unpopular among a core (and vocal) group of citizens. In their eyes, the project was yet another example of public sector incompetence and its propensity to waste resources. However, while these views may be widespread and prevalent, this does not mean that they are accurate. Although the fast ferries had technical deficiencies, time delays, cost overruns and a problematic business case, these deficiencies, while real, have been generally exaggerated and sensationalised by the media and by the Liberal government that took power in British Columbia in 2001. In addition, the inability and unwillingness of the political sphere to make any attempts to recover from the project’s failed path is, from one point of view, a more egregious failure of the policy process than the initial technical and financial failures of the project itself. Although several Canadian provinces enjoy multi-​party competition in their legislatures, BC’s politics have long been dominated by only two political parties, one linked to organised labour movements and another that is usually composed of an alliance between conservative political interests and the business community. While these factions have undergone numerous name changes over the years, the former has been known as the NDP since the 1960s and the latter has been called the Liberal Party since the 1990s. Like many two-​party jurisdictions, electoral politics in British Columbia have been heavily polarised and competition is particularly fierce (Phillips, 2010). Since British Columbia is Canada’s most westerly province, many of its 4.6 million residents live on islands or coastal communities requiring marine-​based transport. Accordingly, marine transportation has frequently become the subject of intense political interest. With ferry passenger numbers rising in the 1980s and early 1990s, the NDP government embarked on a major capital expansion of the terminal infrastructure and ships of BC Ferries in 1994. One central pillar of this plan was to address the congestion problems that plagued

99

Policy Learning and Policy Failure

the route between Horseshoe Bay (near Vancouver) and Nanaimo on Vancouver Island. Both terminals of this route were confined due to their geographic locations, which meant that during high use times, on-​and off-​loading vehicles created serious traffic problems for local communities (BC Ferries, 1993).The fast ferries project was a proposed solution with a plan to replace larger, slower ferries with smaller, faster ships that would operate at a higher frequency, thus reducing the need for long queuing lines. This plan was also intended to revitalise British Columbia’s shipbuilding industry by developing expertise in building high-​speed, aluminium catamarans for potential international export. In June 1994, the NDP government announced a plan to build three high-​speed aluminium ferries at a projected cost of $70 million each, to be delivered one at a time in April 1996, April 1997 and October 1997 (British Columbia, 1999, 16). Despite its laudable objectives, the fast ferries project was plagued by poor governance, including limited information and analysis sharing between project managers, the BC Ferries board, central agencies and the BC Legislature (Watson, 2004). In a review of the project (and the operations of BC Ferries in general) commissioned by the provincial government in 2001, an independent financial consultant concluded that the project was driven by the province’s political executive, not the board of directors of BC Ferries, and that ‘BC Ferries remains vulnerable to influence that is inspired by decidedly non-​commercial motives’ (Wright, 2001, 10). In addition, the fast ferries’ business case had some significant weaknesses. For instance, the short distance of the route mitigated much of the value of running at the higher speeds for which the ships were specially designed. Also, there was a limited appreciation of the risks associated with a novel aluminium construction design, as well as the inherent challenges of establishing a new (aluminium) shipbuilding industry in the province. A profound sense of haste that accompanied the project, due to tremendous pressure from the government onto the project’s managers, meant that plans and designs were not fully developed and evaluated, and timelines were far too optimistic. The project lacked sufficient engineering capacity and suffered from poor management of what would turn out to be an extremely ambitious endeavour. In addition, the project’s cost estimates were off base, with the final costs much higher than original estimates (British Columbia, 1999). And finally, the finished ships were plagued by technical shortcomings: they weighed too much; they consumed twice as much fuel as conventional ferries; their operations were inhibited by high winds; they had a limited capacity for over-​height vehicles; they created

100

Partisan barriers to learning from policy failure

a large wake; and they required relatively large crews to operate and premium pay for employees. Most critically, the ships were unable to operate at the promised 37 knots and they could only travel at 32 knots during regular operations (Armour, 2001). Ultimately, the fast ferries project was brought to an end without the ships entering permanent operation in the BC Ferries fleet. The first two ships operated for a period of time in 1999 and 2000 but were taken out of service shortly thereafter. Joy McPhail, the NDP minister responsible for BC Ferries, referred to the project as a ‘failed experiment’ (Palmer, 2009). A  perception among the public grew, thanks in part to media reports and dogged efforts by the Liberal opposition, that the fast ferries project was an enormously costly NDP blunder. This perception was reinforced by internal factions within the NDP itself, as divergent opinions began to emerge after the resignation of Premier Glen Clark in 1999. Despite all of this, the fast ferries project did not have to end as it did. There were a number of viable policy recovery options available to both the outgoing NDP government and the Liberal government that was elected in 2001 that could have mitigated this failure. The first recovery option was to run the vessels at a slower speed, 22 instead of 32 knots. Reducing the operational speed would diminish maintenance and personnel requirements, as higher speeds are associated with greater wear and tear. This option would have required limited capital investment, approximately $2.5 million per ship, mainly for alterations to vehicle decks. Limited to passenger vehicles only, the ships would have had a restricted capacity, but they could have provided supplementary service on a number of routes (Armour, 2001, 1). Alternatively, there was a more robust policy option available to the government that would include reducing the operational speed of the ships, while simultaneously making significant alterations to the vehicle and passenger decks such that the ships could take all types of vehicles. Since only two of the four engines would be required for this option, the other two could be removed and sold or kept as spares, and the life of the engines would thereby be prolonged. Approximate capital costs would have been between $14 and $18 million per vessel (for a total of just under $60 million), and there was some interest from the private sector for capital investment or some form of public–​private partnership leasing arrangement (Armour, 2001, 10). Equivalent newly built traditional ferries would have cost somewhere between $65 and $95 million each (depending on where they were manufactured) and would have taken a number of years to complete and come into full

101

Policy Learning and Policy Failure

service (Armour, 2001, 10), so this retrofitting option would still have presented substantial savings while enlarging the BC Ferries fleet. These options were never pursued, however, despite being suggested publicly by prominent engineering experts. The incoming Liberal government immediately distanced itself from the fast ferries project:  within 90  days of coming to power, as promised during the election campaign, the new government launched an inquiry, even though at least two similar studies had already been completed (Beatty, 2001). Moreover, political rhetoric emanating from the Liberal leadership was especially disparaging of the project and, of course, of the NDP government that had overseen it. Liberal premier Gordon Campbell, for example, referred to the fast ferries as a ‘debacle’, and stated that ‘they simply should never have been built in the first place’ (Richmond and Villemaire, 2006, 37). The incoming government never explored any options to retrofit the ferries, opting instead to sell the ships as soon as possible. Proposals that would have put the ships into service were summarily dismissed (Bohn, 2003). In the end, the ships were sold at auction in 2003 for about $20 million resulting in a substantial loss to taxpayers (Watson, 2004). Without having been used or modified, the three ships were then re-​sold in 2009 for an estimated $50 million (Mercer, 2009). Case 2: The Sydney Airport Link New South Wales is Australia’s most populous state, and Sydney is the country’s largest urban area in terms of both geographical size and population. Because of Sydney’s sprawling cityscape, which accommodates several densely populated nodes outside of the central business district, transportation issues have been a persistent concern for many decades. Because local government in Sydney is divided among a multitude of small local councils (despite attempts at merger; see Dollery et al, 2008, for example), the responsibility for transportation issues has largely fallen to the state government. Like many other jurisdictions, New South Wales politics are dominated by two factions, in this case a centre-​left Labor Party and a centre-​right coalition between the Liberal Party and the National Party, and interaction between the two sides is particularly adversarial as compared to other similar jurisdictions. After many years of neglect, the state began efforts to address New South Wales’ deteriorating and under-​capacity transportation infrastructure in the late 1980s. Highway expansion in the Sydney area, including the Sydney Harbour Tunnel megaproject, was re-​initiated

102

Partisan barriers to learning from policy failure

at this time. With the election of Nick Greiner’s Liberal–​National coalition in 1988, the state government started to take an interest in public–​private partnerships, in which private sector investment would be sought for public infrastructure projects in return for a share of revenues from user fees. At the time, these arrangements were thought to be foolproof:  major infrastructure projects were supposed to be entirely funded by the private sector and were believed to ‘not require one cent of Government money’ (Baird, 1991). The ideas that would eventually support the Sydney Airport Link began to take shape around 1990. Initially, the plan was to connect Sydney Airport with the city’s north shore by train, including a new tunnel under the harbour. However, when it was found that this option would be too expensive for the private sector to accomplish on its own, more affordable rail routes were explored. Because there were only two private companies involved in negotiations, and because no formal public tender process had been pursued, the state government was able to ask the private bidders to combine their bids and to propose a single viable solution (Larriera, 1991). The end result, which opened for service in 2000, was an extension of an existing suburban train line to the airport and then a new rail line via tunnel to the city’s central business district. The Sydney Airport Link was designed with several goals in mind: first, it was going to provide a quick and affordable public transport option for travel to the airport. Since Sydney Airport was (and still is) Australia’s largest and most-​used airport, convenient transportation to and from the airport was an issue of critical importance. Locally, there were economic reasons for fast and reasonably priced transport to the airport, as the airport has always been a major source of employment in the area, providing jobs for several thousand people. Second, public transport to the airport was intended to reduce congestion on the highway network, which was still under expansion. And third, with stops envisioned to be built between the airport and the city centre, the Airport Link was seen as a way to revitalise several of Sydney’s neglected industrial suburbs through transit-​oriented development (New South Wales, 1993). While the Sydney Airport Link may have been a technical marvel, and although it was built according to schedule and within the bounds of its final construction budget (Jones, 2000), the project was an abject failure on a variety of fronts. First of all, despite initial claims from the Liberal–​National government that the project would be built entirely with private capital, after considerable redefinition of the project, nearly $600 million of public investment was required (Morris, 1996).

103

Policy Learning and Policy Failure

Worse still, the financial model that was used to return revenue to the private investors relied on a ‘station access’ fee that was charged to passengers using stations on the privately built portion of the line, which resulted in a fare increase of about 400 per cent over equivalent trips on the public section of Sydney’s urban rail network. In addition, the absence of special-​purpose trains meant that there was nowhere to store baggage, and it also meant that passengers disembarking from (often long, international) flights would have to compete for limited space with morning commuters on their way into the city. Taken together, the Sydney Airport Link was expensive (both in terms of public sector capital costs and passenger user fees) and ill-​ suited to its purpose. Estimated ridership levels of 46,000 passengers per day, with expected increases to 65,000 (Kinhill Engineers, 1994), were never reached. In the first three months, ridership levels were around 10,500 passengers per day (Wainwright, 2000), and by 2005, after six years of continuous service, the Airport Link still only carried about 11,500 passengers each day (Nixon, 2006). The low levels of patronage in turn meant that the project’s primary goals of traffic abatement, public transport options and local urban development could not be met. In addition, the Sydney Airport Link experienced a surfeit of failures in governance. The informal procurement process garnered attention from the state’s Independent Commission Against Corruption, which delayed the project for more than a year while an investigation was conducted (all parties were eventually exonerated and the project continued as planned once the investigation was completed –​see Jones, 2000 for a more detailed account). Once service on the Airport Link began in 2000, the very low ridership levels meant that the private sector operator was not able to earn enough money to support its debt payments –​since the station access fees were its only source of revenue. This forced the private operator of the Airport Link into bankruptcy only six months after the line opened for business. An ensuing legal battle between the bank-​appointed receivers and the (then Labor) government lasted for five years and resulted in a new partnership and revenue-​sharing arrangement that significantly lowered the public sector’s share of profit (see RailCorp, 2005). When the Labor government under Bob Carr inherited the Sydney Airport Link from its Liberal–​National predecessors, the contract with the private sector had just been signed, having been finalised only a month before the state election. As such, from the beginning, the Labor government was able to characterise the Airport Link as a Liberal–​National project in an attempt to delegate all blame for

104

Partisan barriers to learning from policy failure

the project’s many problems to the opposition. Once it was clear that the Airport Link would experience considerable financial and ridership issues, many avenues were open to the Labor government to remedy the situation. This included lending money to the private sector, subsidising the station access fees, renegotiating the public–​ private partnership contract, or negotiating directly with the project’s capital lenders for re-​financing. In addition, the government could have purchased the entire Airport Link outright, and the fact that it was sold only a year after the renegotiation of the contract in 2005 (Baker, 2006) indicates that this was a potentially economical option. However, the short-​term partisan gains that were to be had from allowing the project to fail outweighed the long-​term social rewards of finding a swift resolution, and the Labor government decided not to participate in any actions that would result in their being blamed for the project’s failures. Instead, the state Labor government and its local transportation authority, RailCorp, stood aside while the Airport Link struggled to improve its ridership levels. Financial aid was not offered in any form. When the receivers initiated legal action against the government, alleging that the public sector had neglected its responsibilities to offer in-​kind support to the Airport Link service under the terms of the original public–​private partnership contract, the government was forced to negotiate. This actually weakened the negotiating position of the government, since it would then be required to negotiate with a well-​funded and patient bank rather than with a desperate small-​time local rail operator. However, this position leant the greatest amount of support to the government’s argument that the contract signed by the previous Liberal–​National government was flawed from the beginning and was the cause of the Airport Link’s troubles. For ten years, the Labor government continued to use the Airport Link as a partisan wedge issue (see for example, Watkins, 2005).

Conclusion Traditional conceptions of democracy support the notion that governments should seek to improve the welfare of society. Of course, this does not mean that they always succeed –​only that their intentions should lean towards overall or particular benefits to societal groups. In theory, we should expect our governing bodies to look out for the interests of the people who elected them. Due to the vagaries of the partisan electoral systems embedded in many modern democracies, however, it is possible for governments

105

Policy Learning and Policy Failure

to select policy options that provide clear disadvantages to society and to taxpayers, without offering benefits to any discernible societal or electoral groups. In these situations, opportunities to learn lessons from domestic policy failures can be ignored. The governments in question can allow or will even actively assist a policy failure to worsen, with the intention of shaming an opposition party that can easily be blamed for the failure. In the cases described in this chapter, infrastructure megaprojects in two countries were allowed to experience a variety of policy failures  –​despite available options for correcting these failures, as would be consistent with lesson-​drawing from past ­examples –​in order to shame the opposition party that had initiated the project when it had been in government. In both cases, the incoming governments that implemented these strategies were rewarded in terms of voter and, more generally, citizen support. Popular policies, particularly in response to policy failures, need not be the best option in some sort of objective sense, but may have widespread appeal. More generally, policy makers and academic observers, likewise, must contend with the fact that empirically sound policy is often at odds with popular sentiment. It is, of course, difficult, if not impossible, to impute specific intentions to policy decisions, not least because policy decisions often affect multiple programmes, target populations and policy sectors, but also because policies may have multiple simultaneous goals. Furthermore, political goals may be kept hidden or may never be publicly declared (McConnell, 2010, 90). Nonetheless, it is our interpretation of the events of the two cases presented here that the incoming governments who inherited these projects intended to avoid blame for the projects’ failures to create situations that would result in the shaming of the opposition party. The BC Liberal party’s insistence on a third inquiry into the fast ferries project, the New South Wales Labor party’s unwillingness to negotiate with the Airport Link’s private sector partners over their bankruptcy, and the BC government’s eagerness to sell the fast ferries at a price that was vastly under market value are all examples of actions that would not have been taken by governments that were truly interested in correcting policy failure. Politicians in both cases stated publicly that they were not interested in initiating efforts to resolve problems with the projects if it required further input of resources from the government. Gordon Campbell, then premier of British Columbia, for instance, stated in an interview with the Vancouver Sun in 2002 that there was no chance the government would keep the fast ferries (McInnes, 2002), and Michael Egan, New South

106

Partisan barriers to learning from policy failure

Wales’ long-​serving treasurer, said that he believed that the Airport Link’s troubles would be best resolved by the project’s private sector partner and its creditors without government intervention (Egan, 2000, 10828). Furthermore, the most obvious alternative explanation, that these incoming governments were inept and that their actions were simply bad decisions of the same scale of incompetence as their predecessors, does not align with the evidence. For one thing, the Liberal government of British Columbia that took power in 2001 embarked on a well-​ documented agenda of massive development of infrastructure with a specific private investment approach (Cohn, 2008).Their careful, deliberative track record on megaprojects, especially in transportation (for example, Newman, 2014), does not support an interpretation in which decisions were made rashly or ineptly. In addition, statements from public officials in both British Columbia and New South Wales are highly suggestive of a partisan approach to dealing with both the fast ferries and the Sydney Airport Link. Liberal politicians in British Columbia frequently used language such as ‘$450 million boondoggle’ (Bell, 2003, 7207) and ‘a monument to NDP incompetence and chicanery’ (Krueger, 2005, 12189) to refer to the fast ferries project. In New South Wales, Labor used words like ‘bungled’ (Watkins, 2005, 18616), ‘failed’ (Keneally, 2006, 22423) and ‘disaster’ (Kerr and O’Malley, 2003) when speaking publicly about the Airport Link. It would seem, then, that there are certain conditions that may prevent governments from learning lessons from policy failures that they would then be able to apply to current public policy. A system that prioritises partisan competition above improved social outcomes is partly to blame: governments that have an opportunity to manipulate public support by demonising their opposition will do so regardless of the benefits or drawbacks to society. Expensive brick-​and-​mortar infrastructure projects are especially vulnerable to this kind of behaviour, because when they go wrong they do so disastrously, and because governments that engage in these megaprojects typically aim to create lasting legacies –​which only serves to cement their association with the project, years or even decades after the project’s conclusion or termination. Governments whose public support may be in doubt and are near the end of an election cycle ought to consider very carefully the risks involved in these kinds of expensive, long-​lasting, resource-​intensive endeavours, because although a successful project may provide some immediate political rewards, an unsuccessful one will become a mark of shame for many years.

107

Policy Learning and Policy Failure

References Armour, R, 2001, Rehabilitate the cats, Oslo: Kvaerner Masa Marine Baird, B, 1991, Light-​rail claims are premature, Sydney Morning Herald (Sydney, NSW), 9 May, 9 Baker, J, 2006, Westpac to bring airport ghost train back from dead, Sydney Morning Herald (Sydney, NSW), 27 July, 5 Beatty, J, 2001, Higher fares, less service urged for BC ferries: Report says some routes should be privatized, Vancouver Sun (Vancouver, BC), 18 December, A1 Bell, P, 2003, Debates of the Legislative Assembly of British Columbia 16, 10, Tuesday 7 October Bennett, CJ, Howlett, M, 1992, The lessons of learning: Reconciling theories of policy learning and policy change, Policy Sciences 25, 3,  275–​94 BC Ferries, 1993, Mid-​island transportation strategy: Final report, Victoria, BC: British Columbia Ferry Corporation BC Ferries, 2011, Response to British Columbia ferries commissioner discussion papers, December, 2011, Victoria, BC:  British Columbia Ferry Services Inc Bohn, G, 2003, Put fast cats on Langdale run, shipyard worker group urges, Vancouver Sun (Vancouver, BC), 22 March, B7 Borins, SF, with Brown, L, 1986, Investments in failure: Five government enterprises that cost the Canadian taxpayer billions, Toronto: Methuen Bovens, M, ‘t Hart, P, 1996, Understanding policy fiascoes, New Brunswick, NJ: Transaction Publishers British Columbia, 1999, A review of the fast ferry project:  Governance and risk management, Victoria, BC: Office of the Auditor General of British Columbia Carpenter, D, 2010, Reputation and power:  Organizational image and pharmaceutical regulation at the FDA, Princeton, NJ:  Princeton University Press Cohn, D, 2008, The new public autonomy? Public–​private partnerships in a multi-​level, multi-​accountable, political environment: The case of British Columbia, Canada, Policy and Society 27, 1, 29–​42 Dollery, B, Byrnes, J, Crase, L, 2008, Structural reform in Australian local government, Australian Journal of Political Science 43, 2, 333–​9 Dolowitz, D, Marsh, D, 1996, Who learns what from whom: A review of the policy transfer literature, Political Studies 44, 2, 343–​57 Downs, A, 1962, The public interest: Its meaning in a democracy, Social Research 29, 1, 1–​36 Dryzek, J, 1982, Policy analysis as a hermeneutic activity, Policy Sciences 14, 4, 309–​29

108

Partisan barriers to learning from policy failure

Dryzek, JS, Berejikian, J, 1993, Reconstructive democratic theory, American Political Science Review 87, 1, 48–​60 Dunleavy, P, 1995, Policy disasters: Explaining the UK’s record, Public Policy and Administration 10, 2, 52–​70 Dur, RAJ, 2001, Why do policy makers stick to inefficient decisions? Public Choice 107, 3–​4, 221–​34 Egan, MR, 2000, Airport rail link, New South Wales Legislative Council Hansard, Friday 24 November Flyvbjerg, B, Bruzelius, N, Rothengatter, W, 2003, Megaprojects and risk:  An anatomy of ambition, Cambridge, UK:  Cambridge University Press Gray, P, 1996, Disastrous explanations –​or explanations of disaster? A  reply to Patrick Dunleavy, Public Policy and Administration 11, 1,  74–​82 Grossman, PZ, 2009, US energy policy and the presumption of market failure, Cato Journal 29, 2, 295–​317 Hall, PA, 1993, Policy paradigms, social learning, and the state: The case of economic policymaking in Britain, Comparative Politics 25, 3,  275–​96 Hart, DK, 1984, The virtuous citizen, the honorable bureaucrat, and public administration, Public Administration Review 44 (Special Issue: Citizenship and Public Administration), 111–​20 Holme, R, 2003, Drinking water contamination in Walkerton, Ontario: Positive resolutions from a tragic event, Water Science and Technology 47, 3, 1–​6 Hood, C, 2007, What happens when transparency meets blame-​ avoidance? Public Management Review 9, 2, 191–​210 Howlett, M, 2012, The lessons of failure:  Learning and blame avoidance in public policymaking, International Political Science Review 33, 5, 539–​55 Jones, K, 2000, Breakthrough: The story of the underground rail link to Sydney Airport, Edgecliff, NSW: Focus Publishing Keneally, K, 2006, Sydney (Kingsford Smith) airport rail link fare, New South Wales Legislative Assembly Hansard, Tuesday 2 May Kerr, J, O’Malley, N, 2003, Costa puts up cheaper options for train link, Sydney Morning Herald (Sydney, NSW), 4 June, 3 Kinhill Engineers Pty Ltd, 1994, New Southern Railway environmental impact statement, Sydney: State Rail Authority of New South Wales Krueger, K, 2005, Debates of the Legislative Assembly of British Columbia 27, 21, Tuesday 1 March Larriera, A, 1991, Second harbour tunnel on govt short list, Sydney Morning Herald (Sydney, NSW), 29 April, 3

109

Policy Learning and Policy Failure

May, PJ, 1992, Policy learning and failure, Journal of Public Policy 12, 4,  331–​54 McConnell, A, 2010, Understanding policy success: Rethinking public policy, London: Palgrave Macmillan McInnes, C, 2002, ‘We knew it was going to be a tough road’: Mistakes, progress: Premier reflects, Vancouver Sun (Vancouver, BC), 16 May, A1 Mercer, K, 2009, Washington Marine Group sells PacifiCats to offshore firm, The Province (Vancouver, BC), 29 July, A3 Moran, M, 2001, Not steering but drowning: Policy catastrophes and the regulatory state, The Political Quarterly 72, 4, 414–​27 Morris, L, 1996, Rail link costs taxpayers extra $27m, Sydney Morning Herald (Sydney, NSW), 17 May, 7 Mossberger, K, Wolman, H, 2003, Policy transfer as a form of prospective policy evaluation:  Challenges and recommendations, Public Administration Review 63, 4, 428–​40 New South Wales, 1993, Integrated transport strategy for Greater Sydney: A first release for public discussion, October, Sydney, NSW: City Planning Department Newman, J, 2014, Measuring policy success: Case studies from Canada and Australia, Australian Journal of Public Administration 73, 2, 192–​205 Nixon, S, 2006, A few more board the airport train, Sydney Morning Herald (Sydney, NSW), 16 March, 3 Palmer, V, 2009, BC fast ferries voyage to oblivion leads to Middle East, Vancouver Sun (Vancouver, BC), 30 July, A3 Phillips, S, 2010, Party politics in British Columbia: The persistence of polarization, in M Howlett, D Pilon, T Summerville (eds) British Columbia politics and government, Toronto:  Emond Montgomery, pp  109–​29 Radaelli, C, 2009, Measuring policy learning:  Regulatory impact assessment in Europe, Journal of European Public Policy 16, 8, 1145–​64 RailCorp, 2005, Restated stations agreement (2005) New Southern Railway/​New Southern Railway settlement deed/​global amending deed/​ deed of release: Contracts summary, November, Sydney: RailCorp Richmond, R, Villemaire, T, 2006, Colossal Canadian failures 2:  A short history of things that seemed like a good idea at the time, Toronto: Dundurn Press Sabatier, PA, Weible, CM, 2007, The advocacy coalition framework: Innovations and clarifications, in PA Sabatier (ed) Theories of the policy process, Boulder, CO: Westview Press, pp 189–​220 Select Standing Committee on Public Accounts, 2000, Governance and risk management of the fast ferry project, Victoria: The Legislative Assembly of British Columbia

110

Partisan barriers to learning from policy failure

Steenberghen, T, López, E, 2008, Overcoming barriers to the implementation of alternative fuels for road transport in Europe, Journal of Cleaner Production 16, 5, 577–​90 Stone, D, 1999, Learning lessons and transferring policy across time, space and disciplines, Politics 19, 1, 51–​9 True, JL, Jones, BD, Baumgartner, FR, 2007, Punctuated equilibrium theory: Explaining stability and change in public policymaking, in PA Sabatier (ed) Theories of the Policy Process, Boulder, CO: Westview Press, pp 155–​87 Wainwright, R, 2000, Safety in numbers as struggling airport rail line cuts fares for group travellers, Sydney Morning Herald (Sydney, NSW), 16 August, 3 Wanna, J, 2006, Insisting on traditional ministerial responsibility and the constitutional independence of the public service: The Gomery inquiry and the Canadian sponsorship scandal, Australian Journal of Public Administration 65, 3, 15–​21 Watkins, J, 2005, Airport rail link, New South Wales Legislative Assembly Hansard, Thursday, 13 October Watson, E, 2004, Public-​sector corporate governance:  British Columbia’s best-​practices reforms, Ivey Business Journal, March/​April 2004,  7–​14 Weaver, RK, 1986, The politics of blame avoidance, Journal of Public Policy 6, 4, 371–​98 Wright, FR, 2001, Review of BC Ferry Corporation and alternative uses for the fast ferries, Report to the Board of BC Ferries, Victoria, BC: Ministry of Transportation of British Columbia Zittoun, P, 2014, The political process of policymaking: A pragmatic approach to public policy, New York: Palgrave McMillan

111

6

Policy failures, policy learning and institutional change: the case of Australian health insurance policy change Adrian Kay

Introduction Although policy learning may stem from assessments of policy failure (Ingold and Monaghan, 2016), establishing such a relationship is not straightforward. As is well known, the drivers of policy change and observations of policy change are not necessarily closely linked across time; indeed, the temporal link between cause and effect may be stretched over relatively long periods in patterns of change (McCashin, 2016). The chapter seeks to introduce a temporal perspective to catalogue the institutional constraints and opportunities embedded in sequences of repeated policy failure; and the subsequent ability of reform advocates to learn how to use the opportunities presented in future reform struggles. The first section of the chapter investigates the concept of policy failure in terms of several types, values and timings in order to assist understanding of how policy failures play out in longer sequences of policy making. The next section explores notions of policy learning in terms of the different types of failure identified. This leads into the main focus of the chapter: exploring connections between repeated assessments of policy failure, the catalysts of deinstitutionalisation and subsequent opportunities for system-​wide policy learning and reform. Selected evidence from the reform trajectory of Australian health insurance policy from the mid-​1970s to late-​1990s is used to explore these possible relationships. The current Australian health insurance system has its origins in a tumultuous and foundational sequence of policy change between 1972 and 1984, which settled down in the subsequent 15  years or so to leave Australia with a distinctive but important health insurance model (for example, Colombo and

113

Policy Learning and Policy Failure

Tapay, 2003; Hurley, 2002); an intriguing combination of universal public health insurance with financial and regulatory support for one of the highest percentages of private health insurance cover in the Organisation for Economic Co-​operation and Development (OECD). The Australian health insurance case allows us to catalogue at least one pattern of the relationships between policy failure, deinstitutionalisation and learning. Three core analytical arguments underpin this pattern. First, policy failures create opportunities for learning at a system-​wide level, following Elmore (1987), only after institutions have been eroded and exhausted by repeated failure. Second, this first claim holds in both the expert and political inquiry dimensions of policy failure. Third, learning processes are related to the particular sequence of deinstitutionalisation processes; in particular, initial deinstitutionalisation in the expert domain creates the conditions for political learning processes. Under pinning these three arguments is the concept of deinstitutionalisation. This chapter argues for this concept to be considered an important intervening variable between policy failure and policy learning. The concept labels a pattern of change where in institutions are gradually, over time, no longer observed, or where they fall into disuse, or wear out through imperfect reproduction over long sequences of time (DiMaggio, 1988). Conventionally in the new institutionalism in policy analysis, institutions are defined as sets of regularised practices with a rule-​like quality that structure the behaviour of actors in policy making and they endure and are not easily changed (for example, Mahoney and Thelen, 2010).The crux of new institutionalism for policy studies is the claim that institutions matter in the analysis of policy change, providing constraints on as well as opportunities for change, and they emerge and develop within a wide variety of historical processes and sequences. The second-​order questions of how, why and when institutions change, have often led institutionalisms, particularly the historical variety, to be judged as over-​emphasising positive feedback processes and the sensitivity of small events in initiating institutional development but under-​emphasising the subsequent opportunity for endogenous change in the process of institutional reproduction over time (van der Heijden, 2010). Echoing the problem of specifying change in policy studies, the first-​generation institutionalism in policy theory relied on perturbations occurring outside of the institutionalised policy system often characterised as societal or political upheaval; and the stability of institutions was explained by the absence of such exogenous shocks. More recently, however, new institutional theory approaches to change have become more refined in their attention to historical

114

Policy failures, policy learning and institutional change

causality and encourage an extension of the conceptual repertoire of policy studies to apprehend types of policy change beyond the relatively straightforward path dependency pattern periodically punctuated by surprises or crisis (McCashin, 2016). This chapter argues that deinstitutionalisation needs to be considered as a central part of this new conceptualism repertoire of policy analysis. The current literature reveals relatively little about the questions of why and how institutions fail to be observed, or cease to function, or gradually wear out. Although policy studies has described deinstitutionalisation in terms of displacement, where new institutions displace old, as well as forms of incremental institutional change through conversion, layering or drift (Béland, 2007; 2010), we still lack accounts of why and how institutions fail to reproduce themselves over time. In this chapter, we focus on deinstitutionalisation as a slow disassembling of institutions, rather than a process of forced displacement, which may be reinforced by agents interested in institutional extinction or reacted against at different points by agents of institutional preservation. Reaching back to a more sociological strand of new institutional theory, following Oliver (1992, 564), we can conceptualise deinstitutionalisation as ‘the process by which the legitimacy of an established or institutionalized organizational practice erodes or discontinues. Specifically, deinstitutionalization refers to the de-​ legitimation of an established organizational practice or procedure as a result of organizational challenges to or the failure of organizations to reproduce previously legitimated or taken-​for-​granted organizational actions’. Kay and Boxall (2015, 39) employ this lens for the study of public policy change by arguing that ‘failure repeated can gradually weaken institutions, exhaust them; failures are both a symptom but also a cause of deinstitutionalization’. In a similar policy studies vein, Jacobs and Weaver (2015) argue that policy feedback accounts have tended to emphasise reinforcement over reaction and that this is the source of institutionalisation and policy stability; but have failed to fully articulate and account for the feedback processes that might undermine institutionalised and previously stable policy.

The concept of policy failure: who judges, when and against whose standards? Any reasonable claim of policy failure requires the statement of a standard or benchmark against which to brace that assessment. This might be an absolute line or a standard relative to some moving target.

115

Policy Learning and Policy Failure

In the most recently developed literature on policy failure (Bovens et al, 2001; Bovens, 2010; Marsh and McConnell, 2010; McConnell, 2010a; 2010b) there is no scientifically objective standard –​absolute or relative –​for judgements of policy failure; instead such judgements are constructed socially through two types of inquiry: those undertaken by experts against those emerging from public politics. The two types of inquiry can lead to the emergence of different standards of policy failure. For the expert inquiry, the standard of failure is premised on substantive rather than procedural rationality and stresses evidence-​based policy making (EBPM). In the political inquiry, the stress is on the construction of, and impact on, the perceptions of policy performance by different publics. While the former is embedded in conventional research on policy evaluation in a range of social science disciplines in EBPM, the latter occurs at the public stage and the institutional arenas of politics and affects the translation of evidence into policy (Hulme and Hulme, 2012). The results of these two types of policy inquiry may be closely aligned: policy may fail to achieve substantive policy results, and also result in political punishment. Alternatively, governments may do well in policy expert terms and at the same time reap the political benefits of this. These are the sorts of outcomes one might hope for from various accountability and evaluation institutions in any advanced democracy. However, the outcomes of expert and political types of inquiry in policy performance may also conflict. Governments may succeed in policy terms on a given issue yet find that politically significant groups and forums judge themselves to be worse off. Likewise, governments may be able to ‘hide’ or ‘reframe’ any failures against expert standards so as to minimise the political costs which they might incur. The empirical diagnosis of an asymmetry between expert judgement and public judgement of policy failure presents an important predicament for (critical) policy studies approaches to the study of policy failure. The dilemma of the democratic expert has an ancient pedigree with roots in Plato, enjoyed great salience in the first half of the 20th century in the United States in celebrated debates between Lippmann (1925) and Dewey (1927) about democracy in a technological age, and has recently made a return in both public policy and democratic theory in the notion of epistemic communities (Haas, 1992) and in the promise of expert-​led politics and evidence-​based decision-​making, embodying policy processes that optimise efficiency and rationality (Head, 2016). For this chapter, however, the asymmetry between judgements is a key driver in apprehending the relationship between policy failure

116

Policy failures, policy learning and institutional change

and policy learning. In order to be useful analytically, the concept of policy failure must be more than simply what powerful interests that might benefit from judgements of failure think: a conventional analysis of interest and power can do that job. Instead, policy failure needs to refer to a meaningful and influential judgement that satisfies some generally accepted standards for credibility. Although the key insight of the policy failure literature to date is that judgements of failure are socially constructed, it is important for the concept to avoid this insight leading to a self-​defeating relativism where all claims of policy failure are treated as analytically equivalent with the only relevant variation for understanding policy processes being the power of those advancing the claim. Acknowledging that the process of expert inquiry will construct different standards of policy failure from those from the public arena does not inhibit the analytical value of the notion of policy failure.

Learning from policy failure Public political inquiry May (1992) has influentially distinguished between policy learning (instrumental and social) and political learning. The insight that policy learning is not always instrumental but sometimes social is a seam that runs through many models of the policy process, including institutionalist analyses (Kern et al, 2014; Niemela and Saarinen, 2012); and has roots back to the earliest work by Lasswell on the distinction between contemplative and manipulative policy analysis and its cognate, analysis of versus analysis for policy. In principle, policy failure should leave opportunities for policy learning (Bennett and Howlett, 1992); yet May finds that policy failure contexts tend to be conducive to political learning about the viability of different political strategies in the policy process; how better to advance ideas; or clues about coalition building. For the purposes of this chapter, the literature provides few firm propositions about policy failure and policy learning, but this does suggest that there is no simple one-​to-​one mapping between types of failure inquiry (expert and public political) and types of policy learning. Following May (1992), the concept can be used analytically to refer to the potentials and limitations in actors’ ability to reflect over the policy processes within which they are involved. While lacking complete information, political actors have some capacity to learn. They are reflective, routinely monitoring the consequences of their action, and their knowledge of the sources of failure will generally increase over time. Of course, since the broader and policy-​specific

117

Policy Learning and Policy Failure

context is both complex and evolving, any knowledge acquired is likely to be incomplete and asymmetrically distributed across actors in the policy-​making process. Through their actions and the consequences of these, policy makers learn, apprehending more clearly the informal institutions in policy and the constraints/​opportunities they impose, and providing the basis from which subsequent policy strategy might be formulated. What is important is that policy makers are in situations of strategic interdependence, regulated in varying degrees by institutions, and have to make estimates on the strategic motivations, intentions and likely actions of other significant players in a constantly evolving context as a result of strategic interaction. Thus the presumption of cumulative learning towards a ‘better’ public policy is difficult to sustain: there is no guarantee that actors draw the ‘right’ lessons from failure; or have the capacity to judge which lessons are right and which ones are wrong in a shifting and potentially deinstitutionalising context. The next section locates learning in the expert inquiry dimension of policy failure, where failure is conceived in terms of performance measurement, more accurate information and new estimates of key parameters in a policy model. Here in the public politics inquiry, learning mechanisms are investigated in the beliefs held by government officials and various other types of policy actors continuously and actively involved in policy formulation and implementation. When beliefs concerning failure are updated, changed by reflection on processes or outcomes, then we observe learning. This learning may act causally in accounts of subsequent policy development. Crucially, such policy learning by actors occurs within an institutionalised structure that may constrain certain ways of thinking and particular actions and facilitate others. In this basic form of institutionalism, institutional structures may facilitate and constrain which lessons are learned from policy failure and which are ignored. In the broad historical institutionalism variety of analysis, these institutions often reflect the outcome of past power struggles as well as inherited commitments and policy legacies. The public political dimension of failure may reveal institutional constraints and opportunities for reflective and skilful reform advocates who want to change policy in particular directions and for those who desire the status quo ante and are the most willing to accept limited adjustments to address the most immediate problems presented as failures. These institutions influence the power balance within the policy process by empowering some actors with particular interests and weaken others who hold differing views on the desired direction of policy evolution. Institutions facilitate and constrain which lessons are learned from policy experience and

118

Policy failures, policy learning and institutional change

which are ignored; in short, institutional change may be an intervening variable between policy failure and policy learning. The politics of failure may reveal constraints and opportunities to reflective and skilful reform advocates who want to change policy in particular directions and for those who desire the status quo and are the most willing to accept limited adjustments to address the most immediate policy problems. These perceived opportunities and constraints influence the power balance within the policy process by empowering some actors with particular interests and weaken others who hold differing views on the desired direction of policy evolution regardless of whether or not these views are motivated by power concerns or are outcomes of genuine learning processes. Health policy reform provides for a set of useful illustrative cases because there are many more reform attempts than successful reforms, and failure in healthcare reform often has high political costs. While governments are understandably keen to distance themselves from policy failures, both experts and politicians are not always so keen to move on. This may be desirable for instrumental or social learning reasons; in the literature on public policy as democratic experiment (for example, Campbell, 1999) failure is to be expected and succour drawn from failure as a prelude to establishing ‘what works’ through strategies of ‘trial and error’ and ‘learning by doing’. Expert inquiry From the basic connection between institutions and learning outlined earlier, we are alerted to the possibility of deinstitutionalising effects playing out alongside conventional learning processes in policy sequences where failures are observed. The current literature lacks analytical leverage on such processes of institutional unsettling. The Australian health insurance case presented here reveals an empirical pattern suggesting that deinstitutionalisation in expert inquiry can sometimes be a necessary precondition for system-​wide learning for ‘big’ reforms. In expert inquiry terms, while the large literature acknowledges that performance measurement is a form of policy learning (Braithwaite et al, 2007; Sabel and Zeitlin, 2012); less discussed is its related role as an advanced indicator of policy failure. For example, Pollitt (2004; 2006) identifies institutions such as the mandate of data collection and reporting agencies, the extent of formal reporting requirements, and the informal routines for data collection, reporting and feedback as relevant to the learning that occurs. These institutions can impose

119

Policy Learning and Policy Failure

limits and constraints on learning, to the point where experts might not see failure coming or indeed have the means of extracting useful policy lessons from observed failure. Policy scholarship has also documented how identifying and learning about policy performance demands a detailed mix of different types of indicators (those related to input, process, outcomes for example) developed to improve policy analysis and the relative strengths of different means to ends rather than exclusively for compliance purposes (Ingold and Monaghan, 2016). In broad terms, the effectiveness of policy learning depends on the level of trust among agents, where high trust relationships, often away from the public arenas of politics, can encourage candour about the nature of shared policy challenges and observed failures (De Bruijn, 2007). These environments require institutional designs to support dialogue among expert peers about policy performance and socialising shared norms and policy ideas about what works (Trubek and Trubek, 2005). For example, Lewis and Triantafillou (2012) argue that performance measurement counts as policy learning when it is characterised by bottom-​up rule-​making and displays careful attention to processes and sophisticated and comprehensive performance measures. Accounts of policy learning are strongly related to learning in the ‘new governance’ literature, which stresses institution building and sustaining in multi-​level contexts in its account of policy learning. It also draws explicitly on pragmatism to value diverse sources of (local) information and advances the claim that iterative deliberation and collaborative reflection on ‘what works’ will lead to (continuous) mutual policy learning (Sabel and Zeitlin, 2012). Although in essence a normative rather than an empirical claim, it does allow, for descriptive and explanatory purposes, recognition that knowledge about policy failure or success is often incomplete, and more specifically, individual policy solutions are unlikely to respond to all the facets of a given problem (Cohen, 2010). This presents the dilemma of commensurable evidence for expert inquiry learning within an institutional context: how to use information on the policy experience of local units across multiple contexts and then aggregate and convert this into an overarching assessment of policy performance. For new public governance frameworks, learning occurs through deliberation among expert peers and other stakeholders associated with the same broader policy domain. Such deliberation compares and articulates policy performance achieved in a given setting and attempts to rationalise how and why this has occurred. What is important is that the inclusion of a range of actors who exchange their policy knowledge

120

Policy failures, policy learning and institutional change

has the potential to destabilise the status quo or taken-​for-​granted ways of doing things. This is the process of deinstitutionalisation in expert inquiry into repeated policy failures. As we explore in the following Australian health case, this was a necessary condition for failure in public policy inquiry to then lead to eventual system-​wide reform. This is just one possible pattern of policy failure leading to reform through deinstitutionalisation. As an ideal processual model, policy learning is a recursive and transparent procedure, including periodic monitoring and deliberation about the progress made on different objectives and indicators (Hulme and Hulme, 2012; Ingold and Monaghan, 2016). Over time this presents participants with the opportunity to report back on what they have done to resolve earlier difficulties in achieving objectives. It also anticipates (continually) changing the overall framework of objectives themselves, as new solutions emerge through the process of problem solving (Sabel and Zeitlin, 2012). This institutional plasticity allows for expert learning from failure in order to prevent or limit institutional erosion. Policy learning from failure takes place without deinstitutionalisation.

Policy failure as a catalyst for institutional change Deinstitutionalisation is a plausible link between repeated policy failures, policy learning and large-​scale, system-​changing policy reform. This section develops the argument that policy failure can create the opportunity for subsequent reform by gradually preparing the broader political system for change through deinstitutionalisation. Elmore (1987, 175) outlines system-​changing reform as the transfer of authority among individuals and agencies in order to alter the system within which public policy is being made and delivered. Such transfer may come about by institutional design; or through emergence as repeated policy failures unsettle and exhaust existing institutions, existing policy instruments and their settings thereby shifting the distribution of authority. We can identify two institutional dynamics to aid analysis of the links between policy failure and policy learning. First, the process of creating new institutional arrangements; and the capacity of existing institutions to use their incumbency advantage to limit or undermine the development of new institutions. In other words, the extent of deinstitutionalisation matters for the probability of displacement of old institutions by new ones. Second, in the case where new institutions do emerge, deinstitutionalisation can affect how agents use new institutions. In particular, deinstitutionalisation affects the extent to

121

Policy Learning and Policy Failure

which coalitions under old institutions are residual or actually able to organise to convert new institutions to replicate the functions and power distributions of the old institutions. These two institutional dynamics combine in some form or other in any system-​changing policy reform. The overall effect depends on the extent to which existing institutional interests and incentives are able to survive and endure through driving new alternatives out of existence before they have had a chance to form and institutionalise fully. Policy learning offers the possibility for an account of agency in the policy process beyond the relatively simple information processing associated with instrumental learning, where beliefs are updated about the best means to achieve already established interests and ends. Instead, the concept of learning also includes the capacity for deeper reflection about the contingent and corrigible purposes of policy action. In the spirit of policy reforms as democratic experiments (Campbell, 1999), policy learning is the antecedent to changes in thinking as well as actual practices that may subsequently become institutionalised, whether in hard or soft forms. As discussed, policy failure inquiries may trigger deinstitutionalisation processes as well as learning. For example, in certain policy contexts as policy failures begin to occur they tend to exhaust institutions, leading them to be less widely observed or become imperfectly reproduced. At other times, policy failure may surprise and induce institutional change (for example, wholesale replacement or adaptation or conversion or layering) and subsequent policy change. The relationship between relatively informal and formal institutions is important to understanding how judgements of policy failure may trigger deinstitutionalisation and subsequent institutional change. The Australian health insurance case supports the claim that the informal tends to subvert the formal as policy failure undermines institutions. In particular, formal or law-​like institutions which are codified and enforced by third parties in public political inquiry into policy failure are by their nature ‘stickier’ than informal institutions that attend learning in expert inquiry. Policy failure as a negative feedback affects informal institutional practice in expert inquiry more immediately, absent any grand reform, as things are done slightly differently in response to failure inquiries. This may be unconscious but can steadily accumulate over time to the point where related formal policy institutions are undermined. Sometimes institutions can undermine their own foundations (Jacobs and Weaver, 2015). Even a self-​correcting capacity embedded in an institution in which policy failure leads to equilibrating adjustments

122

Policy failures, policy learning and institutional change

can bring down or exhaust an institution over time. For example, the density of rules to cope with policy failures qua exceptions increases, and such density and complexity can lead to rules over time become self-​disorganising. Some policies fail slowly; imperceptibly undermining their own political support by failing to meet changing contexts, and there can be expert discussions of new policy options or wholesale paradigm change. But there is always a double movement, resistance back. It is when the pushback in response to failure no longer provides resistance to change. Repeated failure can gradually weaken institutions, exhausting them; failures are both a symptom but also a cause of deinstitutionalisation. Where learning breaks the paradigm is the interesting point, when social learning takes over from instrumental learning that we see change.

Australian health insurance policy reform from early 1970s to late 1990s While judgements of policy failures in public political inquiry terms may be volatile and epiphenomenal, expert judgements are slower burning in the wider political system and often endure over longer periods. This is relevant to deinstitutionalisation and policy learning; as noted, in both the expert and public political inquiry domains, if failure is to be expected and succour drawn to establish ‘what works’ through strategies of ‘trial and error’ and ‘learning by doing’ then the different tempos of expert and political judgements of failure are an important point of analysis. The case of Australian health insurance policy reform between 1972 and 1984 stands analytically as one possible set of relationships between failure and learning. Australia remains the only OECD country to have introduced universal public health insurance, Medibank in 1974, abolish it and then subsequently reintroduce the same scheme in 1984 (Medicare).This empirical pattern provides a means to explore the three claims of the chapter because it displays, within a decade, two major reforms sitting at both ends of a period of gradual but transformative change and reveals the interactions between failure, deinstitutionalisation and learning. Australia’s first universal health insurance scheme, Medibank, was introduced by the Australian Labor Party (ALP).The ALP had won power at the federal level in 1972 after 23 years in opposition, a victory attributed largely to two factors: the popularity of its charismatic and domineering leader, Gough Whitlam, and Labor’s ambitious social reform programme, which promised to improve quality of life for all

123

Policy Learning and Policy Failure

Australians (Mayer, 1973).The overall centrepiece of Labor’s policy programme was Medibank. The Whitlam bills to introduce Medibank were at the top of the agenda of the unique joint sitting of the House of Representatives and Senate in August 1974, following a rare double dissolution federal election. This remains the only time in Australian history that members of both houses of the Commonwealth Parliament have sat together as a single legislative body under section 57 of the Constitution. This seminal ‘big bang’ moment initiated a sequence of policy change which was far from ‘locked-​in’, nor immediately transformative. The evidence from contemporary research is that key actors in the decade after 1974 regarded Medibank variously  –​in the different inquiry domains –​as a response to policy failure, as a failure itself but a constraint to be observed and possibly converted, but also at other times failure that itself required major reform (Boxall and Gillespie, 2013; Kay and Boxall, 2015). Medibank was a hybrid policy design; a public health insurance system layered onto an existing and strongly institutionalised private health insurance system. As Kay (2007) described, this design was introduced in response to widespread expert judgements of policy failure in Australian health insurance policy from the mid-​1960s onwards. In turn, the Medibank design introduced new tensions and failure tendencies into the health policy-​making system in Australia identified by widespread expert inquiry as (i) an opportunity for exit from the public element of the health insurance system possibly encouraging a reduction in loyalty to and undermining political support for that public element; (ii) policy tensions in the trade-​off between the goals of equity and efficiency in a healthcare system; (iii) segmentation in health financing, particularly where private health insurance has high political and public visibility, which may introduce significant collective action problems in policy implementation (see, for example, Cass and Whiteford, 1989; De Voe and Short, 2003; Gray, 2004; Hunter, 1984; Scotton, 2000; Scotton and Macdonald, 1993). In terms of tracing the policy path from Medibank to Medicare, Sax (1984) along with Boxall and Gillespie (2013) provide comprehensive accounts of the period. The Liberal–​N ational Party Coalition government (the Coalition) led by Malcolm Fraser came to power in Australia in December 1975 in highly controversial circumstances. After the dismissal of the Whitlam government on 11 November 1975, the Governor General appointed the Coalition as the caretaker government, and Fraser immediately called a general election which he won by a landslide.

124

Policy failures, policy learning and institutional change

After waging an energetic and highly public campaign against the introduction of Medibank during the Whitlam years, in the lead-​up to the 1975 election, Malcolm Fraser announced in that, if elected, the government would keep the newly established Medibank scheme –​ the same scheme it had voted against in Parliament at least five times during the previous three years. Fraser’s decision to promise the electorate that the Coalition would maintain Medibank was a firmly political rather than policy judgement. Little evidence of expert learning from failure can be discerned: Medibank was popular, and promising to keep it would help win over disillusioned Labor voters. In May 1976 during a television interview, Fraser gave the clearest explanation for why he changed his mind on Medibank: ‘Look, time marches on. Circumstances change and you deal with circumstances as they are. Medibank was introduced. Among many people it was plainly popular. It would have been destructive and unreasonable to attempt to break Medibank’ (Fraser, 1976). Underneath the political judgement of the PM of the day, assessment is required of the critical endogenous processes operating during the period, such as the construction of failure in institutionalised policy areas; and policy failure as a destabilising force in institutional reproduction; as well as learning in reform design. However, most work attempting to explain health policy development in Australia in this period relies on variants of a pluralist interest group approach to political analysis (for example, Duckett, 1984; Gardner, 1989; Gray, 2004; Hunter, 1984; Sax, 1984). Recent primary research on the period suggests that in both the expert or public political inquiries into failure, the presumption of cumulative learning towards a ‘better’ public policy is difficult to sustain:  there is no evidence of a guarantee that experts or public political actors would draw the ‘right’ lessons from failure; or show the capacity to judge which lessons are right and which ones are wrong (Boxall and Gillespie, 2013). Although the expert consensus on the failure of the pre-​1972 health insurance system has remained firmly established, there was no equivalent consensus on the lessons for managing a hybrid health insurance system in the period from 1984 until the late 1990s (Braithwaite, 1995). For example, between 1976 and 1981, the Fraser government introduced a series of confusing and contradictory reforms to health insurance that ultimately ended with the abolition of Medibank, and with it, temporarily, universal healthcare in Australia. Hawke defeated Fraser at the 1983 election and, drawing lessons from Whitlam’s failures, Labor developed a policy strategy to introduce Medicare that linked

125

Policy Learning and Policy Failure

economic reform to social policy. The introduction of Medicare was made contingent upon the roll-​out of large-​scale economic and industrial relations reforms (outlined in the Price and Incomes Accord). Although Medicare has endured since 1994, both the Whitlam and Hawke–​Keating governments failed to make substantial changes to the long-​standing private health insurance scheme that sat alongside Medibank/​Medicare. Whitlam was not in power long enough after the introduction of Medibank to have to deal with the problem of funding both universal health insurance and, to a lesser extent, private health insurance (through subsidies and rebates).The Hawke–​Keating government dealt with the budgetary problem by progressively withdrawing all subsidies for private insurance. However, near the end of its time in power, the Keating government was forced to restore support for private insurance in order to prevent the collapse of some private insurance funds (Boxall and Gillespie, 2013). Neither the Hawke–​Keating governments nor any government since has heeded the lessons of the Fraser years about the challenges of balancing a universal, tax-​funded insurance scheme alongside a substantial duplicative private insurance scheme (Banks et al, 1997). The lessons learned by John Howard, Coalition Prime Minister from 1996 to 2007, on health insurance serve as an interesting example of public political learning in the case. As Treasurer during the Fraser years (1977–​83), Howard and his department fought successfully to abolish the Medibank scheme. Howard strongly opposed the idea of compulsory, tax-​funded insurance and advocated instead for the restoration of a market for private health insurance, which had been in operation since the early 1950s. Ultimately, Howard’s view prevailed and Medibank was abolished. As opposition leader in 1987, Howard declared that his government would move quickly to dismantle the hugely popular Medicare scheme if elected (it lost). By 1995, and during his second term as opposition leader, Howard had changed his mind about Medicare. He explained, when Medicare was first introduced I was critical of it … and so were a lot of other Australians. But over the years people have grown to support it. It gives them a sense of security and it now has our total support. And there’s no law in politics that says that you can’t over a period of time change your view about an issue. (Elliot, 2006) Whitlam was the chief protagonist for Medibank in the Labor party, both in opposition and government. He succeeded in pushing his policy

126

Policy failures, policy learning and institutional change

preference upon the party (even though others supported alternative proposals) because his party knew that it was his electoral appeal that gave them the best chance of winning the 1972 election. Fraser, too, dominated his party in opposition and government. It was Fraser’s decision to reverse the Coalition’s long-​standing commitment to abolishing Medibank in the lead-​up to the 1975 election. Although the Fraser government did eventually abolish Medibank, there is substantial archival evidence showing that Fraser’s commitment to ‘maintaining Medibank’ was genuine and highly influential in policy deliberations during his years in government (Boxall and Gillespie, 2013). The Australian health insurance case of the sequence of reforms from the early 1970s to the late 1970s suggests that the sequence in which different failure, learning and deinstitutionalisation occur is critical in identifying how and when health insurance policy changes in the period. The reproduction of policy over time is not an automatic or self-​perpetuating process; and judgements of failure play an important and contingent role in deinstitutionalisation in different policy subsystems of Australian health policy. There is a degree of plasticity built into the design and implementation of mixed public–​private health insurance systems: while they are relatively durable, costly to overturn and may enter as causal factors in the behaviour of different groups, they generally remain politically contested settlements sustained by specific configurations of coalitions. The concepts of policy failure, learning and deinstitutionalisation put these tensions at the foreground of the analysis and seem to show that those who benefit from policy institutions, while they will have a preference for ensuring institutional continuity, must mobilise political support on an ongoing basis as well as actively seeking to resolve the tensions in processes of policy failure in their favour. The establishment, maintenance and erosion of institutional foundations are contingent and conflicted processes that can move forwards or backwards, by design or accident, but always producing unintended consequences. When these institutions malfunction, policy failures are likely to ensue which can accumulate over time to the point that institutions disassemble and without correction, opportunities for system-​wide policy learning present in both the expert and political domains.

Conclusion This chapter contributes to both the literature on policy learning as well as the most recent generation of new institutionalist work

127

Policy Learning and Policy Failure

describing patterns of endogenous and incremental institutional change. The various relationships between policy failure, policy learning and institutional change are difficult to model, varying in terms of political contexts and temporal sequencing of policy episodes. This chapter has sought to develop the distinction between different types of policy failure in order to connect these through the concept of deinstitutionalisation to different types of policy learning. In particular, the chapter contributes to the institutionalist literature on policy learning by uncovering how repeated policy failures affect the institutions that underpin learning in both the public political and expert inquiry domains by eroding and exhausting them. This deinstitutionalisation in turn affects the context of policy learning, and allows its scope to broaden beyond adjustments to system-​wide reform. In the case of Australian health insurance, the reformulation of a whole policy system emerged in response to widespread and accumulated acknowledgements of policy failure. While evidence of policy learning in this process is strong, this learning is related empirically to changing institutional contexts. The central dynamic in this interrelationship is provided by repeated policy failures leading to institutional erosion, exhaustion and immanent deinstitutionalisation that in turn allowed both public political learning and expert learning to broaden in scope from adjustments to stabilise the system to system-​ wide reforms. The chapter contributes to the catalogue of endogenous and incremental institutional change patterns in three ways. First, it has presented a pattern of change where policy failures create opportunities for policy learning at a system-​wide level through initiating institutional change. In particular, it is only after institutions have been eroded and depleted by repeated failure that system-​wide policy learning can take place. Second, this pattern of deinstitutionalisation may be observed in both the expert and political inquiry dimensions of policy failure. In turn, policy learning is related to the particular sequence of deinstitutionalisation processes. This third point highlights the position of deinstitutionalisation in the expert domain as critical to creating the conditions for subsequent political learning processes. Acknowledgements This research was supported by grants awarded to Kay from the Australian Research Council Discovery Program (DP120103676) and from the Australia and New Zealand School of Government. This funding is acknowledged gratefully. An earlier version of this chapter

128

Policy failures, policy learning and institutional change

was presented at a workshop, ‘Policy failures and governance design’, organised and supported by the Lee Kuan Yew School, National University of Singapore on 20–​21 February 2014. The input of workshop participants was constructive, interesting and valuable in revising the paper. References Banks, G, Owens, H, Kearnery, B, 1997, Report no 57: Private health insurance, Canberra: Industry Commission Béland, D, 2007, Ideas and institutional change in social security: Conversion, layering and policy drift, Social Science Quarterly 88, 1, 20–​38 Béland, D, 2010, Policy change and health care research, Journal of Health Politics, Policy and Law 35, 4, 615–​41 Bennett, CJ, Howlett, M, 1992, The lessons of learning: Reconciling theories of policy learning and policy change, Policy Sciences 25, 3,  275–​94 Bovens, M, 2010, A comment on Marsh and McConnell: Towards a framework for understanding policy success, Public Administration 88, 2, 584–​5 Bovens, M, ‘t Hart, P, Peters, BG (eds) 2001, Success and failure in public governance: A comparative analysis, Cheltenham: Edward Elgar Boxall, A, Gillespie, J, 2013, Making Medicare: The politics of universal health care in Australia, Sydney: UNSW Press Braithwaite, J, 1995, Health-​care reform under President Clinton: Issues, ideas and implications, Australian Journal of Public Administration 54, 1,  102–​12 Braithwaite, J, Makkai, T, Braithwaite, V, 2007, Regulating aged care: Ritualism and the new pyramid, Cheltenham: Edward Elgar Campbell, DT, 1999, Social experimentation, Thousand Oaks, CA: Sage Publications Cass, B, Whiteford, P, 1989, Social policies from Fraser to Hawke, in BW Head and AE Patience (eds) Australian public policy in the 1980s, Melbourne: Longman, Chapter 11 Cohen, A, 2010, Governance legalism: Hayek and Sabel on reason and rules, organization and law, Wisconsin Law Review 2, 357–​87 Colombo, F, Tapay, N, 2003, Private Health Insurance in Australia: A Case Study, Paris: OECD Health working paper Commonwealth of Australia, 2009, A Healthier Future for All Australians, Final Report of the National Health and Hospitals Reform Commission, Barton: Commonwealth of Australia

129

Policy Learning and Policy Failure

De Bruijn, H, 2007, Managing performance in the public sector, New York: Routledge De Voe, JE, Short, SD, 2003, A shift in the historical trajectory of medical dominance: The case of Medibank and the Australian doctors’ lobby, Social Science and Medicine 57, 2, 343–​53 Dewey, J, 1927, The public and its problems, University Park, PA: PSU Press DiMaggio, P, 1988, Interest and agency in institutional theory, in L Zucker (ed) Institutional patterns and culture, Cambridge, MA: Ballinger, pp 3–​22 Duckett, SJ, 1984, Structural interests and Australian health policy, Social Science and Medicine 18, 11, 959–​66 Elmore, RF, 1987, Instruments and strategy in public policy, Policy Studies Review 7, 1, 174–​86 Fraser, M, 1976, Transcript of Interview with George Negus, This Day Tonight, 28 May Gardner, H (ed) 1989, The politics of health: The Australian experience, Melbourne: Churchill Livingstone Gray, G, 2004, The politics of Medicare: Who gets what, when and how, Sydney: UNSW Press Haas, PM, 1992, Epistemic communities and international policy coordination, International Organization 46, 1, 1–​35 Head, B, 2016, Toward more ‘evidence-​informed’ policy making? Public Administration Review 76, 3, 472–​84 Hulme, R, Hulme, M, 2012, Policy learning? Crisis, evidence and reinvention in the making of public policy, Policy & Politics 40, 4,  473–​89 Hunter, T, 1984, Medical politics:  Decline in the hegemony of the Australian Medical Association? Social Science and Medicine 18, 11,  973–​80 Hurley, J, 2002, Parallel private health insurance in Australia: A cautionary tale and lessons for Canada, Canberra: Centre for Economic Policy Research Jacobs, A, Weaver, K, 2015, When policies undo themselves:  Self-​ undermining feedback as a source of policy change, Governance 28, 4,  441–​57 Ingold, J, Monaghan, M, 2016, Evidence translation: An exploration of policy makers’ use of evidence, Policy & Politics 44, 2, 171–​90 Kay, A, 2007, Tense layering and synthetic policy paradigms:  The politics of health insurance in Australia, Australian Journal of Political Science 42, 4, 579–​91

130

Policy failures, policy learning and institutional change

Kay, A, Boxall, A, 2015, Success and failure in public policy: Twin imposters or avenues for reform? Selected evidence from 40 Years of health-​care reform in Australia, Australian Journal of Public Administration 74, 1, 33–​41 Kern, F, Kuzemko, C, Mitchell, C, 2014, Measuring and explaining policy paradigm change:  The case of UK energy policy, Policy & Politics 42, 4, 513–​30 Lewis, J, Triantafillou, P, 2012, From performance measurement to learning: A new source of government overload?, International Review of Administrative Sciences 78, 4, 597–​614 Lippmann, W, 1925, The phantom public, Piscataway, NJ: Transaction Publishers Mahoney, J, Thelen, K (eds) 2010, Explaining institutional change:  Ambiguity, agency, and power, Cambridge:  Cambridge University Press Marsh, D, McConnell, A, 2010, Towards a framework for establishing policy success, Public Administration 88, 2, 564–​83 May, PJ, 1992, Policy learning and failure, Journal of Public Policy 12, 4,  331–​54 Mayer, H (ed) 1973, Labor to power:  Australia’s 1972 election, Sydney: Angus and Robertson McCashin, A, 2016, How much change? Pierson and the welfare state revisited, Policy & Politics, 44, 2, 313–​29 McConnell, A, 2010a, Understanding policy success:  Rethinking public policy, Houndmills: Palgrave Macmillan McConnell, A, 2010b, Policy success, policy failure and grey areas in-​between, Journal of Public Policy 30, 3, 345–​62 Niemela, M, Saarinen, A, 2012, The role of ideas and institutional change in Finnish public sector reform, Policy &Leon Politics 40, 2,  171–​91 Oliver, C, 1992, The Antecedents of Deinstitutionalization, Organization Studies 13, 4, 563–​88 Pollitt, C, 2004, Performance information for democracy: the missing link? Paper presented at the European Evaluation Society conference in Berlin, September 30–​October 2 Pollitt, C, 2006, Performance management in practice: A comparative study of executive agencies, Journal of Public Administration, Research and Theory 16, 1, 25–​44 Sabel, C, Zeitlin, J, 2012, Experimentalist governance in the European Union: Towards a new architecture, Oxford: Oxford University Press Sax, S, 1984, A strife of interests: Politics and policies in Australian health services, Sydney: George Allen and Unwin

131

Policy Learning and Policy Failure

Scotton, RB, 2000, Milestones on the road to Medibank and Medicare, Medical Journal of Australia, 3 July, 173, 1, 5–​7 Scotton, RB, Macdonald, CR, 1993, The making of Medibank: Australian studies in health service administration no 76, Sydney: University of New South Wales, School of Health Service Management Trubek, DM, Trubek, LG, 2005, Hard and soft law in the construction of social Europe:  The role of the open method of coordination, European Law Journal 11, 3, 343–​64 van der Heijden, J, 2010, A  short history of studying incremental institutional change:  Does explaining institutional change provide any new explanations? Regulation and Governance 4, 2, 230–​43

132

7

Policy myopia as a source of policy failure: adaptation and policy learning under deep uncertainty Sreeja Nair and Michael Howlett

Introduction Policy failure can broadly be characterised as the failure of policy efforts to attain policy objectives, the occurrence of outcomes that have adverse impacts on target populations, or ones which have negative political outcomes for policy makers (McConnell, 2010; Newman and Head). This chapter focuses on one reason why such failures are a common phenomenon in policy making: the problems and limitations that policy makers encounter dealing with imperfect knowledge and their resulting inability to often accurately predict the future. Of course, the fact that policy makers and policy making is ‘boundedly rational’ in the sense that some limits always apply to knowledge of future events and consequences is well known (Lindblom and Cohen, 1979; Simon, 1991). As Simon (1955) noted, many of the limits on rational policy making have to do with information limits which prevent consideration of possibly superior alternative courses of action, limits which in many cases cannot be overcome through better data collection and analysis as they are imposed simply by the fact that they relate to future events and activities which are difficult to model or predict. The exact nature and impact of these boundaries on knowledge and cognition are, however, less well known. Some behaviour, such as the willingness of automobile drivers to stop at red lights, is in normal conditions quite predictable as is also, for example, the willingness of a certain percentage of the population to take up tax incentives aimed at having them voluntarily contribute more funds to their pension plans, or for bankers to respond to interest rate hikes by withdrawing funds from circulation and thus help dampen inflation. Policies which fail in such circumstances may do so for other reasons than knowledge

133

Policy Learning and Policy Failure

limits; for example, due to poor implementation practices, malfeasance or many other similar less cognitive causes (May, 2014). The fundamental problem of an uncertain future, however, as Lindblom (1959; 1979) noted, is of a different order. This source of failure can be termed ‘policy myopia’ or the difficulty of seeing far enough into the future to discern its general shape and contour in enough detail to be able to properly anticipate and plan in the present. Like the optical condition it analogises, ‘policy myopia’ or the inability to clearly see the horizon of the future policy environment in which impacts of the policy will develop, requires corrective lenses to help clarify and offset the uncertainties with which policy makers are dealing (Ison et  al, 2015; Maxim and van der Sluijs, 2011). These lenses include building ‘robustness’, ‘resilience’ and ‘agility’ into policies or the ability for them to change and adapt over time to build capacity for better learning on the part of policy makers and decision-​takers about the lessons of the past concerning the nature of future uncertainty and risk and other techniques such as the promotion of ‘evidence-​based or informed’ policy making; or better data collection and analysis of past, present and future behaviour in order to help steer policies in desired directions over the medium and long term. In order to better understand this condition, this chapter presents a review of efforts to characterise policy uncertainty and the learning processes and adaptive techniques that can help deal with the various cognitive and normative challenges stemming from ‘policy myopia’. It sets out a taxonomy of different types and levels of uncertainty and draws out the implications of each for policy making, along with possible corrective actions which can be taken in each case.

Modelling policy myopia Policies are continually being designed for current and future conditions, about which policy makers have incomplete or little information. While some government policies are crafted in response to future events that are ‘reasonably predictable’ –​such as cycles of commodity price swings or periods of inflation and unemployment, or longer-​term and anticipatable demographic changes such as aging of populations or increasing urbanisation for which reasonable time-​ series data exist –​others are affected by policy events and futures which make them less predictable and potentially subject to ‘unforeseen’ and ‘unprojectable’, and sometimes catastrophic, failure (Wardekker et al, 2010; Watson et al, 2015).

134

Policy myopia as a source of policy failure

Long-​term policy making in particular is a vexing problem for policy makers since policies developed to deal with a specific problem or issue, if left in place unchanged long enough, are likely not to achieve their intended goals at some point in the future (Swanson and Bhadwal, 2009). Such long-​term policies are often rendered ineffective or irrelevant because of change in the original environment or the context of a policy intervention, owing to unplanned or unexpected shifts in actor configurations, ideas or the political conditions and coalitions which supported and created them (Jacobs, 2008). Policy making is thus by nature an exercise which often involves projections of current trends which do not foresee, or discount, the likelihood of severe challenges to the status quo operation of a policy (Howlett, 2014; Manski, 2011; 2013).This affects both good-​faith efforts to deal with problems which are inherently difficult to predict and also problem responses based on denial of this uncertainty, hubris, delay or simply attempts to ‘sweep the problem under the rug’ and wait for it to emerge at a future point in time (Howlett and Newman, 2013; Newman and Howlett, 2014). Failing to identify the bounds and range of uncertainties, is thus a major cause of policy failures leading to over-​and under-​reactions in policy responses (Maor, 2012; 2014) and policy over-​and under-​design. This issue of identifying different types of uncertainty, their diagnosis and their implications for policy success and failure are elements of policy making which have received some attention in the past but which have not been resolved (Jarvis, 2011; Morgan and Henrion, 1990; Stirling, 2010). Avoiding policy failure by properly diagnosing the level and type of uncertainty which a government faces in dealing with a problem is complex because not all policy environments change as rapidly as others and not all information problems are as acute as others. Under conditions of high or ‘deep uncertainty’ (Brugnach et al, 2008; Walker et al, 2010) in particular there is little agreement on the choice of models to characterise a system’s variables and their interactions or to even assign a likely probability distribution (McInerney et al, 2011) or valuation of diverse possible future outcomes (Walker et al, 2010). Furthermore, in some cases the interpretation of uncertainty signals may be convoluted because of interference or ‘noise’ from associated moral, political and social issues, such as those which occurred in the efforts to deal with AIDS (Day and Klein, 1989), and global environmental issues, such as climate change with skewed distributional impacts (Parry et al, 2007). A major problem, however, in devising strategies to deal effectively with this fundamental kind of policy uncertainty has been the

135

Policy Learning and Policy Failure

inadequacy of various schemes and models used to classify different levels and types of uncertainty and assess their effects. Historically students of policy problems such as Churchman (1967), Rittel and Webber (1973) and Simon (1973) considered uncertainty in a purely ‘objective’ sense, that is, focusing on the extent to which problem causes and solutions were known or unknown. Bivariate concepts of ‘wicked’ and ‘tame’ or ‘well-​structured’ and ‘ill-​structured’ problem contexts introduced by these authors have dominated thinking in this area. However, at minimum, a more complex range of possible policy-​making scenarios or contexts exists which includes both the level of ‘objective’ knowledge of problems as well as the relative nature of decision-​makers’ knowledge of that ‘fact-​base’. Moreover, uncertainties can arise not only from purely epistemic concerns such as incomplete information on the level of employment involved in unemployment insurance payouts, but can also relate to inherent ontological problems in policy making such as the variability and unpredictability of a system (physical system, human behaviour, technological advancement) or its potential to ‘surprise’ (Manzo et al, 2015; Perrow, 1984; Walker et al, 2003). Contextual factors within the policy-​making system itself are also important. Maxim and van der Sluijs (2011), for example, have noted a continuing tendency for most policy typologies to remain focused on the ‘producer’ of information and ignore uncertainty related to process and communication between producer and the end-​user, that is, the decision-​maker. These uncertainties can relate to ‘qualification of knowledge base’ or the degree of agreement upon or the absolute size of the evidentiary support for models, or the ‘value-​ladenness’ of policy choices, which includes different actor perspectives on the worth and value of the knowledge and information being utilised for decision-​making and the presentation of arguments concerning preferred policy alternatives and pathways (Mathijssen et al, 2008). In an early article Hansson (1996) termed these different kinds of uncertainties as: (1) uncertainties of demarcation, where it was not well determined what the options for action are; (2) uncertainties of consequences, where it was not well known what the consequences of options are; (3) uncertainties of reliance, where it is not clear what information is reliable; and (4) uncertainties of values, where the values of decision-​makers and others were not well determined. He noted that much policy analysis focused on the second set of uncertainties and how to address them, while often ignoring the other three (Hansson, 1996). Hirsch-​Hadorn et al (2015), following Hansson (1996), set this out as dealing with three separate locations of uncertainties in 1) policy

136

Policy myopia as a source of policy failure Figure 7.1: Typology of uncertainties by options, outcomes and values Location of uncertainty

Source of uncertainty Incomplete information

Inherent indeterminacy

Unreliable information

Options

Unfinished list of unclear options

Unfinishable list of options

Contested framing of decision problem

Outcomes

Subdivided into statistical uncertainty, scenario uncertainty, ignorance

Subdivided into statistical uncertainty, scenario uncertainty, ignorance

Questionable information base

Values

Pragmatic incompleteness of rankings

Pragmatic incompleteness of rankings

Completed rankings despite fuzzy or ambiguous values

Source: Hirsch-​Hadorn et al, 2015

options, (2) their outcomes and 3) the values associated with these outcomes, owing to incomplete information and/​or indeterminate and unreliable information (Figure 7.1). This is a useful classification but still assumes that decision-​makers are able to visualise or understand what they do not know. However, even at a relatively simple level when knowledge is available on a subject, policy makers may be unaware of it and thus undertake decision-​ making on the basis of ignorance rather than knowledge. This situation becomes more complex as collective or absolute knowledge of a subject or phenomenon is lacking so that ignorance is less culpable although no less real and consequential (Figure 7.2). Decision-​makers may be aware of this gap and function with an attitude of prudent awareness or, when they are unaware of their ignorance, with a hubristic attitude or over-​confidence. This is often alleged to be the case with issues such as crime, for example, which often engenders poor policies based on over-​and under-​reactions to real or perceived problems (Becker and Brownson, 1964; Hilaire et al, 2015; Perl et al, 2018). This situation adds the dimensions of (sometimes wilful) ignorance to the ‘ambiguities’ contained in Hansson’s model (Proctor and Schiebinger, 2008). Ambiguity in this sense is a persistent issue in policy making and refers to the ‘simultaneous presence of multiple frames of reference about a system among different actors’ (Kwakkel et al, 2010) while ignorance relates to either or both a lack of knowledge about what is known and knowable or what is unknown (Schrader et al, 1993).

137

Policy Learning and Policy Failure Figure 7.2: Policy maker’s knowledge and comprehension matrix Nature of existing collective knowledge of a phenomenon

Aware Nature of decision-makers awarenss of existing knowledge of a phenomenon Ignorant

Aspects of a problem and possible solutions are known

Aspects of a problem and possible solutions are unknown

Known-known: Key policy actors are aware of the known aspects of a phenomena (INFORMED AWARENESS)

Known-unknown: Key policy actors are aware that certain aspects of the phenomena are unknown (PRUDENT AWARENESS)

Unknown-known: Key policy actors are aware of aspects of a phenomena (UNINFORMED IGNORANCE)

Unknown-unknown: Key policy actors are aware that certain aspects of the phenomena are unknown (IMPRUDENT ignorance)

Source: authors, based on Becker and Brownsen, 1964; Chow and Sarin, 2002; Knight, 1921

Examining these different scenarios from the standpoint of risk assessment, Stirling (2010) argued it is important to distinguish not only between ‘uncertainty’, ‘ambiguity’ and ‘ignorance’ as set out earlier but also to show how these conditions differ from traditional notions of ‘risk’, such as those first set out by Knight in his 1920s work on the subject (Jarvis, 2011; Knight, 1921). Stirling has argued each type is characteristic of several very different kinds of policy-​making environments which extend well beyond traditional actuarial and other definitions of risk (Bond et al, 2015; Jarvis, 2011; Leung et al, 2015; Schrader et al, 1993) and directly affect the kinds of remedial actions which decision-​makers take in the attempt to avoid or overcome them (see Figure 7.3). Kwakkel et al (2010) and Walker et al (2003; 2013b) have continued to work in this same direction, developing more precise schemas for describing future uncertainties and detailing potential correctives them. Their model of policy uncertainty (see Figure  7.4) includes ‘Level I’, ‘shallow’ or ‘parameter’ uncertainties where alternative states of a system are well known and established, a category which is roughly equivalent to Stirling’s idea of ‘risk’. These are distinguished from ‘Level II’ ‘uncertainty’, as Stirling would have it, where multiple alternatives exist within the same scenario given the ‘fuzzy’ or imprecise nature of problem and solution parameters and the difficulties with precise prediction it implies. As Kwakkel and others have noted, both of these first two types are relatively straightforward and differ from a second order or range of uncertainty which involves multiple, competitive or clashing scenarios. Level III situations are the least complex of these second major taxa where different scenarios exist but can still be ranked in terms of their

138

Policy myopia as a source of policy failure

Likelihood/probabilities

Known

Figure 7.3: Different kinds of risk faced by policy makers and potential solutions

Unknown

Risk Risk assessment Optimising models Expert consensus Cost-benefit analysis Aggregated beliefs

Interactive modelling Participatory deliberation Focus on dissensus groups Multicriteria mapping Q-method, reporting grid

Interval analysis Scenario methods Sensitivity testing Decision rules Evaluative judgement

Monitoring and surveillance Reversibility of effects Flexibility of commitments Adaptability, resilience Robustness, diversity

Ambiguity

Uncertainty Known

Ignorance Unknown

Outcomes

Source: Stirling, 2010

Figure 7.4: Characteristics of different types of uncertainty Level 1 Context

A clear enough future

Level 4 Level 3 Deep uncertainty Alternate futures A multiplicity of Unknown future (with plausible futures probabilities) A Level 2

System model

A single-system A single-system model model with a probabilisitic parameterization System A point estimate Several sets of outcomes and confidence point estimates interval for each and confidence outcome intervals for the outcomes, with a probability attached to each set Weights A single Several sets of on estimate of weights, with outcomes the weights a probability attached to each set

Source: Walker et al, 2010, Figure 1: 919

139

Several system models, with different structures A known range of outcomes

Unknown system model; know we don’t know

A known range of weights

Unknown weights; know we don’t know

Unknown outcomes; know we don’t know

Total ignorance

Determinism

B C

Policy Learning and Policy Failure

likelihood. This is similar to what Stirling described as ‘ambiguity’. It still allows a form of boundedly rational decision-​making where alternative courses of action can be ranked in terms of their likely impact in the future. Level IV uncertainty in Kwakkel and Walker et al’s schema, however, represents a more complex situation in which multiple plausible alternative future scenarios can be enumerated. In Level III there may be some agreement on at least the number of possible general models of the future but it is not possible to rank alternatives in terms of their likelihood (Stirling’s ‘ignorance’ category). In the most complex ‘Level IV’ situation there is an inability to even present or agree upon the range of possible alternative scenarios, and the ‘possibility of being surprised’ in such a scenario must be acknowledged as being very high (Walker et al, 2013b). These final two situations of multiple, contested scenarios fall into a category Walker et al refer to as ‘deep’ uncertainty. These are the most myopic situations because they cannot be corrected simply through better data collection and analysis which might allow probabilities of future event occurrences to be approximately estimated. With multiple perspectives regarding the nature of an issue as well as multiple potential solutions whose prospects for success are unknown these situations approximate the ‘wicked’ problems first articulated by Rittel and Weber (1973) but with even more dimensions of ‘wickedness’ (Lewin et al, 2012). Policies dealing with climate change, for example, fall into this class as any delay in action towards addressing climate change now only makes a very uncertain future scenario more difficult to deal with over time (Levin et al, 2012). Such problems face an ‘uncertainty explosion’ as uncertainty gathers and often magnifies over time (Schneider and Kuntz-​Duriseti, 2002) as any errors introduced in fuzzy parameter estimates accumulate and multiply over time as events unfold (Parry et al, 2007). These kinds of issues have been termed ‘super-​wicked’ problems by Levin et al (2012) but should more properly be termed ‘inherently myopic’ ones.

Dealing with different types of uncertainty Distinguishing between these uncertainty types and understanding which uncertainty circumstances a decision-​maker finds oneself in is crucial in dealing with or attempting to overcome policy myopia. It bears repeating that many policies exist in circumstances in which the uncertainty is minimal or manageable, such as what happens in transportation policy making or health policy making where historical

140

Policy myopia as a source of policy failure

data and linear relationships greatly reduce risks associated with uncertain futures and uncertainties operate at manageable levels. Even in these circumstances, however, it is very common for decision-​makers and analysts to underestimate the nature of the knowledge and limits with which they are dealing (Committee on Decision Making under Uncertainty et  al, 2013; Lempert et  al, 2004). Hubris, or the inability or unwillingness to recognise limits to knowledge and certainty is all too commonly a key contributing factor to conscious and unconscious short-​term thinking (Desai, 2015; Lindblom and Cohen, 1979; Zeisel, 1981) and such actions must be guarded against (Head, 2010; 2016). Inherently myopic policy efforts under deep uncertainty, however, require additional actions and techniques beyond improved forecasting and future scenario analysis (van der Steen et al, 2016; Vink et al, 2016).1 A critical task for policy makers involves learning to design better and enable policy responses in accordance with the different levels and types of uncertainties identified earlier. While in Level I and II circumstances this involves policy makers in efforts to either reduce uncertainty through, for example, data analysis and analytical capacity building (Wu et al, 2015), in cases of deeper uncertainty, however, additional techniques are required (Bredenhoff-​Bijlsma, 2010; O’Neill et al, 2006). That is, policy problems characterised by Level I  or parameter uncertainty are at least in theory not very difficult problems to handle and likely to be resolvable by standard treatments. Hence, for example, controlling housing markets though interest (mortgage) rate manipulations is a well-​known stratagem when based on reliable data. Improving data collection and information management while ensuring policy-​making incorporates this knowledge is thus a standard and reasonable technique for dealing with this kind of issue. Level I  uncertainty offers only a very limited range or space for failure, for example when low probability catastrophic events such as hurricanes or earthquakes or financial collapse occur and upset otherwise stable policy environments. Various techniques can be used to cover even this kind of uncertainty, however, such as using conventional forecasting methods like Monte Carlo simulations or other kinds of statistical analyses (Brugnach et al, 2008; Manski, 2013; Walker et al, 2010). Furthermore, some flexibility and additional capacity can be designed into a policy to deal with what are reasonably expected to be ‘60-​year’ or ‘100-​year’ one-​off events. Level II uncertainty is only slightly more complex and may involve some unexpected results from policy interventions  –​such as when tobacco price hikes designed to discourage cigarette consumption run

141

Policy Learning and Policy Failure

into problems with smuggling, black markets and tax evasion. Again this behaviour can often be anticipated (Boulding, 1947) and factored into predictions of future events and trajectories by, for example, increasing police patrols or using a mix of tools such as anti-​smoking advertising and health educational campaigns and indoor smoking bans rather than simply relying on market-​or price-​based signals (Lambin et al, 2014; Morgan and Henrion, 1990). At higher levels of uncertainty, however, such strategies are not able to deal with the unknowns and ambiguities which exist at these levels (Roggema et  al, 2012).Taking the case of climate change policies, for example, when the degree of external change associated with global warming becomes high, a large change in adaptation response or ‘transformative adaptation’ may be required if, for instance, major ocean currents shift or change direction (Kates et al, 2012; Vermeulen et al, 2013). Under such conditions, policy responses based on the problems ‘as we know them now’ while ‘factoring in a margin for them becoming worse’ are not sufficient (Heazle et al, 2013). Policies that are designed based on ‘same scenario tomorrow’ forecasts and impact assessments using strict ‘baselines’ typically cannot deal with multiple possible futures (Bond et al, 2015). Under such higher levels of uncertainty, coping strategies are more complex and difficult. When there is such a high level of uncertainty about the nature of a policy problem and the potential effects of policy decisions, policy makers often ‘grope along’ and prefer innovation along the way, after little if any initial planning and analysis. In such cases policy ‘corrections’ are common (Deyle, 1994). However, such problems involve very different possible alternative scenarios –​such as when transport planners try to increase the number of walkers and cyclists at the expense of car drivers and need to evaluate different levels of take-​up and road design involved in each possible future walker-​r ider-​driver state. Such problems involve a number of possible variations in each scenario and the difficulty of accurately forecasting road usage climbs dramatically and changes over time given the trade-​ offs and linkages between alternate scenarios (Taeihagh et al, 2013). Groping along is inadequate in such circumstances as it bears with it a high probability of failure. Policies designed to deal with such multiple future scenario issues thus must not only be more flexible and adaptable than those dealing with lower levels of uncertainty but also be of a different type. Under these higher levels of uncertainties policy solutions must be designed to be robust over different potential scenarios and time periods (Walker et al, 2013a), and designed to be adaptive or resilient, that is, accepting

142

Policy myopia as a source of policy failure

the likelihood of an adverse future but focusing on quick recovery (Chandler, 2014). A core characteristic of such policies is their capacity to learn and adapt with change and their ability to ‘self-​organise’ under dynamic conditions (Anderies et  al, 2012; Carpenter and Brock, 2008). Acknowledging the limitations of ‘rational calculation, planning and forecasting’ (March and Olsen, 1975) policy making in such situations draws attention to the role of learning through experience, experimentation and feedback to create adaptive policies which are flexible and agile over time (Hall et al, 2012; Pahl-​Wostl,  2007). By deploying such approaches to deal with dynamic futures, policy makers can learn to accept the ‘irreducible character’ of future uncertainties (Walker et al, 2013a) and the almost inevitable failure of static policies in such dynamic environments and allow plans to change over time as the conditions change and new knowledge and relationships emerge (Bond et al, 2015; Polasky et al, 2011). Resilience, however, needs to be understood as a ‘process’ characteristic of systems (Park et  al, 2013) as compared to robust policies which are ‘fail-​safe’ within a range of uncertainty (Davoudi et  al, 2012). Under conditions of deep uncertainty, policies must be in a constant state of adjustment as new knowledge emerges of system parameters, problem structure and scenario likelihood. In such adaptive policy making, decision-​makers need to operate as ‘continuous policy-​fixers’ (Ingraham, 1987) and their role oscillates between that of a policy ‘architect’, ‘facilitator’ and ‘learner’ in the policy process to appropriately adjust and re-​constitute policies in response to changing conditions (Swanson and Bhadwal, 2009).2 Such adaptive policy making is based on learning over time, operating on available ‘best’ scientific information until new knowledge comes up, consciously experimenting with policy alternatives to identify better strategies as new conditions emerge and providing opportunities for policy review and renewal (Coglianese, 2011; Swanson et al, 2010; Walter, 1992). Brown (2000) conceptualises the kinds of policy learning needed as involving ‘new information including policy feedback and new causal understandings that result in more effective policies, defined as more complex, integrated, flexible, and implementable legislation that enhances goal achievement’. Such learning begins with the recognition of the need for policy content to be continually updated in order to deal with changes in context (Jones and Baumgartner, 2012; Lejano and Shankar, 2012; Swanson and Bhadwal, 2009; Walker et al, 2003).

143

Policy Learning and Policy Failure

Learning to deal with policy myopia:  promise and problems Adaptive policy making in situations of high uncertainty emphasises the need not only for the policy itself to be adaptive or ‘robust’ but for the process of decision-​making or ‘pathway’ to policy making to also share this characteristic (Haasnoot et al, 2013; Wise et al, 2014). For long-​term policies that address complex and dynamic policy problems there is a need for constant monitoring and evaluation to ascertain if policies are still continuing to meet their intended goals and objectives (Ramjerdi and Fearnley, 2014). Swanson and Bhadwal (2009), for example, emphasise the importance of formal policy review and continuous learning through a regular review process even when the policy is functioning well as one of several process tools that can help policies deal effectively with anticipated and unanticipated conditions. Learning is, however, not a panacea, and learning from experience can be subjective based on different interpretations of intended goals, especially when these are vague, and evaluation of outcomes and their causal mechanisms. One challenge for policy making in situations of high uncertainties, for example, is often that there may be little or no scope for the decision-​maker to respond from history or experience (Lempert et al, 2003; Mumford, 2015; Walker et al, 2010), meaning that traditional reliance on single or even double-​loop kinds of learning may be ineffective. This may be due to turnover in office or lack of institutional memory or from the novelty of new challenges, as is often the case in foreign policy decision-​making, for example, such as when frequent large-​scale terror attacks on civilians emerged in the late 1990s and early 2000s with few precedents in terms of their scope and origins. It is also often difficult to introduce any radical changes in a policy mix even if new policy objectives are put forth given lock-​in and the distributional-​electoral aspects and impacts of such changes (Kern and Howlett, 2009). A  key challenge while designing far-​sighted policies for the future is thus to operate effectively and plan in a space where pre-​existing policy mixes have developed over time through a series of incremental changes such as ‘layering’, ‘drift’, ‘conversion’ or reformulation such as ‘redesign’ (Howlett and Rayner, 2013). For example, in the context of technological innovations, these need to compete with existing technologies that have already been absorbed into the socio-​economic context (Callander, 2011; Giordano, 2012) requiring processes not only of learning but also often ‘coercion and negotiation’ (Christiansen et al, 2011; Rip and Kemp, 1998).

144

Policy myopia as a source of policy failure

The capacities and structures required for such ‘continuous anticipation and learning’ are also often lacking in contemporary policy making (Volkery and Ribiero, 2009; Wu et al, 2015). That is, the translation of the lessons of learning for effective policy design may not occur if there are ‘organisational rigidities’ and capacity gaps that impede changes, even if these are absolutely warranted, from entering into the policy realm (March and Olsen, 1975). This often occurs in both developed and developing countries due to the political, social and institutional challenges, that is, in terms of a lack of the resources needed to conduct relevant analysis and justify the need to deploy additional resources to plan for ‘unknown unknowns’ (Wu et al, 2015).

Conclusion A critical challenge that policy makers deal with in responding to policy problems concerns the conditions of uncertainty they face in ensuring their efforts will prove effective (Morgan and Henrion, 1990; Simon, 1991; Swanson et al, 2010).The types and levels of uncertainties must be correctly understood and diagnosed by policy makers if policy failures are to be avoided. In the short and long term both, inaction, delayed action or wrong policy action due to incomplete information and uncertainties about the future can lead to policy failure. This is a serious problem in policy making. As the literature on the subject over the past half-​century has shown, outcomes, outputs, impacts and target behaviour are only a few of the many aspects of policy making which are uncertain. Many of these uncertainties stem from the lack of knowledge or differential interpretation of cause and effect relationships between policy interventions and outcomes. If policy failures are to be avoided it is important that policy makers learn to carefully evaluate the types of risks they face in order to ensure that a policy deals with an issue effectively and dynamically, as intended (Hood, 2002; Twight, 1991). While many efforts of governments and other policy actors have aimed towards the reduction of uncertainty through better provision and processing of information, there are limits to the kinds of uncertainties that better information processing can address. An acknowledgement of the limitations to rational calculations and forecasts about the future in many problem areas characterised by deeper levels of uncertainty has led to an increased focus on the role of adaptation and learning in order to constantly update the knowledge base, gain experience and design and implement flexible and robust policies and resilient policy processes to deal with anticipated and unanticipated futures.

145

Policy Learning and Policy Failure

Deeper types of uncertainty result in situations where policy makers cannot rule out the possibility of ‘surprise’ or ‘unknown unknowns’ and, in fact, often can count on them occurring. Under such conditions, considering policy making as a dynamic process that is open to feedback and has a self-​correcting element (Da Costa et al, 2008) can help to eliminate or limit the extent of failure due to myopia. The concept of adaptive policies has been introduced by authors such as Swanson and Bhadwal, and Walker et al in order to help avoid policy failure and aid policy makers in developing policies which will continue to function effectively in achieving their end objective(s) even when conditions change (Rammel and van der Bergh, 2003). Such policy-​making processes and ideals, however, require high levels of policy capacity and both the capacity and willingness of policy makers to learn from their and from others’ experiences. In many cases either one or both of these fundamental pre-​conditions may be lacking. Overcoming such limitations, therefore, is a crucial pre-​requisite for successfully dealing with policy myopia. Notes Although another course of action in the face of such uncertainties is doing nothing, this can be prudent in some circumstances but in others can exacerbate a problem. Policy makers often face punishment by their constituents for their failures of commission and opt instead to fail through omission (Howlett, 2012; 2014; Saward, 1992). 2 Ingraham (1987) notes that such ‘corrections’ can be limited to suggesting the need for better goal clarity in the original policy programme even after it has been fully implemented. Consequently, such evaluation findings might sometimes refine or reshape the goals, and these might only have limited semblance to the original policy goals and objectives. 1

References Anderies, JM, Folke, C, Ostrom, E, Walker, B, 2012, Aligning key concepts for global change policy:  Robustness, resilience, and sustainability, Center for the Study of Institutional Diversity (CSID) Working Paper Series, #CSID-​2012-​002 Becker, SW, Brownson, FO, 1964, What price ambiguity? Or the role of ambiguity in decision-​making, Journal of Political Economy 72, 1,  62–​73 Bond, A, Morrison-​Saunders, A, Gunn, JAE, Pope, J, Retief, F, 2015, Managing uncertainty, ambiguity and ignorance in impact assessment by embedding evolutionary resilience, participatory modelling and adaptive management, Journal of Environmental Management 151,  97–​104

146

Policy myopia as a source of policy failure

Boulding, KE, 1947, A note on the theory of the black market, The Canadian Journal of Economics and Political Science/​Revue Canadienne d’Economique et de Science Politique 13, 1, 115–​18 Bredenhoff-​Bijlsma, R, 2010, Policy development under uncertainty: A framework inspired by cases of water management, Gildeprint, Enshede: The Netherlands Brown, L, 2000, Scientific uncertainty and learning in European Union environmental policymaking, Policy Studies Journal 28, 3, 576–​96 Brugnach, M, Dewulf, A, Pahl-​Wostl, C, Taillieu, T, 2008, Toward a relational concept of uncertainty: About knowing too little, knowing too differently, and accepting not to know, Ecology and Society 13, 2, 30 Callander, S, 2011, Searching for good policies, American Political Science Review 105, 4, 643–​62 Carpenter, SR, Brock, WA, 2008, Adaptive capacity and traps, Ecology and Society 13, 2, 40 Chandler, D, 2014, Beyond neoliberalism: Resilience, the new art of governing complexity, Resilience:  International Policies, Practices and Discourses 2,1, 47–​63 Chow, CC, Sarin, RK, 2002, Known, unknown, and unknowable uncertainties, Theory and Decision 52, 2, 127–​38 Christiansen, L, Olhoff, A, Trærup, S (eds) 2011, Technologies for adaptation:  Perspectives and practical experiences, Roskilde:  UNEP Risø Centre Churchman, CW, 1967, Wicked problems, Management Science 14, 4, B141–​42 Coglianese, C, 2011, Process choice, Regulation and Governance 5, 2,  250–​61 Committee on Decision Making Under Uncertainty, Board on Population Health and Public Health Practice, Institute of Medicine, Washington (DC), 2013, Environmental decisions in the face of uncertainty, Washington, DC: National Academies Press, doi: 10.17226/​12568 Da Costa, O, Warnke, P, Cagnin C, Scapolo F, 2008, The impact of foresight on policy making: Insights from the FORLEARN mutual learning process, Technology Analysis and Strategic Management 20, 3,  369–​87

147

Policy Learning and Policy Failure

Davoudi, S, Shaw, K, Haider, LJ, Quinlan, AE, Peterson, GD, Wilkinson, C, Fünfgeld, H, McEvoy, D, Porter, L, Davoudi, S, 2012, Resilience: A bridging concept or a dead end? ‘Reframing’ resilience: Challenges for planning theory and practice; Interacting traps:  Resilience assessment of a pasture management system in Northern Afghanistan; Urban resilience:  What does it mean in planning practice?; Resilience as a useful concept for climate change adaptation?; The politics of resilience for planning: A cautionary note, Planning Theory and Practice 13, 2, 299–​33 Day, P, Klein, R, 1989, Interpreting the unexpected: The case of AIDS policymaking in Britain, Journal of Public Policy 9, 3, 337–​53 Desai, M, 2015, Hubris: Why economists failed to predict the crisis and how to avoid the next one, New Haven, CT: Yale University Press Deyle, R, 1994, Conflict, uncertainty and the role of planning and analysis in public policy innovation, Policy Studies Journal 22, 3, 457–​73 Giordano, T, 2012, Adaptive planning for climate resilient long-​lived infrastructures, Utilities Policy 23, 80–​9 Haasnoot, M, Kwakkel JH, Walker, WE, ter Maat, J, 2013, Dynamic adaptive policy pathways: A method for crafting robust decisions for a deeply uncertain world, Global Environmental Change 23, 2, 485–​98 Hall, JW, Brown, A, Nicholls, RJ, Pidgeon, N, Watson, R, 2012, Proportionate adaptation, Nature Climate Change 2, 12, 833–​4 Hansson, S, 1996, Decision making under great uncertainty, Philosophy of the Social Sciences 26, 3, 369–​86 Head, BW, 2010, Reconsidering evidence-​based policy: Key issues and challenges, Policy and Society 29, 2, 77–​94 Head, BW, 2016, Toward more ‘evidence-​informed’ policy making? Public Administration Review 76, 3, 472–​84 Heazle, M, Tangney, P, Burton, P, Howes, M, Grant-​Smith, D, Reis, K, Bosomworth, K, 2013, Mainstreaming climate change adaptation: An incremental approach to disaster risk management in Australia, Environmental Science and Policy 33, 162–​70 Hilaire, J, Bauer, N, Brecha, RJ, 2015, Boom or bust? Mapping out the known unknowns of global shale gas production potential, Energy Economics 49, 581–​7 Hirsch-​Hadorn, G, Brun, G, Soliva, CR, Stenke, A, Peter, T, 2015, Decision strategies for policy decisions under uncertainties: The case of mitigation measures addressing methane emissions from ruminants, Environmental Science and Policy 52, 110–​19 Hood, C, 2002, The risk game and the blame game, Government and Opposition 37, 1, 15–​54

148

Policy myopia as a source of policy failure

Howlett, M, 2014, Why are policy innovations rare and so often negative? Blame avoidance and problem denial in climate change policy making, Global Environmental Change 29, 395–​403 Howlett, M, Newman, J, 2013, After ‘the regulatory moment’ in comparative regulatory studies: Modeling the early stages of regulatory life cycles, Journal of Comparative Policy Analysis: Research and Practice 15, 2, 107–​21 Howlett, M, Rayner, J, 2013, Patching vs packaging in policy formulation: Assessing policy portfolio design, Politics and Governance 1, 2, 170–​82 Ingraham, P, 1987, Toward more systematic considerations of policy design, Policy Studies Journal 15, 4, 611–​28 Ison, RL, Kevin, BC, Philip, JW, 2015, Institutionalising social learning: Towards systemic and adaptive governance, Environmental Science and Policy 53, Part B, 105–​17 Jacobs, AS, 2008, The politics of when: Redistribution, investment and policy making for the long term, British Journal of Political Science 38, 2, 193–​220 Jarvis, DSL, 2011, Theorising risk and uncertainty in international relations: The contributions of Frank Knight, International Relations 25, 3, 296–​312 Jones, BD, Baumgartner, FR, 2012, From there to here: Punctuated equilibrium to the general punctuation thesis to a theory of government information processing, Policy Studies Journal 40, 1, 1–​20 Kates, RW, Travis, WR, Wilbanks, TJ, 2012, Transformational adaptation when incremental adaptations to climate change are insufficient, Proceedings of the National Academy of Sciences 109, 19, 7156–​61 Kern, F, Howlett, M, 2009, Implementing transition management as policy reforms: A case study of the Dutch energy sector, Policy Sciences 42, 4, 391–​408 Knight, FH, 1921, Risk, Uncertainty, and Profit, Chicago, IL: University of Chicago Press Kwakkel, JH, Walker, WE, Marchau, VAWJ, 2010, Classifying and communicating uncertainties in model-​based policy analysis, International Journal of Technology, Policy and Management 10, 4, 299–​315 Lambin, EF, Meyfroidt, P, Rueda, X, Blackman, A, Börner, J, Cerutti, PO, Dietsch, T, 2014, Effectiveness and synergies of policy instruments for land use governance in tropical regions, Global Environmental Change 28, 129–​40

149

Policy Learning and Policy Failure

Lejano, RP, Shankar, S, 2012, The contextualist turn and schematics of institutional fit: Theory and a case study from Southern India, Policy Sciences 46, 1, 83–​102 Lempert, RJ, Popper, SW, Bankes, SC, 2003, Shaping the next one hundred years:  New methods for quantitative, long-​term policy analysis, Santa Monica, CA: RAND Lempert, R, Nakicenovic, N, Sarewitz, D, Schlesinger, M, 2004, Characterizing climate-​change uncertainties for decision-​makers: An editorial essay, Climatic Change 65, 1–​2, 1–​9 Leung, W, Noble, B, Gunn, J, Jochen, AGJ, 2015, A  review of uncertainty research in impact assessment, Environmental Impact Assessment Review 50, 116–​23 Levin, K, Cashore, B, Bernstein, S, Auld, G, 2012, Overcoming the tragedy of super wicked problems: Constraining our future selves to ameliorate global climate change, Policy Sciences 45, 2, 123–​52 Lindblom, CE, 1959, The science of muddling through, Public Administration Review 19, 2, 79–​88 Lindblom, CE, 1979, Still muddling, not yet through, Public Administration Review 39, 6, 517–​26 Lindblom, CE, Cohen, DK, 1979, Usable knowledge: Social science and social problem solving, New Haven, CT: Yale University Press Manski, CF, 2011, Policy analysis with incredible certitude, The Economic Journal 121, 554, F261–​89 Manski, CF, 2013, Public policy in an uncertain world: Analysis and decisions, Cambridge, MA: Harvard University Press Manzo, S, Nielsen, OA, Prato, CG, 2015, How uncertainty in input and parameters influences transport model output a four-​stage model case-​study, Transport Policy 38, 64–​72 Maor, M, 2012, Policy overreaction, Journal of Public Policy 32, 3,  231–​59 Maor, M, 2014, Risk and policy underreaction, Working Paper, Department of Political Science, Jerusalem:  Hebrew University, www.moshemaor.net March, JG, Olsen, JP, 1975, The uncertainty of the past: Organizational learning under ambiguity, European Journal of Political Research 3, 2,  147–​71 Mathijssen, J, Petersen, A, Besseling, P, Rahman, A, Don, H, 2008, Dealing with uncertainty in policymaking, The Hague/​Bilthoven/​ Leiden: CPB/​PBL/​Rand  Europe Maxim, L, van der Sluijs, JP, 2011, Quality in environmental science for policy: Assessing uncertainty as a component of policy analysis, Environmental Science and Policy 14, 482–​92

150

Policy myopia as a source of policy failure

May, PJ, 2014, Implementation failures revisited:  Policy regime perspectives, Public Policy and Administration 30, 3–​4, 277–​99 McConnell, Allan, 2010, Understanding policy success: Rethinking public policy, Basingstoke: Palgrave Macmillan McInerney, D, Lempert, R, Keller, K, 2011, What are robust strategies in the face of uncertain climate threshold responses? Climatic Change 112, 3, 547–​68 Morgan, MG, Henrion, M, 1990, Uncertainty: A guide to dealing with uncertainty in quantitative risk and policy analysis, Cambridge: Cambridge University Press Mumford, A, 2015, Parallels, prescience and the past:  Analogical reasoning and contemporary, International Politics 52, 1, 1–​19 Newman, J, Head, BW, 2015, Categories of failure in climate change mitigation policy in Australia, Special Issue:  Policy Failure, Public Policy and Administration 30, 3–​4, 342–​58 Newman, J, Howlett, M, 2014, Regulation and time: Temporal patterns in regulatory development, International Review of Administrative Sciences 80, 3, 493–​511 O’Neill, BC, Crutzen, P, Grübler, A, Ha Duong, M, Keller, K, Kolstad, C et al, 2006, Learning and climate change, Climate Policy 6, 5, 585–​9 Pahl-​Wostl, C, 2007, Transitions towards adaptive management of water facing climate and global change, Water Resources Management 21,  49–​62 Park, J, Seager, TP, Rao, PSC, Convertino, M, Linkov, I, 2013, Integrating risk and resilience approaches to catastrophe management in engineering systems, Risk Analysis 33, 3, 356–​67 Parry, ML, Canziani, OF, Palutikof, JP, van der Linden, PJ, Hanson CE (eds) 2007, Contribution of Working Group II to the fourth assessment report of the Intergovernmental Panel on Climate Change, 2007, Cambridge: Cambridge University Press Perl, Anthony, Michael Howlett and M Ramesh, 2018, Policy making and truthiness:  Can existing policy models cope with politicized evidence and willful ignorance in a ‘post-​fact’ world? Policy Sciences 51, 4, 581–​600, https://​doi.org/​10.1007/​s11077-​018-​9334-​4 Perrow, C, 1984, Normal accidents: Living with high risk technologies, New York: Basic Books Polasky, S, Carpenter, SR, Folke, C, Keeler, B, 2011, Decision-​making under great uncertainty: Environmental management in an era of global change, Trends in Ecology and Evolution 26, 8, 398–​404 Proctor, R, Schiebinger L (eds) 2008, Agnotology: The making and unmaking of ignorance, Stanford, Calif: Stanford University Press

151

Policy Learning and Policy Failure

Ramjerdi, F, Fearnley, N, 2014, Risk and irreversibility of transport interventions, Transportation Research Part A 60, 31–​39 Rammel, C, van den Bergh, JCGM, 2003, Evolutionary policies for sustainable development: Adaptive flexibility and risk minimizing, Ecological Economics 47, 2–​3, 121–​33 Rip, A, Kemp, R, 1998, Technological change, in S Rayner, L Malone (eds), Human choice and climate change, Vol 2 Resources and technology, Washington, DC: Battelle Press, pp 327–​99 Rittel, HWJ, Webber, MM, 1973, Dilemmas in a general theory of planning, Policy Sciences 4, 2, 155–​69 Roggema, R, Vereen, T, van den Dobbelsteen, A, 2012, Incremental change, transition or transformation? Optimizing change pathways for climate adaptation in spatial planning, Sustainability 4, 2525–​49 Saward, M, 1992, Co-​optive politics and state legitimacy, Aldershot: Dartmouth Schneider, SH, Kuntz-​Duriseti, K, 2002, Uncertainty and climate change policy, Chapter 2, in SH Schneider, A Rosencranz, J-​O Niles (eds) Climate change policy: A survey, Washington, DC: Island Press Schrader, S, Riggs, WM, Smith, RP, 1993, Choice over uncertainty and ambiguity in technical problem solving, Journal of Engineering and Technology Management 10, 1–​2, 73–​99 Simon, HA, 1973, The structure of ill structured problems, Artificial Intelligence 4, 3–​4, 181–​201 Simon, HA, 1955, A behavioural model of rational choice, Quarterly Journal of Economics 69, 1, 99–​118 Simon, HA, 1991, Bounded rationality and organizational learning, Organization Science 2, 1, 125–​34 Stirling, A, 2010, Keep It complex, Nature 468, 7327, 1029–​31 Swanson, D, Bhadwal, S (eds) 2009, Creating adaptive policies: A guide for policymaking in an uncertain world, Ottawa: IDRC/​New Delhi: Sage Swanson, D, Barg S, Tyler S, Venema H, Tomar S, Bhadwal S, Nair S, Roy D, Drexhage, J, 2010, Seven tools for creating adaptive policies, Technological Forecasting and Social Change 77, 6, 924–​39 Taeihagh, A, Bañares-​Alcántara, R, Givoni, M, 2013, A  virtual environment for the formulation of policy packages, Transportation Research A, Policy and Practice 60, 53–​68 Twight, C, 1991, From claiming credit to avoiding blame: The evolution of congressional strategy for asbestos management, Journal of Public Policy 11, 2, 153–​86 van der Steen, M, Chin-​A-​Fat, N, Vink, M, van Twist, M, 2016, Puzzling, powering and perpetuating: Long-​term decision-​making by the Dutch Delta committee, Futures. Policy making for the long term: puzzling and powering to navigate wicked futures issues 76, 7–​17

152

Policy myopia as a source of policy failure

Vermeulen, SJ, Challinor, AJ, Thornton, PK, Campbell, BM, Eriyagama, N, Vervoort, J, Kinyangi, J, Jarvis, A, Läderach, P, Ramirez-​Villegas, J, Nicklin, K, Hawkins, E, Smith, DR, 2013, Addressing uncertainty in adaptation planning for agriculture, Proceedings of the National Academy of Sciences 110, 8357–​62 Vink, M, van der Steen, M, Dewulf, A, 2016, Dealing with long-​term policy problems: Making sense of the interplay between meaning and power, Futures. Policy making for the long term: puzzling and powering to navigate wicked futures issues 76, 1–​6 Volkery, A, Ribeiro, T, 2009, Scenar io planning in public policy:  Understanding use, impacts and the role of institutional context factors, Journal of Technological Forecasting and Social Change 76, 9, 1198–​207 Walker, W, Marchau, V, Swanson, D, 2010, Addressing deep uncertainties using adaptive policies, Technological Forecasting and Social Change 77, 6, 917–​23 Walker, WE, Haasnoot, M, Kwakkel, JH, 2013a, Review. Adapt or perish: A review of planning approaches for adaptation under deep uncertainty, Sustainability 5, 3, 955–​79 Walker, WE, Lempert, RJ, Kwakkel, JH, 2013b, Deep uncertainty, in SI Gass, MC Fu (eds) Encyclopedia of operations research and management science (3rd edn), New York: Springer Walker, WE, Harremoes, P, Rotmans, J, van der Sluijs, J, van Asselt, MBA, Janssen, P, Krayer von Krauss, MP, 2003, Defining uncertainty:  A conceptual basis for uncertainty management in model-​based decision support, Integrated Assessment 4, 1, 5–​17 Walter, CJ, 1992, Perspectives on adaptive policy design in fisheries management, K Jain, LW Botsford (eds) Applied population biology, Dordrecht: Kluwer Academic Publishers Wardekker, JA, de Jong, A, Knoop JM, van der Sluijs, JP, 2010, Operationalising a resilience approach to adapting an urban delta to uncertain climate changes, Technological Forecasting and Social Change 77,  987–​98 Watson, J, Gross, R, Ketsopoulou, I, Winskel, M, 2015, The impact of uncertainties on the UK’s medium-​term climate change targets, Energy Policy 87, 685–​95 Wise, RM, Fazey, I, Smith, MS, Park, SE, Eakin, HC, Archer Van Garderen, ERM, Campbell, B, 2014, Reconceptualising adaptation to climate change as part of pathways of change and response, Global Environmental Change 28, 325–​36

153

Policy Learning and Policy Failure

Wu, X, Ramesh, M, Howlett, M, 2015, Policy capacity: A conceptual framework for understanding policy competences and capabilities, Policy and Society 34, 3–​4, 165–​71 Zeisel, H, 1981, Social science hubris? A  review of Lindblom and Cohen’s usable knowledge, Law and Social Inquiry 6, 1, 273–​81

154

Index Note: page numbers in italic type refer to Figures; those in bold type refer to Tables. A absorptive capacity (ACAP) in epistemic learning 31, 31–2 BTB (bovine tuberculosis) policy making 32–3, 43 acquis communitaire 86 adaptation, in policy transfer 52, 63, 83 adaptive learning 10, 11, 51, 54, 56, 56, 63 adaptive policies 142–3, 144, 146 administrative capacity (ADCAP) in epistemic learning 31, 33–4 BTB (bovine tuberculosis) policy making 34–6 Advanced Institute of Science, Korea 63 agility 134 ALP (Australian Labor Party) 123–4, 125–6, 126–7 see also Labor Party, New South Wales ambiguity 137, 138 analytical capacity (ANCAP) in epistemic learning 31, 36–8 BTB (bovine tuberculosis) policy making 24, 38–40, 43 Argyris, C 4 assemblage 72, 85, 87 AstraZeneca 61 Australian health insurance policy study 12, 113–14, 119, 122, 123–7, 128 Australian Labor Party (ALP) 123–4, 125–6, 126–7

B badger culling 25–6, 32–3, 34, 35–6, 38–40, 41–2 see also BTB (bovine tuberculosis) policy making BC Ferries 94, 99, 100, 101 see also British Columbia fast ferries study behavioural psychology 6, 14 beliefs, and policy learning 26–7 Bell Labs 58, 59 Berry, FS 73–4 Berry, WD 73–4 Bhadwal, S 144, 146 Bird, Malcolm G. 11–12, 13, 93–111

‘borrowing scenarios’ 51 bounded rationality 12, 133 Bourne, John 32 Bovens, M 5, 15 Boxall, A. 115 Bresnahan, T 57 bricolage 72, 85, 87 British Columbia fast ferries study 11–12, 93–6, 98–102, 106, 107 Brown, L 143 BTB (bovine tuberculosis) policy making 10 absorptive capacity (ACAP) 32–3, 43 administrative capacity (ADCAP) 34–6 analytical capacity (ANCAP) 24, 38–40, 43 communicative capacity (COMCAP) 24–5, 41–2, 43 epistemic learning 10, 24–8, 27, 30–42, 31

C Campbell, Gordon 102, 106 Canada: ‘sponsorship scandal’ 96 see also British Columbia fast ferries study Carr, Bob 104 certification of actors, and policy learning 27, 27 China, and policy transfer 85 Churchman, CW 136 Civil Service Reform Act 1978, US 77 Clark, Glen 101 coalition government, UK 1, 42 Cohen, W 53 communicative capacity (COMCAP) in epistemic learning 40–1 BTB (bovine tuberculosis) policy making 24–5, 41–2, 43 competition, and policy diffusion 74 Conservative government, UK see coalition government, UK contingent learning 6 ‘contributor’ role in epistemic learning 28–9, 29, 31

155

Policy Learning and Policy Failure Copenhagen, Denmark see Medicon Valley, SVM (Silicon Valley Model) of policy transfer copying, in policy transfer 52, 62, 75 cultural factors, impact on policymaking 9

D Davis, K 57 de Jong, M 85 decision-makers, relationships with epistemic communities 29–30, 36–7 deep uncertainty 135, 139, 140, 141, 143, 146 see also uncertainty Defence Advanced Research Projects Agency, US 62 defence industry 57, 59 DEFRA (Department for Environment, Food and Rural Affairs) 25, 26, 32–3, 34, 35, 36, 38–9, 41–2 see also BTB (bovine tuberculosis) policy making degenerate learning 13, 23–4 BTB (bovine tuberculosis) policy making 10, 24–8, 27, 30–42, 31, 43 deinstitutionalisation 114, 115, 119, 121–2, 123, 127, 128 democracy 95, 105, 116 Denmark see Medicon Valley, SVM (Silicon Valley Model) of policy transfer Department for Environment, Food and Rural Affairs see DEFRA (Department for Environment, Food and Rural Affairs) Dewey, J 116 diffusion see policy diffusion divergence 83 Dolowitz, D 50, 52, 55, 76 ‘double-loop’ learning 4, 144 Dror, Y 15 Dunlop, Claire A. 1–22, 23–47 dysfunctional learning 8, 10, 12, 15, 28, 42, 43, 78

E East Asia, policy transfer 82–3 EBPM (evidence-based policy making) 23–4, 35, 41–2, 116 ‘eclecticism’ 85 Egan, Michael 106–7 Elmore, RF 9, 114, 121 emulation, in policy transfer 52, 62, 75, 78 epistemic communities, and policy transfer 81

epistemic learning 27, 27, 43 absorptive capacity (ACAP) 31, 31–3, 43 administrative capacity (ADCAP) 31, 33–6 analytical capacity (ANCAP) 25, 31, 36–40, 43 BTB (bovine tuberculosis) policy making 10, 24–8, 27, 30–42, 31 communicative capacity (COMCAP) 24–5, 40–2, 43 unpacking of 28–30, 29 EU (European Union) 95 policy transfer and convergence 74, 82–3, 85–6 European Commission 86 Eurozone, policy failures 8, 9–10 evidence-based policy making (EBPM) 23–4, 35, 41–2, 116 expert inquiry in policy failure 116, 117, 118, 119–21 exploitative learning 53, 54 exploratory learning 53

F ‘facilitator’ role in epistemic learning 29, 29, 31 financial crisis, West’s response to 9 Flyvberg, B. 87 food banks, demand for 1–2 Fraser, Malcolm 124–5, 126, 127 Freedom of Information Act, US 79 Freeman, R 84

G Gambardella, A 57 General Dynamics 59 Giest, Sarah 10–11, 13, 49–70 Godfray, Charles 33, 36, 39 governments: avoidance of blame by 96, 97 exacerbation of policy failures 97, 106 ‘gradualism’ 85 Graduate Research Centre, Texas 59, 63 Greiner, Nick 103 groupthink 7, 31

H Hall, Peter 8 Hansson, S 136, 137 Harrison, K 79 Hassink, R 54 Hawke, Bob 125, 126 health policy reform 119 see also Australian health insurance policy study

156

Index Hirsch-Hadorn, G 136–7, 137 Hirschman, AO 31 Howard, John 126 Howlett, Michael 5, 12, 13, 133–54 hybridisation, in policy transfer 75, 83

I ignorance 137, 138, 138 Illical, M 79 IMF (International Monetary Fund) 76–7 implementation analysis 23 inappropriate transfer 76, 77, 78 incomplete transfer 76, 77, 84 Independent Scientific Group see ISG (Independent Scientific Group) information overload, and absorptive capacity 31–2 information sources, and absorptive capacity 32 infrastructure megaprojects 97–8, 106–7 see also British Columbia fast ferries study; Sydney Airport Rail Link study innovation see SVM (Silicon Valley Model) of policy transfer inspiration, in policy transfer 52, 62, 75 Institute of Science and Technology (IST), New Jersey 58, 59, 60 institutionalism 118–19, 128 definition of institutions 114 see also deinstitutionalisation; new institutionalism International Monetary Fund (IMF) 76–7 IR (international relations) 74, 75 ISG (Independent Scientific Group) 25, 26, 32–3, 35–6, 38–9, 40–1, 42, 43 see also BTB (bovine tuberculosis) policy making IST (Institute of Science and Technology), New Jersey 58, 59, 60 Italy, cultural factors in policy-making 9

KIST (Korea Institute of Science and Technology) 60, 61 Knight, FH 138 Korea Advanced Institute of Science (KAIS) 60 Korea Institute of Science and Technology (KIST) 60, 61 Krebs, John 35, 41 Kwakkel, JH 138, 140

L

Jacobs, A. 115 James, O 53 Janis, IL 7

Labor Party, New South Wales 102, 104–5, 106, 107 see also ALP (Australian Labor Party) Lagendijk, A 54 Landwehr, C 55 leadership studies 23 learning see policy learning learning clusters 54 learning in the shadow hierarchy 27, 28 learning regions 54 learning through bargaining 27, 28 Lendvai, N 83, 84 Leslie, S 50, 57, 58, 60 Levin, K 140 Levinthal, D 53 Lewis, J 120 Liberal Party, British Columbia 99, 102, 106, 107 Liberal Party-National Party coalition, New South Wales 102, 103, 104–5 Liberal-National Party Coalition, Australia 124–5, 126, 127 life sciences industry see Medicon Valley, SVM (Silicon Valley Model) of policy transfer Lindblom, CE 134 Ling-Temco-Vought (LTV) Corporation 59 Lippman, W 116 listening, and absorptive capacity 31 localisation 72 ‘norm localisation’ 84–5 Lockheed 59 Lodge, M 53 LTV (Ling-Temco-Vought) Corporation 59 Lundbeck 61

K

M

KAIS (Korea Advanced Institute of Science) 60 Kargon, R 50, 57, 58, 60 Kay, Adrian 12, 13, 113–32 Keating, Paul 126 Kenney, M 57 King, David 26, 36, 39, 42, 43

MAFF (Ministry of Agriculture, Fisheries and Food) 35 Major Projects Authority (MPA) 1 Malmberg, A 54 Marsh, D 50, 52, 55, 76 Martin, Paul 96 Martin, Thomas 59

J

157

Policy Learning and Policy Failure Maskell, P 54 Maxim, L 136 May, PJ 117 McCann, E 84 McConnell, A 6, 23, 73, 87 McPhail, Joy 101 Medibank health insurance scheme, Australia 123, 124, 125, 126–7 Medicare health insurance scheme, Australia 123, 124, 125–6 Medicon Valley, SVM (Silicon Valley Model) of policy transfer 50, 61–2, 63, 64 Ministry of Agriculture, Fisheries and Food (MAFF) 35 Moore, G 57 ‘moral hazards,’ and epistemic communities 37 MPA (Major Projects Authority) 1 Mukhtarov, F 74 mutation, in policy transfer 83, 85, 87

N Nair, Sreeja 12, 13, 133–54 NASA, US 62 National Farmers Union (NFU) 39–40, 43 NDP (New Democratic Party), British Columbia 98–9, 100, 101, 102, 107 negative lesson-drawing 71, 75, 78–80 new institutionalism 114–15, 126–7 New Jersey, SVM (Silicon Valley Model) of policy transfer 50, 58–9, 62, 63, 64 New Labour government, UK, EBPM (evidence-based policy making) approach 35, 41 new public governance 120–1 Newig, J 55 Newman, Joshua 11–12, 13, 93–111 NFU (National Farmers Union) 39–40, 43 ‘norm localisation’ 84–5 Novo Nordisk 61

O OECD (Organisation for Economic Cooperation and Development) 123 Oliver, C. 115 organisational capacity 23–4, 30

P Park Chung Hee 60 Patton, D 57 performance measures 115–16, 119–20 PFP (‘pay-for-performance’) 77–8 Pharmacia 61

policy diffusion 71, 72, 73–4, 86–7 ‘dud’ 76–8 see also policy transfer policy evaluation 2 policy failure 2, 13–15, 133 as catalyst for institutional change 121–3 definitions 3–4, 73 as a ‘degeneration’ of policy learning 23–4 expert inquiry 116, 117, 118, 119–21 failure to learn from 93, 95–8 future research 13–15 judgements and standards 115–17 learning from 24, 42–3, 113, 114, 117–21 macro level 9–10, 14, 15 McConnell’s typology of 6, 23 meso (group) level 7–8, 13, 14–15 micro (individual) level 6–7, 13–14 processes of 5 products 5–6 public political inquiry 116, 117–19 see also Australian health insurance policy study; British Columbia fast ferries study; BTB (bovine tuberculosis) policy making; Sydney Airport Rail Link study policy learning 2, 13–15, 78–9, 95–6, 143 and beliefs 26–7 definitions 3–4, 26–7 future research 13–15 link to policy failure 24, 42–3 macro level 8–10, 14 meso (group) level 7–8, 13, 14–15 micro (individual) level 6–7, 13–14 and policy transfer 53–5 and political learning 117 processes of 4–5 products 5 typology of 27, 27–8 see also epistemic learning policy myopia 12, 134, 145–6 learning to deal with 144–5 modelling of 134–8, 137, 138, 139, 140 and uncertainty 12, 13, 134, 135–8, 137, 138, 139, 140–3, 145–6 policy stages explanations of policy failure 23 policy tool analysis 23 policy transfer 11, 13, 50, 51–3, 71–3, 74–6, 86–7 intermediaries and interests 80–3 learning and adaptation in 53–5 negative lesson-drawing 71, 75, 78–80

158

Index and policy learning 53–5 processes in 75 soft forms of 53, 81, 83 success and failure in 55–6, 56 temporal transfers 85–6 truncated transfer and ‘dud’ diffusion 76–8 see also SVM (Silicon Valley Model) of policy transfer policy transformation, and regime change 96 policy translation 83–6, 87 policy variation 83 political inquiry in policy failure 116, 117–19 political learning 117 politics type of policy failure 23 Pollitt, C 119 Princeton University 58 problem tractability, and policy learning 27, 27 process type of policy failure 23 ‘producer of standards’ role in epistemic learning 29, 29, 31 programme type of policy failure 23 public political inquiry in policy failure 116, 117–19 public-private partnership schemes: British Columbia fast ferries study 94 Sydney Airport Rail Link 103–4 Putnam, Robert 9

R Radaelli, CM 8, 27 RBCT (Randomised Badger Culling Trial) 25, 34–5, 38 see also badger culling; BTB (bovine tuberculosis) policy making RCTs (randomised control trials) 34–5, 41 reflexive learning 27, 28 regional learning 54 resilience 142–3, 145 Resolution Foundation 1 risk 138, 145 Rittel, HWJ 136, 140 robustness 134, 142, 143, 144, 145 Rose, R 52–3, 54 Rothberg, RI 9 Rudd, Amber 2

S Sax, S. 124 Saxenian, A 57 Scania, Sweden see Medicon Valley, SVM (Silicon Valley Model) of policy transfer

Schön, D 4 selective borrowing 52 semiconductor industry 56–7 see also Silicon Valley ‘Silicon Somewheres’ see SVM (Silicon Valley Model) of policy transfer Silicon Valley 49, 51, 52, 56–7, 61, 62 see also SVM (Silicon Valley Model) of policy transfer Simon, HA 133, 136 ‘single-loop’ learning 4, 144 Small Business Innovation Research Program, US 62 SMU (Southern Methodist University) 59 soft forms of policy transfer 53, 81, 83 South Korea, SVM (Silicon Valley Model) of policy transfer 50, 60–1, 62, 63, 64 ‘sponsorship scandal,’ Canada 96 Stanford University 57, 58, 59 state corruption 9 Stirling, A 138 Stone, Diane 11, 13, 52, 53, 55, 71–92 structural adjustment policies 76–7 Stubbs, P 83, 84 ‘super-wicked’ problems 140 SVM (Silicon Valley Model) of policy transfer 10–11, 49–50, 62–5 case selection 50–1 Medicon Valley case study 50, 61–2, 63, 64 New Jersey case study 50, 58–9, 62, 63, 64 South Korea case study 50, 60–1, 62, 63, 64 Texas case study 50, 59–60, 62, 63, 64 see also Silicon Valley Swanson, D 144, 146 Sweden see Medicon Valley, SVM (Silicon Valley Model) of policy transfer Sydney Airport Rail Link study 11–12, 93–6, 102–5, 106–7 synthesis, in policy transfer 52, 75

T ‘t Hart, P 5, 15 ‘teacher’ role in epistemic learning 28, 29, 31 Teal, Gordon 59 Terman, Frederick 57, 58, 59, 60 Texas, SVM (Silicon Valley Model) of policy transfer 50, 59–60, 62, 63, 64 Thatcher, Margaret 96 TI (Texas Instruments) 59

159

Policy Learning and Policy Failure Toens, K 55 transformative learning 53–4 trial and error, in policy transfer 84 trial-and-error learning 54, 55, 56, 63, 64–5 Triantafillou, P 120 Trussell Trust 1

uninformed transfer 76, 84, 787 unintended consequences 84 un-learning 54, 63

U

Walker, WE 138, 139, 140, 146 Ward, K 84 Washington Consensus 76–7 Weaver, K. 115 Webber, MM 136, 140 Whitlam, Gough 123–4, 125, 126–7 ‘wicked’ problems 140 Wildavsky, Aaron 81 World Bank 76–7, 80

UC (Universal Credit) 1–2 uncertainty 12, 13, 134, 135–8, 137, 138, 139, 140–3, 145–6 deep uncertainty 135, 139, 140, 141, 143, 146 Levels I–IV 138, 139, 140, 141–3 options, outcomes and values 137, 137

V van der Sluijs, JP 136

W

160

The changing nature of politics and the contemporary challenges facing public and social policy demand new perspectives that contest existing assumptions and deliver new ways of understanding a changing world. From institutional reform to network governance and from public expectations to political inequality the New Perspectives in Policy & Politics book series will focus on state-of-the-art contributions that aim to reorient perennial debates or open up emerging seams of research. The series is interdisciplinary in approach, international in scope and seeks to publish books that are both theoretically informed and policy relevant. Most of all, the New Perspectives in Policy & Politics series seeks to challenge and redefine debates concerning both politics and public and social policy through fresh and innovative contributions to the field.

Policy successes and failures offer important lessons for public officials, but often they do not learn from these experiences. The studies in this volume investigate this broken link. The book defines policy learning and failure and organises the main studies in these fields along the key dimensions of processes, products and analytical levels. Drawing together a range of experts in the field, the volume sketches a research agenda linking policy scholars with policy practice.

Claire A. Dunlop is Professor of Public Policy at the University of Exeter.

“This book brings together two aspects of policy analysis in interesting and creative ways. Policy learning is often treated as a remedy for policy failures, but we find that learning can have its own pathologies. And failures may be a source of learning and improvement if considered properly. The analytic and empirical work in this book make significant contributions to our understanding of both failure and success in public policy.” B. Guy Peters, University of Pittsburgh

Edited by Claire A. Dunlop

“How do we know if policies have failed and in what way? Do we really want to learn, or to bury our heads in the sand? This marvellous collection of insights and case studies tackles the intersection of these issues in innovative and thought-provoking ways.” Allan McConnell, University of Sydney

Policy learning and policy failure

First published as a special issue of Policy & Politics, this updated volume explores policy failures and the valuable opportunities for learning that they offer.

New Perspectives in Policy & Politics Edited by Sarah Ayres, Steve Martin and Felicity Matthews

Policy learning and policy failure

ISBN 978-1-4473-5200-6

PolicyPress

policy.bristoluniversitypress.co.uk

@policypress

9 781447 352006

Edited by Claire A. Dunlop